diff --git a/.github/workflows/build_whitepaper.yml b/.github/workflows/build_whitepaper.yml index 84f0a46d1..1a9b6ea2f 100644 --- a/.github/workflows/build_whitepaper.yml +++ b/.github/workflows/build_whitepaper.yml @@ -21,6 +21,7 @@ on: paths: - 'whitepaper/**' - '!whitepaper/PUBLISHER_CONTEXT.md' + - '.github/workflows/build_whitepaper.yml' workflow_dispatch: # Allow manual trigger env: @@ -55,11 +56,16 @@ jobs: working-directory: ./whitepaper run: bash make-md.sh + - name: Verify PDF CI failure handling + run: python3 -m unittest discover -s whitepaper/tests -p test_pdf_ci_contract.py + - name: Build PDF Version working-directory: ./whitepaper run: | echo "Building PDF version..." - bash make-pdf.sh || echo "PDF build had issues, continuing..." + rm -f build/Synchronism_Whitepaper.pdf + bash make-pdf.sh + test -s build/Synchronism_Whitepaper.pdf - name: Build Web Version working-directory: ./whitepaper @@ -81,7 +87,7 @@ jobs: with: name: synchronism-pdf path: whitepaper/build/Synchronism_Whitepaper.pdf - if-no-files-found: warn + if-no-files-found: error - name: Upload Web Artifact uses: actions/upload-artifact@v4 @@ -175,4 +181,4 @@ jobs: echo "[📖 Read the Whitepaper](https://dp-web4.github.io/Synchronism/whitepaper/)" >> $GITHUB_STEP_SUMMARY echo "" >> $GITHUB_STEP_SUMMARY echo "---" >> $GITHUB_STEP_SUMMARY - echo "*Generated from commit: ${{ github.sha }}*" >> $GITHUB_STEP_SUMMARY \ No newline at end of file + echo "*Generated from commit: ${{ github.sha }}*" >> $GITHUB_STEP_SUMMARY diff --git a/manuscripts/publisher/PUBLISHER_LOG_synchronism.md b/manuscripts/publisher/PUBLISHER_LOG_synchronism.md index 7eaba6ea9..7cadc6a46 100644 --- a/manuscripts/publisher/PUBLISHER_LOG_synchronism.md +++ b/manuscripts/publisher/PUBLISHER_LOG_synchronism.md @@ -12,6 +12,402 @@ every entry below is byte-identical to what §6 carried. --- +### 2026-11-20 (autonomous-publisher-cbp): **One CHECKED note on §5.15 TEST-10. The 11-16 dSph note gave escapes at a local 3σ right after naming the 3.51 trials threshold, and the site's 11-19 figures (M/L ≥ 5.3 / 6.9) are the 3.51 ones. Both are right; the note now says which bar each answers. The site's new Walker+2009 row ("estimator-conditional": Crater II 2.9σ) sets the mass inside a sphere of radius R_e equal to half the light, which repeats the aperture slip it corrects. Plummer puts 0.354 of the light there. With the aperture matched, Walker gives 3.96 / 4.31, past 3.51. Count HELD at 6. Web4: no change, rebuilt and verified.** + +- **Step 0:** this run is the `autonomous-publisher-cbp` timer fire. #14 OPEN at `5cc49536` (the 11-19 pass), MERGEABLE, CI `build`/`summary` SUCCESS, `reviewDecision` empty, last review still NOVA's 11-13 HOLD. Fresh worktree `ai-workspace/.worktrees/syn-pub-1120` on `publisher/wp-2026-11-20`, branched from `5cc49536`. `origin/main` (`6905fd1b`) merged in cleanly (`3b72b52d`; it adds `PREDICTIONS.md` and one proposal, no whitepaper paths). +- **Window `0ac548ed..6905fd1b`:** `b26f6fa9` (11-19 back-annotation, five items) and `6905fd1b` (two design fixes to the 2×2 reinvention ablation before registration; gates on dp, nothing printed in the paper). I searched every live section with `/meta` excluded, printing each matched span: + - (i) **dSph estimator correction.** Reached. §5.15 kill-criteria item (3) already carries the 4/3 correction (11-16 note: z 4.48 / 4.21, ceiling ~11.0), which agrees with the site's 4.5σ / 4.2σ / ≈ 11. Two things do not agree, and both are dealt with below: the escape values (paper 6.3 / 8.7, site 5.3 / 6.9), and the site's new Walker+2009 row. + - (ii) **LLR bound on a measured gradient:** `d ln ρ`, `Berengut`, `Olive`: **0** hits in the paper's own claims. The one `d ln ρ` hit (L342) is the covariant branch's Ġ/G, a different quantity. Not reached. + - (iii) **"+2843 is a tilt":** the paper prints +2,843 at L101 as the source's figure on a 122-galaxy cut. The site now says where the loss comes from: 0.21 dex over-prediction in bright inner discs and 0.21 dex under-prediction in the deep regime. The paper makes no claim about where the loss sits, and its "close to a constant 2.6× stretch, and still loses" agrees with a tilt. The site attributes the 122-vs-166 cut to the loader's V_flat > 0 requirement; the paper says only "the sample cut differs", which is not wrong. Additive, not a correction, so not edited. + - (iv) **δ-family widths:** `1.006`, `1.167`, `1.17 dex` **0**. The one `δ = 6` (L144) is the Cassini-row sharpness at Υ_d ≈ 0.54. Not reached. + - (v) **2×2 reinvention ablation:** a frame proposal for the project lead. Gates on dp. +- **Escape values: the two sets answer different bars.** `synchronism-site/maintainer/scripts/cap6p39_vs_dsph.py` L48–55 computes the escapes "so that z = 3" and prints "passes 6.39 at 3 sigma only if M/L ≥ 8.4 / 11.6". Its tidal factor is σ_obs/(σ_max + 3·err), with the error not scaled. The 11-16 note printed these, and its 3/4 rescaling (6.3 / 8.7) is exact, because the escape M/L is linear in the coefficient. But the sentence before them in the note names the trials threshold 3.51. The site's 11-19 `cap6p39_vs_dsph_estimator.py` computes the escapes at 3.51, with the error scaling with σ: 5.3 / 6.9 at coefficient 3 (7.0 / 9.2 at 4), and inflation 1.15× / 1.31×. Both are correct; the paper had not said which bar it meant. Hand check, Draco at 3.51 and coefficient 3: (9.1 − 3.51·1.2)/√6.39 = 1.934 km/s → M/L = 1.934²·3·0.29·2/(G·2.2×10⁵) = 6.88. +- **New finding on the re-run: the Walker row repeats the aperture slip.** The site's 11-19 script writes Walker+2009 as M(R_e) = 2.5σ²R_e/G with R_e = ¾ r_3D, then sets M(R_e) equal to the half-light mass, so σ_N² = G·M_half/(1.875 r_3D). Walker's M is the mass inside a sphere of radius R_e, the projected half-light radius, and that sphere holds less than half the light. For a Plummer profile the fraction is 2^−1.5 = 0.354, with R_e = r_3D/1.305. Matched to its own aperture, Walker is the half-light form with coefficient **2.71**. That is 10% below Wolf's 3, the usual agreement between the two estimators, not a factor of 1.6. `simulations/publisher_20261120_dsph_walker_aperture.py` (run from `/tmp` with `python3 -I`, rc 0; identity control PASS, all six source rows reproduced at k = 4 / 3 / 1.875) gives, at M/L = 4: Crater II **z = 3.96**, Draco **4.31**, both past 3.51. Escape M/L at 3.51 is 4.7 / 6.2, the trials-corrected ceiling 7.6 / 9.9, and the inflation needed 1.09× / 1.25×. **The site's "estimator-conditional at its edge" does not survive the aperture match.** The exclusion stays conditional on M/L, on equilibrium (the site reports Crater II as tidally disturbed) and on the inputs. The 11-16 L_V × 1.3 stress test (Crater II 3.54) is still the edge. Limits: Plummer for the aperture fraction (exponential or King profiles move it by a few per cent). Walker's estimator was taken in its published form as the site quotes it; I did not re-read the paper here. Not edited in the site repo (other lane). Routed here and in the state file so the site maintainer can correct the Walker row and the "estimator-conditional" sentence in `PREDICTIONS.md` 11-19 (i). +- **Edit (`05-quantum-macro/15-dark-matter/dark_matter.md`, kill-criteria item (3)):** `[CHECKED 2026-11-20]` after the 11-16 note. Words added, none removed: the old text equals the new text with the note cut out (asserted in Python). The word-diff's one "removed" token is `]*;`, the tokenizer split at the insertion point, as on 11-16. Section CHANGELOG 05 (CRLF throughout; the entry was written with CRLF) and the top-level CHANGELOG. +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc 0 with `CI=true` and absolute script paths. PDF **314** pp locally (312 on 11-18), `Missing character` **0**, U+FFFD **0**. The note is in the text layer. In `section_5.html` it opens as `[CHECKED 2026-11-20`, with 0 literal `*[CHECKED 2026-11-20`. Geometry: **1** word past width − 36 (the p6 table-cell baseline), **0** past the page width. Artifacts restored by explicit path, not committed. +- **Deployed surface:** `docs/whitepaper/Synchronism_Whitepaper.pdf` on `main` is blob `320e7fa0`, the same blob as `c3e88fec` (#16's deploy). No whitepaper path has changed on `main` since, so the 11-19 measurement (292 pp, U+FFFD 0, 1 past w − 36, 0 clipped) stands. +- **Cross-repo: Web4. No change, verified by rebuild rather than by blob.** `origin/main` = `1ea6fb21`, and `bbaf2673..origin/main` (5 commits) touches no `web4-standard/core-spec`, `ontology`, `whitepaper` or `docs/whitepaper-web` path. Fresh worktree `ai-workspace/.worktrees/web4-pub-1120`, all three scripts rc 0. Diff: the `*Generated:*` line in both monoliths and the PDF text layer, nothing else; `index.html` byte-identical; `build/` = `docs/` (`cmp`). PDF 24 pp, `Missing character` 0, U+FFFD 0. Geometry: 8 words past 545 pt, the 11-02 baseline, max 596 (p15). Churn restored by explicit path. No web4 PR. Open web4 PRs touching the paper or the core spec: none (#984 hub mesh, #994 AIC demo). +- **So what:** the find has the same shape as 11-16's, one layer further down. The 08-23 table paired a coefficient with the wrong radius. The 11-19 check of that table then paired a second estimator with the wrong mass, and that row became a new qualifier ("estimator-conditional") that would have weakened a correct exclusion. An identity control cannot catch either slip, because each reproduces its own table. What catches them is reading what the estimator's M is the mass *of*. Productive rather than corrective: estimator-aperture pairing is now a two-instance defect class in this leg, and the check is one line to state when quoting any dSph boost. #14 now carries 35 commits and 33 dated notes that the deployed paper does not have; the merge decision is dp's. + +### 2026-11-19 (autonomous-publisher-cbp): **No paper edit. None of the six items in the 11-18 back-annotation (`0ac548ed`) is printed in the paper. The deployed PDF is clean. #14 now carries 32 dated notes that the published paper does not have. Count HELD at 6. Web4: no change.** + +- **Step 0:** this run is the `autonomous-publisher-cbp` timer fire (12:30 UTC). Shared Synchronism tree on `main` = `origin/main` (`0ac548ed`). #14 OPEN at `15d64c36` (the 11-18 pass), MERGEABLE, CI `build`/`summary` SUCCESS, `reviewDecision` empty. The last review is still NOVA's 11-13 HOLD. Fresh worktree `ai-workspace/.worktrees/syn-pub-1119` on `publisher/wp-2026-11-19`, branched from `15d64c36`. `origin/main` merged in cleanly (`ef9c77d9`; it adds `PREDICTIONS.md` and one proposal, no whitepaper paths). The shared `web4` checkout is still on `kimi/open-issue-audit-20261103` with dirty files from another lane, so it was read at `origin/main` only. +- **Window:** `0ac548ed` (11-18 06:10 PST). I searched every live section file with `/meta` excluded, using a Python scan that prints the matched span. Control: `Decoheren` gives 29 lines. + - (i) **Door #3 is closed as a measurement mechanism by its basis** (L ∝ p̂ damps momentum superpositions and leaves position coherence alone). Tokens searched: `door #3`, `Lindblad`, `L ∝ p`, `momentum diffus`, `preferred basis`, `Rovelli`, `CSL`, `measurement is synchroni`: **0** hits. Not reached. (The 11-14 entry had already found door #3 absent.) + - (ii) **LLR register quoted at the central value** (|d ln G_eff/d ln ρ_s| ≲ 4×10⁻⁹–4×10⁻⁸; at the 2σ edge it is ≲ 1×10⁻⁸–1×10⁻⁷). `d ln G`, `gradient factor`, `five to six orders`, `smooth 100`: **0**. Each `4×10⁻` hit is a different quantity (ρ_crit 4.4×10⁻⁶, rms 7.1×10⁻⁴, Wilcoxon p, Brouwer g_bar). The paper's LLR text is the covariant branch's Nordtvedt/Ġ failure (§5.15 L170/L185/L342/L366), which is ≥ 900σ and not the register bound. Not reached. + - (iii) **"Width saturates ~1.17 dex" is the γ → ∞ limit.** `1.17 dex`, `acts as a width`, `sharp member`, `transition width`: **0**. The one `δ = 6` (§5.15 L144, the 11-08 Cassini row note) is MOND's interpolating-family sharpness at Υ_d ≈ 0.54, not the compander width claim. Not reached. + - (iv) **Ciocan a₀(z), one bar convention for both sides.** The paper's Ciocan block (§5.15 L263–L275) tests branch (A) a₀ ∝ E(z) across four anchors (≈10σ low … 0.5σ … ≈2σ high) and already reads Ciocan's a₁ interval as a 95% CI. It does not print the bin-slope reading that the annotation corrects (n = 1 against 0.71 ± 0.13, "unresolved, straddled"): `straddl`, `0.71`, `2.2σ`, `4.4σ` **0**. Not reached. + - (v) **DR4 companion rate:** `DR4` **0**. The paper's wide-binary text is the old DR3 C(a) fit (L285) and the S685 MOND+EFE reduction (exec L190). Neither one is touched. Not reached. + - (vi) **Post-cutoff arm default:** `post-cutoff` **0**. Gates on dp. +- **Verdict:** no `sections/` edit, and no rebuild owed on #14. Nothing in `sections/` has changed since the 11-18 build (312 pp, clean), and CI's `build` passed on `15d64c36`. +- **Deployed surface (re-measured, not carried):** `docs/whitepaper/Synchronism_Whitepaper.pdf` on `main` (`c3e88fec`, #16's build) is `cmp`-identical to `whitepaper/build/`. It has **292** pp and U+FFFD **0**. Geometry: **1** word past width − 36 (p6 table cell, the baseline), **0** past 612 pt. #16's fix holds. +- **Magnitude of the hold (new this pass; the 11-17/11-18 entries gave it only as "N passes deep").** At `ef9c77d9` against `origin/main`, `whitepaper/sections/` is 14 files, +191/−29 lines, over 33 commits. A scan for dated bracket notes (`[CHECKED|SCOPED|NOTE|UPDATE|AMENDED|CORRECTED … 2026-MM-DD`; a lower bound, since it misses notes whose date is not in the bracket head) finds **79** on #14 and **47** on `main`. **32** are dated after 10-27, and `main` has **0** of those. The last content deploy is #13 (10-27), so a reader of the published paper has been without 23 days of corrections. Those include the 1/φ, a₀ and S₈ status-cell qualifiers that NOVA's own HOLD asked for. +- **Cross-repo: Web4. No change.** `origin/main` = `1ea6fb21`, unchanged since 11-13. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`), so the 11-02 geometry baseline carries over unmeasured. Open web4 PRs: #984 (draft, hub mesh), not this lane's. +- **So what:** the yield of this pass is a negative. The back-annotation stream has moved to items the paper never printed: door #3, the DR4 plan, the post-cutoff arm and the site's register numbers. Two of the last six passes (11-14, 11-19) found nothing that reaches the paper. That is the expected steady state once the paper's printed claims have been swept, and it is a signal about cadence, not a failure. The open lever is no longer finding the next sibling sentence. It is whether the 32 notes ship: NOVA's HOLD asked for the status-cell qualifiers, which have been on #14 since 11-13, and the merge decision is dp's. I am stating the number rather than restating the 11-17/11-18 frame point a third time. + +### 2026-11-18 (interactive maintenance pass): **§6.4 OQ-Coarsening printed "x is a virial ratio, capped at ~0.02 for every gravitationally bound system" and "~40× below the knee" with no condition on the density law. Back-annotation `5d2d684c` (ii) says the cap holds only for A·V² with one shared length. Re-run on SPARC: under the pairing the archive's fits actually use (S53's 0.028·V^0.5 against Σ/2h at h = 0.3 kpc), 62 of 175 discs reach the knee on stellar Σ alone. Annotated CHECKED at both sites. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`5d2d684c`), clean. #14 OPEN at `f99ebc65` (the 11-17 pass), MERGEABLE, `reviewDecision` empty. The last review is still NOVA's 11-13 HOLD, which the 11-13 pass answered. No review and no dp comment since 11-17. Fresh worktree `ai-workspace/.worktrees/syn-pub-1118` on `publisher/wp-2026-11-18`, branched from `f99ebc65`. `origin/main` merged in cleanly (`2c27ce11`; it adds `PREDICTIONS.md` and one proposal, no whitepaper paths). +- **Window:** `5d2d684c` (11-17 06:12 PST, after the 11-17 pass). Live sections were searched per datum, with `/meta` excluded: + - (i) **B1 "S ≤ 2 no-signaling bound" is the local-realist bound.** `no-signal`, `S≤2`, `2√2`, `Tsirelson`, `PR box`: **0** hits in the paper's own claims. The paper prints B1 only as "S = 1.98 ≤ 2 on both arms" (§5.15 count item (6)), which makes no no-signaling claim. `no-signaling` appears once, inside L134's own statement that the token has zero occurrences. Not reached. + - (ii) **Three knee laws, and the virial argument holds only for A·V² with one shared length.** Reached at §6.4 OQ-Coarsening. The table's first row already says the crossing is "estimator-dependent, not settled by margin", but the prose below it (L51) states the cap "for every gravitationally bound system", and the correction note (L55) says the two-sided convention "keeps every bound system ~40× below the knee". **Edited** (below). + - (iii) Ciocan constant-a₀ σ (23/21/19.7σ) and the knee's velocity-exponent 3.5σ vs 3.6σ: `21σ`, `23σ`, `19.7`: **0** hits. The paper's Ciocan block gives no σ for constant a₀. The paper's one `3.6σ` (§5.15 L138) is TEST-09's |Δn| from zero, a different quantity. Not reached. + - (iv) **Frame proposal** (`the_external_value_has_moved_to_other_peoples_theories_20261117.md`): direction, outreach, B1 row edit and the stopping question all gate on dp. No paper edit. +- **Re-run, not relayed:** `simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py`, run from `/tmp` with `python3 -I`, rc 0. Identity controls PASS: two-sided caps 0.0190 (4π form) and 0.2387 (primary form) against the table's 0.019 and 0.24; Jeans A at 8 kpc = 4.565×10⁻⁵ against the paper's 4.56×10⁻⁵. Source check: Session 53 L40–59 derives ρ_crit = V²/(Gα²R_half²) and substitutes R_half = R₀V^0.75 to get A·V^0.5. So S53's law *is* the V² Jeans law with the galaxy's R_half as the length in A, and the cap appears only if ρ is smoothed over the same R_half. SPARC (`MassModels_Lelli2016c.mrt`, innermost point, Σ = 0.5·SBdisk + 0.7·SBbul, gas omitted so x is a lower bound, V = mean of the last 3 Vobs): **62/175** discs reach x > 1; median 0.45; NGC 2403 2.83, NGC 3198 2.64, NGC 7331 3.91, DDO 154 0.07, UGC 128 0.21. These match the back-annotation's "2.6–3.9 in the bright centres … below 1 on DDO 154 and UGC 128" (it used gas too, and the fits' Υ). They also agree with §5.15's 11-15 note (63/153 on the solved midplane density, a different estimator). Side note: S53's printed A ≈ 0.028 comes out as 0.0248 at its own α = 1.1, R₀ = 0.088 kpc, and 0.030 at α = 1. It is not used here. +- **Edits (`06-implications/04-open-questions/open_questions.md`; old text = new text with the two notes cut out, checked in Python; the word-diff's 2 "removed" tokens are tokenizer splits at the insertion points):** full `[CHECKED 2026-11-18]` after the 08-28 qualifier at L51, and a short pointer after "~40× below the knee" at L55. Section CHANGELOG 06 (LF in the entry region) and the top-level CHANGELOG. +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc 0 with `CI=true` and absolute script paths. PDF **312** pp (311 on 11-17), `Missing character` **0**, U+FFFD **0**. Both notes are in the text layer. In `section_6.html` both open as `[CHECKED`, with 0 literal `*[CHECKED`. Geometry: **1** word past width − 36 (the page-6 baseline), **0** past the page width. Artifacts restored by explicit path, not committed. +- **Cross-repo: Web4. No change.** `origin/main` = `1ea6fb21`, unchanged since 11-13. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`, the same blobs as 11-17), so the 11-02 geometry baseline carries over unmeasured. Deployed PDF 24 pp, U+FFFD 0. Open web4 PRs: #984 (draft, hub mesh), not this lane's. +- **Frame point, restated (gates on dp):** #14 is now 31 passes deep and the deployed paper (`c3e88fec`, 11-15) still has none of the 10-29…11-18 notes. Today's §6.4 edit touches a file #14 does not touch. Its only dependency is the cross-reference to §5.15's 11-15 note, which exists only on #14. A per-day PR from `main` could carry it after a one-clause change, but splitting the log, which lives on #14, is the convention change the 11-17 pass left to dp. Not done unilaterally. +- **So what:** the back-annotation's (ii) is a site-side correction, and the paper had the same compression one section away from where a site reader would look. §6.4's table was right, and its prose summary dropped the condition. This is the visitor log's "a correction lands on one page and survives on the others" failure in the paper's own text. The researcher persona's frame point (iv) applies here too: today's yield is again propagation, not discovery, and the pass that would change that is dp's ruling on (iv), not another note. + +### 2026-11-17 (interactive maintenance pass): **§5.15 printed "SPARC selects γ ≈ 0.489 ≈ ½, the exact MOND point" and "the data select the MOND point rather than merely tolerating it". Back-annotation `d71f1f8b` shows the fitted γ is the incumbent's: the template fitted with no data gives 0.500 for simple μ and 0.595 for McGaugh's RAR function, and nuisance-refit SPARC sits at 0.58–0.62, so it cannot separate the two. Annotated CHECKED. §5.16's "superconductor γ ~2" is BCS's γ at a pinned knee; annotated. Outside the window: the paper prints "BCS ratio derived: 2√π ≈ 3.54, observed 3.52 (<1% error) ✓" at three sites. The 3.52 is BCS's own 3.528 from the gap equation, not an observation. Annotated CHECKED at all three. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`d71f1f8b`), clean. #14 OPEN at `4fc98005` (the 11-16 pass), MERGEABLE/CLEAN, `reviewDecision` empty. The last review is still NOVA's 11-13 exact-head HOLD (COMMENT review), and the 11-13 pass answered it. No review since. Fresh worktree `ai-workspace/.worktrees/syn-pub-1117` on `publisher/wp-2026-11-17`, branched from `4fc98005`, upstream `origin/publisher/wp-2026-10-29`. `origin/main` merged in cleanly (`9d771710`; it adds `PREDICTIONS.md`, one proposal and the Session 1 back-annotation, with no whitepaper paths). +- **Window:** `d71f1f8b` (11-16 06:14 PST, after the 11-16 pass branched). Live sections were searched per datum, with `/meta` excluded: + - (i) **every fitted γ is the incumbent's.** `0.595`, `0.58`, `incumbent`, `Υ-frozen`: **0** hits. `0.489` appears many times, and three of those sites make the "selects ½ / selects the MOND point" claim with no qualifier beside them: the status-table γ = 2 cell (L14), the γ = 2 bullet (L58) and the "What it costs" paragraph (L91). The DE-sector paragraph (L366) already scopes 0.489 as a fixed-Υ slice, but that is 300 lines away from the claim. **Edited** (below). The superconductor leg reaches §5.16's cross-scale table ("Superconductor | tanh(E/2kT) | ~2"). **Edited.** + - (ii) **the N_corr-divergence test answered "diverging".** The paper prints this test at exec summary L221 ("whether N_corr converges … or diverges. Currently mixed"). **Not edited:** the back-annotation says the Bucket 3 edit gates on the project lead, and the paper should not get ahead of the ledger's canonical row. It is recorded here and in the state file as a gated item. + - (iii) the 1/Ω_m floor is the cosmic C₀: `genuinely derived` **0** hits. The paper's `1/Ω_m` sites (L117, L175, L181, L217) already read it as `C_Ω`'s floor and the 10-29 note ties it to `C₀ = Ω_m`. Not reached. + - (iv) Saad & Ting, 9.8σ, 527×: `Saad` **0** hits. The paper's "at least 527×" (L342) is the 11-06 note's own grid statement and is consistent with "grid minimum at γ = 0.3". Not reached. +- **Re-run, not relayed:** all five `synchronism-site/explorer/scripts/gamma_incumbent_map/*.py` (`sc2`, `nest`, `rng`, `dmap`, `dict`) were run from a `/tmp` copy with `python3 -I`: rc 0 for each, **byte-identical** to the committed `_output.txt`. Identity controls hold: the archive points reproduce 2.308 / 2.645 / 2.453, and simple μ returns γ = 0.500, a₀′ = 6.000×10⁻¹¹ exactly. `sc2.py`'s model is the archive's `synchronism_gap_model_v2` at β = 2 with the same bounds (0.5–4), checked against `simulations/chemistry/superconductor_coherence.py` L117–131 and L239–255. **Relayed, not re-run:** the "careful SPARC arms 0.58–0.62 (95% ≈ 0.43–0.95)" are prior site fits, and the note says so. +- **Outside-window find: the BCS gap ratio.** While reading §5.16 for (i), I found "**Derived Gap Ratio**: 2Δ₀/(k_B T_c) = 2√π ≈ 3.54 … Observed: 3.52 (<1% error)". It is printed again with a check mark in the conclusion's "Progress (Apr 2026)" list, and a third time in Appendix B L88. No earlier note covers any of the three sites. `simulations/publisher_20261117_bcs_ratio_and_gamma_check.py` solves the BCS gap equation (weak coupling, ħω_D = 1): 2Δ₀/T_c = 3.5278 / 3.5279 / 3.5289 / 3.5322 at λ = 0.15 / 0.20 / 0.25 / 0.30. The identity control is the closed form 2π·exp(−γ_E) = 3.5278. 2√π = 3.5449 is 0.49% above it. So "observed 3.52" is BCS's own prediction, and "<1% error" is the distance between a coincidence and the theory the paper says it is unifying with. The source (Chemistry Session #1 §3.3) builds √π from Gaussian integrals and never touches the gap equation. The same session's script computes the γ-dependent ratio 2π/tanh(γ ln 2) = 7.12 at γ = 2 and prints "Simple formula gives too high ratio". The paper carried the coincidence and not the script's own verdict. The measured elemental range (about 3.2–4.6) is quoted from standard tables and was not re-read; the note says so. +- **Edits (words added, none removed; `git diff --word-diff=porcelain` shows 0 removed tokens):** §5.15: full `[CHECKED 2026-11-17]` on the L58 bullet, and short pointers in the L14 cell and at L91. §5.16: `[CHECKED]` under the cross-scale table (it also notes that the text above the table reads the same tanh as γ = 1 while the table says ~2), and `[CHECKED]` after the Derived Gap Ratio. Conclusion "Progress" list: `[CHECKED]`, check mark kept. Appendix B L88: short `[CHECKED]`. Section CHANGELOGs 05/07/09 (05 and 09 are CRLF in the entry region; inserted byte-wise with endings copied from the neighbouring entry, 4-line diffs each) and the top-level CHANGELOG. +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc 0 with `CI=true` and absolute script paths. `Missing character` **0**; U+FFFD **0**; PDF **311** pp locally (310 on 11-16). All 7 notes are in the PDF text layer. In `section_5/7/9.html` every note opens as `[CHECKED` / `[NOTE`, with **0** literal `**` and **0** literal `*[`. Geometry: **1** word past width − 36 (the page-6 baseline), **0** past the page width. Artifacts restored by explicit path, not committed. +- **Cross-repo: Web4. No change.** `origin/main` = `1ea6fb21`, unchanged since 11-13. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`). Deployed PDF 24 pp, U+FFFD 0, geometry = the 11-02 baseline (8 words past 545 pt, max 596.2, 0 past 600). Open web4 PRs: #984 (draft, hub mesh), not this lane's. +- **Frame point (surfaced, gates on dp):** #14 now carries 30 commits and 19 dated passes since 10-29. NOVA's HOLD is about the headline status table, and the 11-13 pass answered both exits it named, but no review has come back in four days. Meanwhile the **deployed** paper (`c3e88fec`, 11-15) has none of the 10-29…11-17 corrections. Each daily pass makes #14 bigger and harder to review, so the lane's efficient path (stack another pass) moves away from its purpose (get corrections in front of readers). Two ways out, both dp's call: (a) a ruling on #14 as it stands (merge, or name what still blocks); (b) let the lane open per-day PRs from `main` for notes that do not touch the held status table. Today's §5.16 / conclusion / Appendix B notes are independent of #14's hunks and could ship that way. I did not restructure on my own: it would split a reviewed history, and the stacking convention is dp's. +- **So what:** today's real find came from outside the window. The visitor loop audits the site and the back-annotations follow it, so a claim that lives only in the paper's chemistry-track sections (§5.16, the conclusion's "Progress (Apr 2026)" list, Appendix B) has no channel that will ever check it. The 2√π line stood for seven months, with a check mark, beside the script's own "too high". The rest of the Progress list (Hückel 4n+2 "derived from phase closure (exact)") has the same shape and is untested. Untested is not wrong, so it is left alone here and named for the next pass. + +### 2026-11-16 (interactive maintenance pass): **The TEST-10 note said Ω_m/Ω_b = 6.39 "is excluded … by none after the correction". That is true of discs. The archive's own 08-23 dwarf table excludes 6.39 at z 4.9 in both Draco and Crater II at M/L = 4, past the trials threshold, with no light rule. Annotated CHECKED. On the re-run I also found that the source's mass estimator pairs Wolf+2010's projected-radius coefficient (4) with 3D radii, so the archive's required boosts are 4/3 high. The exclusion survives the correction (z 4.48 / 4.21), and the trials-corrected ceiling falls from ~14.7 to ~11.0. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`f971b1bc`). #14 OPEN at `ed3774cf` (the 11-15 pass). Fresh worktree `ai-workspace/.worktrees/syn-pub-1116` on `publisher/wp-2026-11-16`, branched from `ed3774cf`, upstream `origin/publisher/wp-2026-10-29`. `origin/main` merged in cleanly (`948a3ded`; it adds only `PREDICTIONS.md` and one proposal, no whitepaper paths). +- **Window:** `f971b1bc` (11-15 06:09 PST, after the 11-15 pass), back-annotation + `Research/proposals/the_cap_family_is_closed_by_dwarfs_and_every_fitting_sector_picks_the_known_physics_gamma_20261115.md`. Searched live sections (`/meta` excluded) per datum with a literal-pattern helper (ugrep rejects `.{0,160}` around multibyte text): + - (i) **dSphs exclude the 6.39 cap.** `Draco`, `Crater`, `dSph`, `spheroidal`, `14.7`, `permissive cap` **0**. `6.39` hits once: §5.15 kill-criteria item (3), the 11-09 TEST-10 note, "Ω_m/Ω_b = 6.39 is excluded by 2 discs at χ² ≥ 9 and by none after the correction". Correct for discs, but a reader takes it as the cap surviving the dynamics, and the archive's executed dwarf leg says otherwise. This is the back-annotation's own failure shape: a one-leg verdict printed while a contradicting executed leg sat in the record. **Edited** (below). + - (ii) the 20.3 closure scoped to fiducial M/L: `20.3`, `1/Ω_b`, `dwarf kinematics` **0**. Not reached. + - (iii) which field equation each test used: the paper already keeps TEST-11 "conditional on a modified-Poisson (QUMOND) completion" (L144), states the μ_simple identity as a fact about the compander algebra that "does not carry over" to L2 (L238), and never says L3 or the AQUAL action passed Cassini. `μ_eff`, `x(x+3)`, `2a₀′` **0**. Not reached. Checked the derivation by hand anyway: F(y) = C(√y/a₀′)·y gives dF/dy = C + ½xC′; at γ = ½, C = x/(x+2), so μ_eff = x(x+3)/(x+2)², deep limit 3x/4. Agrees. +- **Re-run, not relayed:** `synchronism-site/maintainer/scripts/cap6p39_vs_dsph.py` (sha256 `c21c153d…`, PREREG site `c7d729c`) from a `/tmp` copy, `python3 -I`: rc 0, **byte-identical** to the committed `_output.txt`. Its inputs match `explorer/findings/scripts/pressure_supported_boost_ceiling.py` (08-23) row for row, including σ errors. Hand check: Draco at 6.39, fiducial M/L: σ_max = 2.28 km/s, z = (9.1 − 2.28)/1.2 = 5.68. +- **New finding on the re-run (estimator coefficient).** The 08-23 docstring reads "sigma_N from Wolf+2010: M_1/2 = 4 sigma^2 r_1/2 / G", and the inputs are labelled `r_half_3D` (Crater II 1.42 kpc ≈ 4/3 of its projected ~1.07). Wolf et al. 2010 give M_1/2 = 3⟨σ²⟩r_1/2/G for the 3D radius and ≈ 4⟨σ²⟩R_e/G for the projected one. The table mixes them, so σ_N² is 3/4 of Wolf's value and every B_req is 4/3 high (Draco 101.5 → 76.1, Crater II 60.2 → 45.1). `simulations/publisher_20261116_dsph_cap_estimator_sensitivity.py` (identity control at k = 4 PASS against all six source values, after I fixed my own first draft, which hard-coded z = 3.51 instead of the source's bisection value 3.5087 and failed on Draco's 14.7): at k = 3 and M/L = 4, Draco z = 4.48 and Crater II 4.21, both past 3.51; the largest cap Draco excludes after trials is 11.0 (was 14.7), Crater II 8.4 (was 11.2). Stress test, k = 3 and L_V × 1.3 (no catalogue value was read; this only states a margin): Crater II 3.54, at the threshold, and Draco 4.04. **The verdict that 6.39 is excluded survives; the "excluded up to ≈ 14.7" headline does not, at the right coefficient.** Not edited in the site repo (other lane). Routed here and in the state file so the site maintainer can correct the 08-23 table, the 11-15 B_max ≈ 14.7 and 18.5, and the 08-23 escape M/L values (8.4 / 11.6 at k = 4 become 6.3 / 8.7 at k = 3; the paper's note gives both). +- **Edit, §5.15 kill-criteria item (3): `[CHECKED 2026-11-16]` after the 11-09 note.** It adds words and removes none: the old line equals the new line with the note cut out (`git diff --word-diff=porcelain` shows one "removed" token, `output.]*;`, which is the whitespace tokenizer splitting the closing `;` onto the note's end). The note scopes "by none" to discs, gives the dwarf z values, the trials threshold, the escapes (M/L ≥ 8.4–11.6 or tidal σ inflation 1.26–1.33×, i.e. non-equilibrium), Draco as a shared MOND+EFE failure, Crater II as McGaugh 2016's a-priori MOND success, the coefficient correction with its numbers, the stress test, and the limits (quoted σ errors only; no anisotropy, r_1/2 or distance uncertainty; equilibrium estimator). Count HELD at 6: TEST-10 already fires as registered. Section CHANGELOG 05 (CRLF, edited byte-wise, 4-line diff) and the top-level CHANGELOG. +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc 0 with `CI=true` and absolute script paths; `Missing character` **0**; U+FFFD **0**; PDF **310** pp locally (308 at the 11-15 local build; 292 deployed, the gap being the local pandoc 3.1.3 / TeX toolchain, which also adds `class="header"` rows to unrelated sections' HTML). The note is in the PDF text layer. A pandoc probe on the edited paragraph renders it as `` with **0** literal asterisks. Geometry: **1** word past width − 36 (page 6, the baseline), **0** past 612. Artifacts restored by explicit path, not committed. +- **Cross-repo: Web4. No change.** `origin/main` = `1ea6fb21`, unchanged since 11-13. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`). Deployed PDF 24 pp, U+FFFD 0, geometry = 11-02 baseline (8 words past 545 pt on pages 6/12/13/15/22, max 596.2, 0 past 600). Open web4 PRs: #984 (draft, hub mesh), not this lane's. +- **So what:** the find is the same shape the 10-27 frame point named, a correction that reaches one leg and not its sibling. This time the re-run also turned up a defect upstream of every number in the leg: a coefficient that has sat in the 08-23 table for 85 days and was copied into the 11-15 test. An identity control that reproduces the 08-23 values cannot catch it, because it checks arithmetic against a table that carries the same coefficient. Same lesson as the 09-01 column misread: read the writer's formula, not just its output. + +### 2026-11-15 (interactive maintenance pass): **Two printed claims in §5.15 were corrected at source weeks or months ago and never reached the paper. (1) The 08-18 locality fork still read "no order-ε⁰ channel … ~10× underpowered". The 09-15 correction says that holds only without a pressure term: read at local density as a fluid, existing P(k) already pins γ = ½ ± 5×10⁻⁶. (2) The dielectric action's striction force still read "≤ 2×10⁻⁵ of g … observationally free". That was computed at the 0.029·V² units slip. At the archive's own knee it is 0.5–1.6% of gravity radially and 9–27% of vertical gravity. Both re-run, both annotated CHECKED. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`c3e88fec`). **#16 merged** 11-14 16:00 PST (`8e23c116`) and CI deployed it (`c3e88fec`), so the deployed PDF no longer clips (the 11-14 entry's frame point 1 is done). #14 OPEN at `a79d9492` (the 11-14 pass), `reviewDecision` empty. Its last review is still NOVA's 11-13 HOLD, which the 11-13 pass answered. Fresh worktree `ai-workspace/.worktrees/syn-pub-1115` on `publisher/wp-2026-11-15`, branched from `a79d9492`, upstream `origin/publisher/wp-2026-10-29`. `origin/main` merges into it cleanly. The only artifact it changes is the PDF, because #16's three build hunks were byte-identical to #14's. +- **Window:** `f3b3c5bc` (11-14 06:10 PST, back-annotation + `Research/proposals/the_record_is_the_novel_content_and_the_cap_family_at_its_permissive_cap_20261114.md`). It landed after the 11-14 pass (12:30Z). Searched live sections (`/meta` excluded) per datum, with a literal-pattern helper: + - (i) **"no disc crosses the knee" was the units slip.** `crosses the knee`, `mass cancels`, `45 of 122` **0**. But the same premise is printed under another name. §5.15 L117 (08-04 dielectric amendment) says the polarization force `C′(ρ)|∇Φ|²/8πG` "is ≤ 2×10⁻⁵ of g across five SPARC galaxies: observationally free". Its source (`Research/proposals/dielectric_completion_and_efe_linearity_equivalence_20260804.md` L35) gives the reason as "because ρ_crit sits far above disk densities". That is the striction the back-annotation calls "~10⁻⁵ of gravity … computed at the hybrid only; at the archive knee it is uncomputed". The site explorer computed it the same day, after the back-annotation (site `6d97ef9`, 11-14 08:12 PST, PREREG `f9b5293`, vertical half post-hoc). That result has not been back-annotated into this repo. **Re-run, not relayed:** see the edit below. + - (ii) the cap family at Ω_m/Ω_b = 6.39: §5.15 count item (3) (11-09 note) already prints "excluded by 2 discs at χ² ≥ 9 and by none after the correction" and the 5.7 edge. The lensing paragraph's 11-11/11-12 notes already handle the deflection rule. `census` hits are the 24-test census and the lensing-ceiling script, not McGaugh 2010. Not reached. + - (iii) what the template forbids: `1.17 dex`, `width–tail` **0**. Not reached. + - (iv) **γ_cos = 0.487 is a mean-density number.** §5.15 L366 prints γ = 0.487 ± 0.02 as the substituted branch's DESI fit. Two sentences earlier it prints the 08-18 locality fork unqualified: "adds **no independent channel** … no order-ε⁰ term exists … the channel is ~10× underpowered". The 09-15 back-annotation corrected exactly that, and the 11-13 backfill audit lists it as **item 6, OPEN**: "rests on site-executed numbers, so it needs a re-run before inscription". `pressure term`, `Jeans term`, `Sandvik`, `Amarzguioui` **0** before this pass. **Re-run:** see the edit below. + - The frame proposal (the methodological record as a third kind of value; 1.4% consistency yield; J ≈ 0) and its three dp decisions (SPINE, DR4 key, author contact): `methodological record` **0**. The paper's `1.4%` hits are #615's disputed rate, already labelled "do not quote alone". Gates on dp. +- **Edit 1, §5.15 L366: `[CHECKED 2026-11-15]` after the 08-18 sources sentence.** It adds words and removes none (`git diff --word-diff=porcelain`: 0 removed tokens). The 08-18 claims hold only without a pressure term. Read at local density with the dark energy comoving (fluid Cardassian), the sector is a barotropic fluid with c_s² ≈ 0.22 ε, and the Jeans term carries all of the effect. Existing P(k) excludes the horn unless |ε| < 9.1×10⁻⁶ / 9.9×10⁻⁶. At γ = 0.487, P/P_ΛCDM = 9.7 at k = 0.01 h/Mpc. So γ = 0.487 survives only on the mean-density reading, which gives up the local evaluation that the same paragraph says the postulate requires. Prior art (Sandvik+2004; Amarzguioui, Elgarøy & Multamäki 2005) is cited. The limit is stated: if only the CDM carries the pressure, galaxy P(k) loosens to |ε| ~ 10⁻⁴ and lensing does not loosen. **Re-run:** `synchronism-site/explorer/findings/scripts/de_horn_l_as_a_fluid_jeans_term.py` (sha256 `4413a9ba…`, PREREG site `7d8efbf`) and `de_horn_l_fluid_controls.py`, from `/tmp` copies, `python3 -I`: rc 0, **byte-identical** to the committed `_output.txt` for both (`cmp`). +- **Edit 2, §5.15 L117: `[CHECKED 2026-11-15]` after "observationally free".** Words added, none removed. At the archive knee, 63 of 153 discs cross on the solved midplane density. Radial striction: median 5.4×10⁻³ (γ = 0.489) and 1.6×10⁻² (γ = 2) of gravity, max 0.26 and 0.51, never outward-dominant in 3,035 points. Vertical striction (post-hoc): median 9% and 27% of vertical gravity, above half somewhere in 40 and 110 of 153 discs, compressive in 98.6% and 99.3% of |ratio|-weighted cells. So the radial term is small but not free, and the rotation-curve verdicts carry over under this action only at fixed disc thickness. The note says it scopes the magnitude only. **It does not claim EFE = 0 is unaffected**, because that is backfill item 4 (OPEN, partly gates on dp), and an earlier draft that said "unaffected" was cut. **Re-run:** `synchronism-site/explorer/scripts/l3_striction_archive_v05_knee.py`, `l3_vertical_striction_archive_v05_posthoc.py` and `l3_vertical_striction_sign_posthoc.py`, run from `/tmp`. They read the site's SPARC build by absolute path and write nothing. All rc 0. Vertical and sign outputs are byte-identical. Radial differs only in its two timing lines (`built … in 25s` vs 29s, `elapsed 99s` vs 107s). The 0.029·V² control row reproduces the 10-13 numbers (L2 249.64, median 6.53×10⁻⁶). Watch item 17 (striction numerics site-executed, never re-run here) is partly answered: these three are now re-run. The 08-26 "164× K_z" figures are not. +- **Section CHANGELOG:** `05-quantum-macro/meta/CHANGELOG.md` (mixed CRLF), one entry written byte-wise with CRLF to match its neighbours. +- **Build (verification only; CI builds and deploys on merge):** `CI=true bash whitepaper/build.sh rebuild`, rc 0. `Missing character` **0**, U+FFFD **0**. PDF **308** pp (292 deployed; the difference is #14's accumulated notes). Geometry: **1** word past width − 36 pt, the p6 table cell, the same as deployed. The PDF text layer's literal-`**` lines stay at **26**, the same as deployed. The notes are on PDF p158 (rendered at 70 dpi and checked) and in `section_5.html`. After edit 2: 309 pp, the same geometry, U+FFFD and `**` counts, `Missing character` 0. The L117 note is on PDF p132 (rendered and checked), and `section_5.html` carries both notes with 0 literal `**`. Artifacts restored by path (`git checkout -- whitepaper/build docs/whitepaper`). +- **Gate, self-caught:** two `mrh.command.unenumerated` refusals. A `$r` loop over the two repos and a `<(…)` process substitution in a diff were both rewritten with literal paths. That makes five passes running with the loop shape. A `$var` that holds a page number also tripped it. +- **Cross-repo: Web4. No change, no rebuild owed.** `origin/main` = `1ea6fb21`, the same head the 11-13 and 11-14 passes read, so the window is empty. Since #978 (11-05) the window holds #971 (4-hub release), #988/#989 (web4-policy `matches` operator), #990/#991 (hub PRDs) and #992 (hub site shell), with **0** `web4-standard/core-spec`, `ontology` or `whitepaper/` changes. §11's Hub paragraph still matches `hub/README.md` (reference POC, `0.1.0-alpha.0`, private production repo). `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`). Deployed PDF: 24 pp, U+FFFD 0. Geometry equals the 11-02 baseline: 8 words past 545 pt on 5 pages (p6, p12, p13, p15, p22), max 596.2, 0 past 600. The shared `web4` checkout is still parked on `kimi/open-issue-audit-20261103` with dirty files from another lane, so it was read at `origin/main` only. +- **So what.** Both corrections were ones the research lane had already done. One was done nine weeks ago (09-15), the other yesterday. Neither reached the printed sentence it corrects. In the 09-15 case the printed sentence even names the reading ("which the framework's one-equation postulate actually requires") under which its own conclusion fails. These are the propagation failures the 10-27 entry predicted: a correction lands on its sibling and not on the sentence. The backfill audit's table found item 6. The audit could not close it, because "re-run before inscription" was an action and the audit was a disposition. **The audit table works as a queue, so the next passes should drain it oldest-first** (items 2, 3, 9 and the L364 half of 5 remain; 1, 3 and 4 partly gate on dp). +- **Frame points for dp:** + 1. **#14 is 17 days and 25 commits.** #16 shows the cost: a formatting repair sat 14 days under a scientific hold. Today's two notes are corrections of printed errors, not new claims, and they wait under the same hold. If the hold is about the headline count and the status-cell labels (NOVA 11-09/11-13), the alternative is to split corrections that only qualify printed text into their own PR, as #16 did for the build. + 2. The 11-14 proposal's three decisions (SPINE "third kind of value", DR4 key, author contact) gate on dp. DR4 is **17 days out** (2 December). + +### 2026-11-14 (interactive maintenance pass): **No paper edit: the window does not show the paper wrong anywhere. The deployed PDF has cut 78 words off at the page edge, on 40 pages, for the 14 days #14 has been held, so the build-only fix is split out of #14 as #16. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`97f638a9`). #14 OPEN at `08227d0b` (the 11-13 pass), MERGEABLE, `reviewDecision` empty. Its last review is NOVA's 11-13 04:14Z HOLD, which the 11-13 pass answered. Nothing from a reviewer since. Fresh worktree `ai-workspace/.worktrees/syn-pub-1114` on `publisher/wp-2026-11-14`, branched from `08227d0b`, upstream `origin/publisher/wp-2026-10-29`. The shared `web4` checkout is still parked on `kimi/open-issue-audit-20261103` with ~100 dirty files (not this lane's), so it was read at `origin/main`. +- **Window:** `97f638a9` (11-13 06:09 PST, back-annotation + `Research/proposals/the_framework_by_piece_and_a_dr4_default_20261113.md`). It relays site `6f25275` (explorer 11-12), which the 11-13 pass named as "next pass", and site `2e4a627` (maintainer 11-13). `git grep -c -F` over live sections at `08227d0b`, control `MOND` = 120: + - (i) density law at the archive's own knee (`0.028·V^0.5`, ΔBIC +8,404, 115/122): `8,404`/`8404` **0**. The paper's `0.028`/`V^0.5` hits (exec L190, §5.15 L128, OQ L51, conclusion L78) are the S687 "coefficient outlived its source" and the 08-28 keying notes, both consistent with it. The solar boost "3.5×10⁴ → 2.7–10.5": `Oort` **0**, and the one `3.5×10⁴` (OQ L41) is a different quantity, the spread of three ℓ conventions for NGC 3198. Not reached. + - (ii) every executed loss by equation: `by-piece` **0**. Bears on the count. See the frame point. + - (iii) which action is "L3": the paper's 8 `L3` hits (§5.15 L158–L208) are its own L1/L2/L3 labels (L3 = the division law `g = g_bar/C`), defined in place. `μ_eff`, `xC′`, `under L3`, `γ ≈ 0.44` **0** in §5.15. The research lane's "L3 action" is a different object under the same name, and the paper does not print it. Not reached. + - (iv) Hyades/open clusters, (v) DR4 default: `Hyades`, `open cluster`, `Torres`, `DR4` **0**. Not reached. The DR4 default (2 December) gates on dp. + - Navier–Stokes banner on FUNDAMENTALS: the paper is already caveated at all four sites. Exec L133 says "retracted by the audit arcs"; A.3 carries the 05-28 inline tension note (S617/S665/S666); §4.1 L135 is "refuted-pending-reformulation"; A.16 is `[PARALLEL-PATHS]` with a status line. The table's "Planck … No (exact)" cell sits under that status line. Not edited. + - Phase-16 "near-reach" decoherence pointer: `intrinsic decoherence`, `door #3`, `near-reach`, `Borexino` **0**. The paper's three `Dk²` hits are the S666 decay-vs-oscillation statement, not the prediction. Not reached. +- **Verdict:** no `sections/` edit. Nothing in `sections/` has changed since the 11-13 build (`08227d0b` is log + state + report), so no rebuild is owed on #14. +- **Deployed-surface lint (new this pass: geometry on the PDF as deployed, not as built on #14).** `docs/whitepaper/Synchronism_Whitepaper.pdf` on `main`: 292 pp, U+FFFD **0**. But **96 words** past page width − 36 pt, and **78 on 40 pages past 612 pt**, i.e. cut off at the physical edge. The 10-31 fix (`62f7610a`, `pdf-breakable-code.lua`) brought this to 1 on #14's build, and #14 has been held since. A formatting fix had been waiting on an editorial ruling. + - **Split out as #16** (`publisher/pdf-breakable-code-1114`, base `main`). Only the three build hunks, byte-identical to `62f7610a`; no section content. Built at `97f638a9` with `CI=true`: past-margin **96 → 1** (p6 table cell, the 10-31 baseline), clipped **78 → 0**, 292 → 292 pp, `Missing character` 0, U+FFFD 0. Text layer, normalised for reflow (whitespace, hyphens, page-number lines): 99 insertions (1,780 chars of restored path tails) + 16 layout moves (mid-line page numbers, a TOC entry 184 → 183, table reading order). The one character-level case (`site+4b9c681` → `site 4b9c681`) matches the source. p15 was rendered before and after. + - **Self-caught:** my first commit of #16 added a "re-measured" clause to the PUBLISHER_CONTEXT line. `git merge-tree` of #14's head over it then CONFLICTED, while the commit message said "identical hunks". Reverted the clause to `62f7610a`'s text, amended, and the merge is clean. The trailer `Seat:` and PR signature were also missing on the first push and were added the same way. +- **Gate, self-caught:** three refusals. (a) `mrh.command.unenumerated` on `$r`/`$b`/`$f` loops, rewritten as a script file or literal paths. That is the fourth pass running with this shape; a `$p` loop over grep patterns passed. (b) `gate.self_access` on `git show 62f7610a -- … > /tmp/pub1114/fix.patch && git apply`. The rule is `governance-closure-opaque-writer`, which reads `git apply` of a redirected patch as an opaque writer. Escalation **`562c7952bf2fe73a`** is open and not needed; the same two one-line edits were made with the Edit tool. (c) Looking for the `hestia` CLI to retire it opened scope request **`scope-b0b7290bb1cc`** (`/home/dp/.local/bin`), also not needed. Both are left for a peer to decline; this seat has no CLI route to retire its own. +- **Instrument note:** `pkill -f "git grep -n -o -E"` killed its own shell (exit 144), which is the `pgrep -f matches its own waiter` shape. The grep it targeted was a catastrophic-backtracking `.{350}X.{350}` regex on 40 KB lines. The context helper used instead is `/tmp/pub1114/ctx.py` (literal find, fixed window). +- **Cross-repo: Web4. No change, no rebuild owed.** `origin/main` = `1ea6fb21` (11-12 22:01 PST), the same head the 11-13 pass read: an empty window. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`). Deployed PDF: 24 pp, U+FFFD 0. Geometry is **exactly** the 11-02 baseline: 8 words past 545 pt on 5 pages (p6 ×2, p12, p13, p15, p22 ×3), 0 past 600. +- **So what.** The one change that reaches readers today is the clipping fix. A reader of the deployed PDF could not see where 78 of the paper's own provenance pointers end. That repair was in hand for 14 days and blocked only by being stacked with content under a scientific hold. This also suggests a rule for this lane: **build fixes do not go on the content stack.** +- **Frame points for dp:** + 1. **#16** is mechanical and independent of #14. Merging it deploys a readable PDF of the current main. + 2. **The count.** The site's `exclusion_table_by_equation_output.txt` (44 rows + 4 withdrawn) states that under its strictest rule (R5: canonical-doc assertion, framework-specific data, standing) **none of the paper's six survive**. Nine other rows do: dens-gal, dens-local, γ–N_corr, de-cov, door #3, T_c, chemistry ladder, B4. Under R0 the six are also only **5 independent roots** (TEST-09 and TEST-10 share the cap). The paper prints "count 6" as a headline at least four times, and the 08-08 amendment already says it is a count of test IDs never enumerated in the paper. The research lane kept the count at 6 and added a by-piece table on the site. Whether the paper's headline should be the registered-and-fired count or a by-equation table is dp's call. It is the same scale question as the 10-18 banner option, now with a table to print. + 3. **#14:** 16 days, 23 commits, unchanged today. This entry adds one log-only commit. + + +### 2026-11-13 (autonomous-publisher-cbp, systemd 04:30): **NOVA's 11-13 review on #14 named two exits. This pass takes both: the three status cells carry their 11-11 qualifiers in place, and the outage-window backfill audit is written up as a disposition table. The window's one paper-reaching item, "η ≡ AG is a label", does not survive a check, and the paper is left alone on it. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`a7078191`). #14 OPEN at `ca78abce`. **New since the 11-12 pass: NOVA exact-head review, 2026-11-13 04:14Z, still HOLD.** It credits the 11-11/11-12 notes ("resolves the *absence of qualifications*"). It holds on two things: the cells stay "stronger and visually prior" to their annotations, and "a one-time publisher-window backfill audit". Fresh worktree `ai-workspace/.worktrees/syn-pub-1113` on `publisher/wp-2026-11-13`, branched from `ca78abce`. Shared `web4` checkout still parked on `kimi/open-issue-audit-20261103`; read at `origin/main`. `synchronism-site` on `main`, read at `origin/main`. +- **WAKE.** The reviewer's request was the work. The 11-12 frame point said #14 needs a merge or a ruling more than another pass, and NOVA's review now names what it would accept short of a ruling. So no new physics notes today unless the window shows the paper wrong. +- **Exit 1, status cells (`53e0214f`).** 1/φ VALIDATED, a₀ = cH₀/(2π) DERIVED and S₈ PREDICTED each carry a short dated qualifier beside the bold label. The S₈ "Independent validation" Notes cell gains "(not independent …)". The note's lead gains one parenthetical saying so. **Labels unchanged, words added, none removed** (word-diff: 0 removed tokens). The same table's γ = 2 row already had an in-cell qualifier, so this follows the table's own precedent. Relabelling (the 11-12 recommendation: RECOVERED / NOT EXCLUDED / RETRODICTED) is still dp's. +- **Exit 2, backfill audit** (`reports/outage_window_backfill_audit_2026-11-13.md`). The 10-18 ten plus 09-23, each read by hand at `ca78abce`. Result: **3 closed (7, 8, 09-23), 3 partial (1, 5, 10), 5 open (2, 3, 4, 6, 9).** The find: the headline sentence "both surviving decisive negatives are galactic (locality no-go; TEST-09 BTFR bounded-boost)" (exec L189, §5.15 L5, conclusion L77) rests on items 4 and 1. Both are open at the headline, and the §5.15 body prints item 1's condition while the headline does not. The source `📌` blocks were not re-read today. Item descriptions are 10-18's. +- **Window.** Synchronism `a7078191` (11-12 06:12, back-annotation + proposal `the_landing_equation_is_the_one_object_no_test_supports_and_the_corner_rule_is_milgroms_20261112.md`). Site `b874662` (explorer 11-12, incl. `6f25275` "every executed loss by equation: the footer's six and the archive's own decisive losses barely overlap": not read this pass, and it bears on the count, so next pass). `git grep -c -F` over live sections, control `MOND` = 84: + - (i) Milgrom 2011 partition rule / open clusters: `1111.1611`, `open cluster`, `Pleiades`, `partition`, `spectral decomposition` **0**. Not reached. + - (ii) landing equation: `landing` 4, none of them this (chemistry "landing at γ~1", "landing the arc's", "landing site"). The decision gates on dp. + - (iii) a₀(z) 2.15 → 1.86: `2.15`, `1.86`, `cH(z)`, `E(1)` **0**. Not reached. **η ≡ AG: reached, 3 sites** (exec L182, superconductivity L147, conclusion L68). +- **η ≡ AG checked, not relayed, and not inscribed.** `manuscripts/publisher/scripts/eta_vs_abrikosov_gorkov_20261113.py`: 607.3 K and 230.8 K reproduce. "AG lowers T_c" is true (the digamma difference is ≥ 0, so T_c ≤ T_c0 for every pair-breaking rate). But the printed formula T_c = Δ/(1.76 k_B η) compares T_c with **the gap**, not with T_c0. On AG's gapped branch, with Δ = the order parameter, η_AG = Δ(0)/(1.764 k T_c) ≥ 1, so η = 0.38 is outside AG. With Δ = the spectral gap (what a measured 35 meV is), η_AG falls monotonically and reaches 0.38 at α/α_c ≈ 0.62. So the 11-12 argument does not decide the identity: it is **untested, not refuted** in either direction. I was one edit from inscribing it. Routed back to the research lane, not to the paper. (A d-wave cuprate is not AG's s-wave setting either, which the paper's "identical" also skips.) +- **Build (verification only; 20 artifact files restored by explicit path, CI deploys):** md, pdf and web-clean rc 0 with `CI=true`. `Missing character` **0**; U+FFFD **0**; PDF **307 → 308** pp. The table (pp 119–120) was rendered and read: the qualifiers wrap in the Status column, italic, legible. `section_5.html` literal `**` **0**, `*[` **0**. Not fixed: `Ω_Λ = (1 - Ω_m)` prints underscores literally. That is the paper-wide plain-text notation (`Ω_m`, `f_DM`, `g_bar`), not a one-off. +- **Gate, self-caught:** one `mrh.command.unenumerated` refusal (`$p` in a page loop), rewritten in Python. This is the third pass running with this shape. Earlier I also opened an unneeded scope request (`scope-9033c5194292`, `/var/log/publisher-cbp.log`) by reading the unit's own log out of scope. Not needed; the request is left for a peer to decline. +- **Cross-repo: Web4. No change, no rebuild owed.** `origin/main` = `1ea6fb21`. The window since `69a2337b` is #992 (hub/site visitor shell). `web4-standard/`, `whitepaper/`, `docs/whitepaper-web/` diff **empty**. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`, unchanged). Open PRs touching those paths: **0**. +- **So what.** Both of NOVA's conditions now have something concrete to accept or reject. The one new physics item turned out to be a correction that would itself have needed correcting. Nothing reached the paper from it, so the paper did not print a second overstatement in the opposite direction. +- **Frame points for dp:** + 1. **#14:** 15 days, now 23 commits with this entry. The cells carry their qualifiers. Merge, or name a cell. + 2. **Headline sentence (audit item 1 + 4):** "both surviving decisive negatives" is the line with the most unpropagated conditions in the paper. A banner (10-18 option a) would carry it in one edit. + 3. **η ≡ AG:** the ledger's "label" verdict needs the comparator stated (order parameter vs spectral gap) before either reading is printed. + +### 2026-11-12 (interactive maintenance pass): **One correction, to this PR's own 11-11 note. The lensing leg of §5.15 does not depend on a light-deflection rule: light on visible matter alone predicts no lensing boost, and Brouwer's measured RAR has one. Stacked on #14. Count HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`15d9e61d`), clean. #14 OPEN, MERGEABLE/CLEAN at `19cc159d`. Its only review is still the NOVA HOLD of 11-09, which the 11-11 table note answered; nothing from a reviewer since. Fresh worktree `ai-workspace/.worktrees/syn-pub-1112` on `publisher/wp-2026-11-12`, branched from `19cc159d`, upstream `origin/publisher/wp-2026-10-29`. +- **WAKE.** The 11-11 frame point 2 said this lane should stop per-item annotation and do one rewrite pass once #14 merges. #14 has not merged, so today's rule was: edit only where the new commit shows the paper wrong. Relays of new site work and new prior art wait for dp. +- **Window:** `15d9e61d` (11-11 06:21, back-annotation + proposal `efe_free_corner_needs_a_system_rule_and_the_light_rule_has_no_free_horn_20261111.md`); site `3d53a42`..`8b0daeb`. `git grep -c -F` per datum over live sections at `19cc159d`, control `MOND` = 117: + - `EFE-free`, `Saturn`, `Crater`, `Antlia`, `Boran`, `emulator`, `Shapiro`, `LHAASO`, `Saad`, `expert review`, `independent review`, `total acceleration`: **0**. (i), (ii), (iv) and most of (v) do not reach the paper. + - `0.057` (10): every hit is the DESI fσ₈ error bar (+0.057/−0.066), not the RAR scatter that (v) re-attributes to Li+2018. **Not reached.** `TEST-15` (5): the paper already closes it (S673); Boran+2018 as prior art is a relay that gates on dp. `0.489` (29) and `Chae` (14) were not traced to (v)'s specific claims; worth a look on a later pass. + - `deflection` (2 lines, one note): **reached.** Part (c) of the 11-11 CHECKED note on the Brouwer paragraph said "dies independently" holds *only* if light follows the boosted potential. Correction (iii) of `15d9e61d` says that was too generous. +- **Checked, not only relayed.** The visible-only branch is decidable from what the paper already prints. Light on baryons alone gives g_obs = g_bar, raised by at most Brouwer's ×10. Brouwer et al. report a measured RAR that "agrees well with the MG predictions". On MOND's curve (a₀ = 1.2×10⁻¹⁰) the boost at g_bar = 1.3, 2.3 and 4×10⁻¹³ is 30.9, 23.3 and 17.8. ×10 baryons leave 3.1×, 2.3× and 1.8× of it unexplained, all above Brouwer's reliable floor of 10⁻¹³ (their Sec. 5.4, quoted by the site; I have not read Sec. 5.4 myself). The site's per-bin σ (7.2–10.6σ at ×10, from the explorer's 10-04 transcription; `maintainer/scripts/efe_free_corner_which_acceleration_output.txt` section 3) is relayed and labelled. The one passing reading puts unseen matter in the lens. +- **Edit.** I replaced part (c) in place rather than adding an UPDATE: it is one clause, this lane wrote it yesterday, and it has not merged. The note now carries "Corrected 2026-11-12" inside it, and `git log` keeps the old wording. One attribution was tightened during the pass ("The paper reports" → "Brouwer et al. report", because "the paper" is ambiguous inside a whitepaper). Changelogs: top-level (+1) and meta 05 (CRLF, written byte-wise). +- **Build (verification only; 20 artifact files restored by explicit path, CI deploys):** md, pdf and web-clean rc 0 with `CI=true`. `Missing character` **0**; U+FFFD **0**; PDF **307** pp, unchanged. The note is on p150 of the text layer, and p259 is the changelog. `section_5.html` has the new text, with literal `**` **0** and `*[` **0**. Words past 576 pt on p150: **0**. The final two-phrase wording edit was not rebuilt; it changes no glyph or markup. +- **Self-caught, gate:** two `mrh.command.unenumerated` refusals (a shell variable holding a worktree path; a `$(cat list)` substitution for the restore). Both were rewritten with literal paths, and no scope request was opened. This is the same shape the 11-11 entry records, so the lane's habit of writing `W=...` has not gone away. +- **Cross-repo: Web4. No change, no rebuild owed.** `origin/main` = `69a2337b`; window since `d0b0dfda` = #991 (hub PRD docs). `git diff --stat d0b0dfda..origin/main -- web4-standard whitepaper docs/whitepaper-web` is **empty**. Open PRs touching `whitepaper/` or `web4-standard/`: **0**. `build/` = `docs/` by blob (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`, the same as 11-11). +- **So what.** One sentence the PR was about to publish made a refutation look softer than its own data allow. It is fixed before merge, and it cost one clause, not a new note. Most of today's window (the corner's undefined acceleration, the dSph result, Boran) does not touch the paper, and I left it alone. +- **Frame points for dp:** + 1. **#14 is still the thing.** It is 14 days old with 21 commits. The 11-11 note answers the hold with the reviewer's own second exit. Merge it, or say which Status cell must change. The recommendation stands: a₀ "RECOVERED (known coincidence)", 1/φ "NOT EXCLUDED", S₈ "RETRODICTED". + 2. **The deflection question now has less at stake than it looked.** Whichever rule is chosen, the lensing leg fails without dark matter. What the rule decides is whether the framework *says* that, which is the TEST-15 / Boran requirement (light and GWs on the potential that moves matter). One FUNDAMENTALS sentence would settle it. + 3. Carried: DR4 reading key and plan before 2 December; the EFE-free corner's partition rule (the site explorer has since named Milgrom's, `82a480b`); `private-context/publisher/log.md` unwritten since 10-16. + +### 2026-11-11 (interactive maintenance pass): **The 11-09 review hold on #14 asked for the §5.15 status table to carry dated qualifications. No pass had answered it. The table now has a CHECKED note on three rows. Also: Brouwer et al. read in full (the lensing note's "abstract does not support that reach" was wrong), and the chemistry T_c row the site blanked as untraced is traced, with a γ that is computed from T_c. Stacked on #14. The count is HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`76a139e0`), clean. #14 OPEN, MERGEABLE/CLEAN at `6502602f`, CI SUCCESS, **one review: NOVA sweep, 2026-11-09 23:51Z, HOLD**. Fresh worktree `ai-workspace/.worktrees/syn-pub-1111` on `publisher/wp-2026-11-11`, branched from `6502602f`. Shared `web4` checkout still parked on `kimi/open-issue-audit-20261103` with uncommitted hub edits; read at `origin/main`. `synchronism-site` on `main`, read at `origin/main` = `b76d7f9`. +- **WAKE.** The 11-10 entry's frame point 1 says "merging it, or saying what blocks it, is worth more than another note". Something had said what blocks it before that pass ran: the NOVA review names the status table (a₀ DERIVED, 1/φ VALIDATED, S₈ PREDICTED) and offers two exits, a ruling from dp or "explicitly annotate the table with the dates/qualifications". The 11-10 pass read `reviewDecision` (empty, because the review is a COMMENT) and did not read the review body. Passes kept stacking notes at line 128 while the table at line 11 was the stated blocker. This pass takes the second exit first. +- **Window.** Synchronism after the 11-10 pass: **`76a139e0`** (11-10 06:18, back-annotation + proposal `lensing_leg_has_no_deflection_law_and_the_dr4_plan_fails_its_dry_run_20261110.md`). Site after `a70f8b8`: maintainer `2110d7c`/`6b05349`, explorer `9fd9639` (PREREG) → `f975060`/`b76d7f9` (Crater II against the EFE-free corner). +- **What reached the paper** (`git grep -c -F` per datum over live sections, control `MOND` = 58 lines in `dark_matter.md`): + - (i) lensing leg: `Brouwer` 1 site (§5.15, "The ceiling also dies independently"). **Reached.** + - (iv) chemistry T_c r = +0.948: 2 sites (§5.12 table, Appendix B.5). **Reached.** Covariant RG "4×10⁴σ": 1 site (the 10-23 CHECKED note). **Reached**, small. + - (ii) DR4 plan, (iii) Boufourou, Crater II / dSph, "Action-derived", "strongest independent", `deflection`: **0 hits**. The "11σ" hits are Chae's EFE detections, not the knee-exponent headline. Nothing to scope. +- **Status table (the hold).** One CHECKED note under the table, three bullets. Status cells untouched; the note says changing them is the project lead's call. + - a₀: ledger Bucket 3 row read at `PREDICTIONS.md` L1399 (1/2π against 1/6, 4.5%); the Υ dependence is the section's own 11-09 UPDATE, re-run on 11-10. + - 1/φ: read from `Research/Session239_Transition_Profile_Analysis.md` L50–93 (best fit 0.688, 1σ 0.609–0.802, χ² 2.95 / 6.96 / 2.20, reduced 0.33 / 0.77, so 9 dof) and `Session218` L98. **Not used:** the ledger's Session #45 quote. S45's "coincidence" is about a virial-scaling exponent B ≈ 1.62, a different quantity from the C(a) exponent, so it does not belong in a note on this row. **Not re-checked:** the Gaia points S239 fitted. + - S₈: `simulations/session102_s8_tension.py` re-run from `/tmp` (suppression factor **0.9425**; × 0.81 = 0.763). Session dated 2025-12-09. The DES Y3 / KiDS-1000 publication years (2020–2022) are from memory, not looked up. **Not relayed:** the ledger's "σ₈ ≈ 0.76 was calibrated to S8 in S102". The script calibrates `ratio_0` so that C_galactic(0) = 0.3; I found no step that tunes to S₈, so the note says "retrodiction" and stops there. +- **Brouwer et al. 2021, read in full for the first time in this lane** (ar5iv HTML via a fetch subagent; it returned verbatim fragments of at most 125 characters, so the footnote is quoted in pieces). Sec. 1: outermost radius at g_bar ≈ 10⁻¹⁵; footnote 3: that value counts stars and cold gas only, and hot gas, dust and missing baryons "could increase this value to g_bar ≈ 10⁻¹⁴"; bins 10⁻¹⁵ to 5×10⁻¹². Three consequences, all in the note. The 09-03 coverage caveat ("the abstract does not support that reach") was **wrong**: the site's 10⁻¹⁵ is the paper's fiducial. The allowance is about ×10, not the ×2 the paragraph used. And "moves ν by √2" is the √f error the 10-18 backlog listed as item 8 (required boost = g_obs/g_bar, which goes as 1/f). New script `simulations/publisher_20261111_lensing_ceiling_census.py`: at ×10 the section's mass-dependent ceilings still fail at 10¹⁰ and 10¹¹ solar masses (1.3× and 2.8× at 10⁻¹⁴; 4.1× and 8.7× at 10⁻¹⁵), and 10⁹ passes at 10⁻¹⁴. **Differs from the proposal:** it has the 6.39 cap passing at 10⁻¹⁴ under ×10 "with the 0.3 dex allowance" (5.5). Without that extra allowance my arithmetic gives 11, which is 1.7× short. The note prints only the section's own ceilings. The deflection condition is relayed as stated and gates on dp. Watch item closed in `whitepaper_sync.json` (stellar-mass binning still unread). +- **The T_c row is traced, and the site's "untraced" was a search miss.** The back-annotation says r = +0.948 "was attributed to Session 62, which has no superconductivity content (grep today)" and blanks it. Core Session 62 is biology (`simulations/session62_universal_coherence.py`). *Chemistry* session #62 is `simulations/chemistry/superconductivity_coherence.py`, commit `e4c903f5` (2026-01-17, "r=0.948 correlation"). Chemistry sessions live as scripts, not as `Research/` files. Traced, the row is weaker than blank: `estimate_gamma_SC` sets γ = 2 × 3.52 / (2Δ/k_B T_c), then multiplies by 0.8 for cuprates and 0.7 for hydrides, and correlates T_c with 1/γ over 28 materials. New control `simulations/publisher_20261111_tc_gamma_control.py` (executes the source's Parts 1–4 unchanged, writes nothing): identity r = **0.948**; category factor alone r = **0.944**; factors removed r = **0.882**; within category +0.57 (cuprates, n = 6), −0.16 (iron, n = 5), +0.73 (type II, n = 11). The label "Tc ∝ exp(−γ/λ)" is never fitted. One CHECKED note at both sites. **For the site lane:** the row can be restored with this provenance and this caveat; "untraced" should become "traced, γ built from T_c". +- **Covariant RG:** one bracketed NOTE labelling the σ figures as this program's unpublished computation. The 10-23 note already separated re-derived from relayed; it did not say the authors present the theory as viable. +- **Edits.** Words added, none removed (`--word-diff=porcelain` on `whitepaper/sections`: 0 removed tokens). Changelogs: top-level (+1), meta 05 and 09 (CRLF, written byte-wise). +- **Self-caught, build:** the first build reported `Missing character` **1**: the solar-mass symbol U+2609 inside the italic Brouwer note. FreeSerif Italic lacks it; the upright face has it. Rewritten as "solar masses". The glyph check is per face, so a glyph that is safe in body text can still drop inside an italic note. +- **Self-caught, gate:** three refusals (`mrh.command.unenumerated`: an unquoted shell variable holding a path or ref, twice, and a heredoc whose prose the gate read as a wide glob), no scope request opened; rewritten as script files under `/tmp`. One `git grep -o -E` with bounded repeats around the match hung for 120 s, the same hang the 11-07 and 11-10 entries record. Stopped; the Python context scan replaced it. +- **Build (verification only; artifacts restored by path, CI deploys):** md, pdf, web-clean rc 0 with `CI=true`, run from `whitepaper/`. `Missing character` **0**, U+FFFD **0**, PDF **307** pp (305 before). Notes in the text layer at pp 109 and 221 (T_c), 119–120 (table), 141 (RG), 150 (Brouwer); pages 120 and 150 rendered at 70 dpi and read. `section_5.html` carries 3 CHECKED + 1 NOTE, `section_9.html` 1; literal double asterisks and stray bracket-asterisk pairs **0** in both. PDF text-layer double-asterisk lines **26**, none on the note pages. `ρ^` **1** (baseline). Geometry: **1** word (baseline). +- **Cross-repo: Web4. No change, no rebuild owed.** `origin/main` = `d0b0dfda`. Window since the 11-06 entry (`bbaf2673`): #971 (hub release), #989 (web4-policy `matches` operator), #990 (first-user frontend PRD); `git diff --stat bbaf2673..origin/main -- web4-standard whitepaper docs/whitepaper-web` is **empty**. Open PRs touching `whitepaper/`: 0. `build/` = `docs/` by blob for all three artifacts (PDF `50ba1b5c`, monolith `05842421`, `index.html` `798a0d8d`). 24 pp, U+FFFD 0, geometry 8 words, max 596 pt (baseline). Pages status `built`. Recorded in `whitepaper_web4.json`; no web4 PR for a no-change pass. +- **So what.** Two things move. The hold on #14 now has an answer in the diff instead of one more note beside it, so the PR can be judged on what the reviewer asked for. And a row the site had given up on as unsourced has a source that settles it: the correlation is between T_c and a γ made from T_c. "Untraced" was a statement about where the search looked. +- **Frame points for dp:** + 1. **#14 needs a decision more than it needs passes.** It is 13 days and 20 commits, 307 pages against a published 292. If the table note is enough for NOVA's hold, merge. If the Status cells themselves must change, that is your ruling, and I recommend making it: a₀ "RECOVERED (known coincidence)", 1/φ "NOT EXCLUDED", S₈ "RETRODICTED". A table whose cells are contradicted by the paragraph under it is still the table people will quote. + 2. **The lane's loop is the problem, not its output.** Each pass waits for the site's daily back-annotation, finds the two or three sites it touches, and annotates. That converges on a paper that is all notes. The 10-18 raise (status banner, or a rewrite of the galaxy-sector verdict paragraphs) has had no answer for 24 days. I would stop per-item annotation once #14 merges and do one rewrite pass of §5.15's head and count list from the ledger. + 3. Deflection rule (new, from `76a139e0`): does the framework commit to light following the effective potential? One sentence in FUNDAMENTALS either way. + 4. Carried: DR4 reading key and plan before 2 December (the plan now fails its dry run); `private-context/publisher/log.md` unwritten since 10-16. + +### 2026-11-10 (interactive maintenance pass): **The paper's "+184 / +10.9" γ = 2 refutation holds Υ_disk = 0.5 for both laws. With one global Υ per law the pre-registered run gives +46.2 by point and +2.73 by galaxy, about 2σ, and the reading rule fixed before the run reads the row as disfavoured, not refuted. The paper's own count item (1) names a marginalised BIC as the registered statistic. Annotated at nine sites, stacked on #14. The count is HELD at 6. Web4: no change; the stuck Pages deployment was rebuilt.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`6fe7b696`), clean. #14 (`publisher/wp-2026-10-29`) OPEN, MERGEABLE/CLEAN, at `07cc72fd` (the 11-09 entry), CI SUCCESS, no review decision. Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1110` on `publisher/wp-2026-11-10`, branched from `07cc72fd`. Shared `web4` checkout is parked on `kimi/open-issue-audit-20261103` with 30+ uncommitted hub edits and was left alone; web4 read at `origin/main`. `synchronism-site` read at `origin/main` = `a70f8b8`. Fourteen `syn-pub-*` worktrees from 10-23 on are still attached (all stacked into #14); two `/tmp` worktrees are prunable. Left alone. +- **Window.** Synchronism after the 11-09 pass (04:49): **`6fe7b696`** (11-09 06:15, back-annotation + proposal `frozen_upsilon_headlines_and_the_first_prospective_test_needs_a_decision_20261109.md`). `synchronism-site` after it: maintainer `5a95614` (PREREG) → `1c6b374` (the run) → `4742931`/`5708200`; explorer `c31d20e` (PREREG) → `f69c8c5`/`a70f8b8` (DR4 plan dry run on DR3, returns no result). Open PRs: #14 only. +- **What reached the paper** (each datum searched with a Python context scan over live sections; `git grep -o -E '.{0,200}…'` hung again, as the 11-07 note says): + - (i) **+184 / +10.9 at frozen Υ.** Printed at **7** sites with the 10-18 SCOPED note (§5.15 L50 in full; copies in §0 L187/L192 and §7 L73/L81) or its short restatement (§7 L118, §8 L47), plus count item (1) at §5.15 L130, which names the registered statistic as "marginalised BIC on SPARC" beside "+184". The paper's own M/L paragraph (L146) already records that `sparc_tanhlog_profile.py`, behind item (1), shares the fixed Υ_disk = 0.5 / Υ_bulge = 0.7 convention. **Reached, at all 8 sites.** + - (iii) **"The scale is identified; what stays unidentified is the shape (the Υ ↔ γ degeneracy)"** is printed verbatim in the 10-04 CHECKED note at §5.15 L360. The back-annotation names it as the ledger's own 10-04 error. **Reached.** + - (ii) every γ is a MOND μ: the paper never prints the B2 column's "forced to differ from MOND" (0 hits), and the 08-04 SCOPED note already says +184 scores the compander "as a MOND interpolating function in acceleration". Folded into the (i) note as a limit, no separate site. + - (iv) wide binaries: `Saad`, `deprojection`, `Chae … Γ`, `0.65`: 0 hits beyond the existing TEST-02 sentences. (v) DR4: 0 hits; gates on dp. (vi) TEST-08's no-power number (r² ≈ 0.001 under MOND+EFE): item (4) makes no claim against MOND, so nothing to scope; `GEDE`/`Shafieloo`: 0 hits; the "η ≡ Abrikosov–Gor'kov" row (printed in §0 and §5.16) is flagged by the source as unverified and unread by it, so it is **untested, not wrong**, and left alone; the EFE-free-corner "≈ 3σ" withdrawal: 0 hits. +- **Re-run, not relayed:** `synchronism-site/maintainer/scripts/gamma2_global_upsilon_per_law.py` (PREREG `5a95614`, written before the script; reads only `MassModels_Lelli2016c.mrt`; `python3 -B -E -s`, 90 s) and `gamma2_global_upsilon_twin_power_posthoc.py`: both **byte-identical** to the committed `_output.txt`. Identity C1/C2 pass (+184.0, γ̂ = 0.489). P1–P6 all inside their registered bands; reading rule R1. The bootstrap sd of about 40 comes from the 16/84 percentiles (7.7, 87.4); the "about 2.2σ" is (44 − (−43))/40. Not re-run: nothing else; the H₀ closure numbers (92.8 / 71.1 / 56.5) are P5 of the same script. +- **Edits (words added, none removed; the word-diff's removed tokens are `+184;` and two `§5.15.]**,` whose punctuation moved behind a note).** (1) §5.15 L50: a full `[UPDATE 2026-11-09]` at the end of the 10-18 SCOPED note (the single-source scope every "+184" in the paper points at). (2) §0 ×2, §7 ×2: a short `[UPDATE 2026-11-09]` on each copy, pointing at §5.15. (3) §7 L118, §8 L47: the parenthetical "(ΔBIC +184 point-level; +10.9 galaxy-level, at threshold; …)" gains "+2.7 galaxy-level once each law gets its own Υ_disk, about 2σ". (4) §5.15 L130, count item (1): `[CHECKED 2026-11-09]`, the registered statistic was never the printed number; S661's own "M/L + distance marginalized" description is **not re-checked** against S661's code. (5) §5.15 L360: `[UPDATE 2026-11-09]` inside the 10-04 note, a₀′/γ is identified only with γ free; pinned at γ = ½ the scale moves 1.44 → 1.10 → 0.87×10⁻¹⁰ across Υ_disk 0.4–0.6 and cH₀/2π closes only at H₀ ≈ 93 / 71 / 57. Changelogs: `whitepaper/CHANGELOG.md` (+1) and meta 00/05/07/08 (05 and 08 CRLF, written byte-wise; CRLF counts 162 → 166 and 14 → 18). +- **Self-caught, markup (fixed before the second build):** the per-paragraph pandoc probe on the 9 edited lines found **4 literal `*`** at §0 L192 and 0 elsewhere. That copy of the SCOPED note sits inside an enclosing bold sentence; pandoc had been pairing the outer opener with the note's opener, and a `**[UPDATE …]**` (or a plain or italic bracket) placed after `]**` flipped the pairing. Seven toy variants tested: only "inside the bracket, before `]**`" and "after the comma, plain" render clean. The note at L192 now sits inside the SCOPED bracket (`… See §5.15. [UPDATE …]]**,`); the other three copies keep the bold form. Probe each copy separately: identical notes have different neighbours. +- **Build (verification only; artifacts restored by path, CI deploys):** md, pdf, web-clean rc 0 with `CI=true`, run from `whitepaper/`. `Missing character` **0**, U+FFFD **0**, PDF **305** pp (302 before). Notes in the text layer at pp 30/198/208 (short copies), 120 (full), 133 (item (1)), 157 (L360), 211/215 (parentheticals); pages 120, 133 and 157 rendered at 75 dpi and read. `section_5.html` carries 3 notes, `section_0/7.html` 2 each, `section_8.html` 1; body literal `**` and `*[` **0** in all four. PDF text-layer `**` lines **28**, all in changelog prose on pp 239–304, none on the note pages. `ρ^` **1** (baseline). Geometry (`xMax` > width − 36): **1** word, p6 (baseline). Artifact diff content = raw (no CRLF churn), restored. +- **Cross-repo: Web4. No change, one deployment repaired.** `origin/main` = `995d0aa5`, unchanged since 11-07; window since the 11-06 log entry (`bbaf2673`) is #971 (hub release) only, `web4-standard`/`whitepaper` diff **empty**. Open PRs touching `whitepaper/`: 0 (#984 hub mesh, #989 web4-policy `matches` operator, neither normative yet). `build/` = `docs/` by blob for all three artifacts. **The 11-07 `pages-build-deployment` failure, which the 11-08/11-09 entries carried as "errored", was a GitHub-side timeout**: the run log shows ten minutes at `syncing_files` then "Timeout reached, aborting!", and #971 touched no `docs/` file. This pass requested a rebuild (`gh api -X POST repos/dp-web4/web4/pages/builds`): built in 48 s at 12:37Z, Pages status `built`, served PDF Last-Modified 11-10 12:38, 149,271 B = committed size. The route is recorded in `whitepaper_web4.json`; no web4 PR opened for a no-change pass. +- **So what.** The paper's loudest refutation now carries, at every site, the number it gets when the nuisance that matters most is freed: about 2σ. That is the third downgrade of the same headline (+184 → +10.9 on 09-29, → +2.7 on 11-09), each from a convention the paper had printed as a scope note rather than a choice. Whether a 2σ row stays in a count of six is a ruling, not a measurement, and the paper says so. +- **Frame points for dp:** + 1. #14 is now **12 days and 12 `[Publisher]` commits** (plus 7 others) unmerged, MERGEABLE/CLEAN, CI passing. The published paper still stops at 10-27, so none of the +184 scoping since then is public. Merging it, or saying what blocks it, is worth more than another note. + 2. **Two decisions have a date.** The DR4 reading key (11-06) and analysis plan (11-09, draft) gate on dp before **2 December**; the explorer's dry run on DR3 already returned no result (calibration bin is a MOND regime), so the plan may need revision before it is adopted. And row #3's membership in the count is now explicitly the 10-07/10-13/10-16/10-18 counting-rule question; the range already carries −1. + 3. Unchanged from 11-09: the §5.15 status table (L11–23) still reads `a₀ = cH₀/(2π)` DERIVED with no dated annotation, while today's L360 note shows that closure holds at one Υ; the a₀ row edit gates on dp (10-16 (i)). `private-context/publisher/log.md` is still unwritten since 10-16. + 4. Not escalated, flagged: S661's "M/L + distance marginalized" (printed in §0 and §7) and the frozen pipeline's fixed Υ = 0.5 cannot both describe the same +184. Deciding which needs S661's code, which no pass has read. + +### 2026-11-09 (interactive maintenance pass): **The 11-08 back-annotation turned TEST-10's "≲ 14" into a bound of 7.5 (5.7 after correction). The paper's ε₀ rescue bullet still argued from the face value, and its TEST-08 row printed a point estimate against the claim. Three CHECKED notes and one markup-only list split, stacked on #14. The count is HELD at 6. Web4: no change.** + +- **Step 0:** shared Synchronism tree on `main` = `origin/main` (`82d9a3a8`), clean. #14 (`publisher/wp-2026-10-29`) OPEN, MERGEABLE/CLEAN, at `76133933`, CI SUCCESS, no review decision. Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1109` on `publisher/wp-2026-11-09`, branched from `76133933`. Shared `web4` checkout is parked on `kimi/open-issue-audit-20261103` with uncommitted hub edits and was left alone; web4 was read at `origin/main`. `synchronism-site` read at `origin/main` = `e1bb1fc`. +- **Window.** Synchronism after the 11-08 pass (04:38): **`82d9a3a8`** (11-08 06:21). `synchronism-site` after it: maintainer `0be9704` (PREREG), `63cb644`, `b6f147b`; explorer `704cd42`/`68abd6b` (PREREG), `6af84fe` (wide-binary mock injection). +- **What reached the paper.** Items were searched with `git grep` on sections excluding changelogs, meta and archive; the control `MOND` hits 56 lines in `dark_matter.md`. + - (ii) TEST-10: `B ≥ 13.7` at the ε₀ rescue bullet, `+67.8`, and count item (3). **Reached.** + - (iv) TEST-08: `r² = 0.0001 against a registered > 0.20` at count item (4). **Reached.** "~900×" itself: 0 hits. + - (i) a₀ prior art: `McCulloch`, `Smolin`, `same geometric`, `cH₀/6`: **0 hits**. Verlinde is printed only as the Path C reparametrization, not as a shared factor. Nothing to correct, but see the frame point on the status table. + - (iii) TEST-25 was already carried by the 11-08 note. Explorer `6af84fe`: `Saad`, `DR4`, `Desmond`'s wide-binary sentence and `Park+`: 0 hits; the paper prints no Saad & Ting number. +- **Re-run, not relayed:** + - `test10_bmax_3sigma_scan.py` and `test10_bmax_3sigma_diag_posthoc.py` (`python3 -B -E -s`; `git grep` found no file writes in them or their imported module): output **byte-identical** to the committed files. This repeats the arithmetic only; the 11-08 entry's warning applies. The bisection edges and the disc counts carry the limits stated in the note. + - New: `simulations/publisher_20261109_test08_upper_limit.py` + `_output.txt`, from the tracked per-galaxy table. Identity holds (r = +0.0118, N = 141). Fisher 95% upper limit on r² = 0.0312. Bootstrap 95% quantile of r² = 0.0299 on all three seeds, 20,000 draws each. Kill bar 2.9×, claim 6.4×. The point-estimate ratio is 649×, so the site's "~900×" also came from dividing by the rounded 0.0001. +- **The ε₀ note is this pass's own reading, not a relay.** The back-annotation does not mention the rescue bullet. 1/7.47 = 0.134 falls between the table's rows 0.120 (+30.8) and 0.156 (about +12). 1/5.71 = 0.175 falls between 0.156 and the optimum 0.220. The framework's 0.315 costs +7.3. The profile was **not re-solved** at those two values, and the note says so. It also says no σ attaches: the profile fixes Υ and does not marginalise D or i, while the bound profiles them. The rows are read from the site finding; 0.120's 157.3 was confirmed in the committed `_output.txt`. The other rows come from `epsilon0_left_edge_convergence.py`, which was not run here. +- **Relayed, not checked:** the "about 0.04 of total variance" reading of the TEST-08 claim. No script in the back-annotation computes it, and a site search did not locate one. The note labels it relayed. For scale, the offset variance is 0.875 of the pooled point variance (0.1244 vs 0.133 dex), so it does not follow from those two numbers alone. +- **Edits.** Three `[CHECKED 2026-11-09]` notes. Words added; the word-diff's two removed tokens are `0.685;` and `0.20;`, each moving its semicolon behind the new note. Separately, the 08-08 count list now has one blockquote paragraph per item, **markup only**: a scripted assertion confirms the text is identical after whitespace normalisation. Reason: page 133 showed the list as an unbroken italic block spanning pages, and this pass's notes made that worse. Changelogs: top-level (+1) and 05 meta (mixed CRLF, written byte-wise). +- **Build (verification only; artifacts restored by path, CI deploys):** md, pdf, web-clean rc 0 with `CI=true`, run from `whitepaper/`. `Missing character` **0**, U+FFFD **0**, PDF **302** pp (301 before). Notes on pp 132–133 and 144, plus the changelog on pp 255–256; pages 133 and 144 rendered and read before the list split; page 132 rendered and read after it, and the split also checked in the text layer and HTML (six paragraphs each). `section_5.html`: 3 notes, literal `**` **0**, literal `*[` **0**. PDF text-layer `**` lines **26**, none on the note pages. `ρ^` **1** (baseline). Geometry (`xMax` > width − 36): **1** word, p6 (baseline). +- **Self-caught, instrument:** a literal-string count reported 1 of 3 notes in `section_5.html`, and two needles as absent. Pandoc wraps HTML lines, so `CHECKED 2026-11-09` can span a newline; normalised, the count is 3. A PDF needle with a straight apostrophe also missed the typographic one. Normalise whitespace and avoid apostrophes in rendered-text needles. +- **Self-caught, gate:** this pass opened three needless scope requests (`scope-ca55f8112cec`, `scope-298bf3904385`, `scope-718fcc99ef97`): `/usr/bin/grep` and `/bin/grep` by absolute path, following this track's own instrument note, and a relative `cd ..` the gate resolved as `/home/dp`. The first two were ruled while the pass ran; neither act was re-issued, since `git grep` covers the need. The note in `manuscripts/publisher/CLAUDE.md` recommending `/usr/bin/grep -a` will trip the same refusal for the next pass. Not edited here, flagged. +- **Cross-repo: Web4. No-change verification.** `origin/main` = `995d0aa5`, unchanged since the 11-08 pass. Open PRs touching `whitepaper/`: **0**. PDF, monolith and `index.html` `build/` = `docs/` by blob. 24 pp, U+FFFD 0. Geometry **8** words, max 596 pt (baseline). **The Pages API now reports `errored`** after the 11-07 `pages-build-deployment` failure. Served PDF: HTTP 200, Last-Modified 11-07 03:50, 149,271 B = committed size, so the paper is not stale. No later deployment has run, so the 11-08 expectation that the next push redeploys it is still untested. +- **So what.** The useful result is a smaller claim: the paper's sharpest-sounding sentence against the ε₀ rescue depended on one galaxy's face value, and with significance attached the squeeze is about the size of what the derived value already costs. The class verdict does not move (126.5 vs MOND's 52.2 on the same points). Nothing here can change the count, which is the 11-08 proposal's point (a). +- **Frame points for dp:** + 1. #14 now holds 11 days and 11 `[Publisher]` commits (plus 7 others) that no reader has seen; the published paper stops at 10-27. It is MERGEABLE and CLEAN with CI passing, so the only thing holding it is a review decision. Merging it, or ruling on what blocks it, is worth more than the next note. + 2. The notes accumulate at line 128 while the §5.15 **status table (L11–23) still reads** `a₀ = cH₀/(2π)` **DERIVED**, 1/φ **VALIDATED**, and S₈ **PREDICTED … Independent validation**, with no dated annotation on those rows. The first thing a reader sees in the section disagrees with what the section concludes, and today's a₀ read says cH₀/2π is the form the MOND literature itself prints. The a₀ row edit gates on dp (10-16 (i)), so it was not touched. Fixing the table may do more than any single note. + 3. `private-context/publisher/log.md`, which this track's CLAUDE.md marks as required on every run, was last written 10-16. Passes have stopped writing it, so either the supervisor's health read is stale or the rule is. Not written here; dp's call which one to change. + 4. Still gated, unchanged: the 10-18 backlog scale question, the convention-dependent kills, the count rule against the Bell row, the DR4 reading key (needs a decision before 2 December), the signal-gate proposal `1ae74f9c`. New from `82d9a3a8`: the correction-retirement rule and the verdict-sentence error-rate study. + +### 2026-11-08 (interactive maintenance pass): **Yesterday's note on §5.15 count item (5) needs two limits. "Every γ from 0.3" reaches too far: the γ = 0.3 and 0.35 rows are solver noise. And every Cassini σ pins Υ: the compander passes only at Υ_d ≈ 1.2–1.4, while MOND's δ = 6 passes at Υ_d ≈ 0.54. An UPDATE is stacked on #14. The count is HELD at 6. Web4: no change.** + +- **Step 0:** shared tree on `main` = `origin/main` (`1ae74f9c`), clean. #14 (`publisher/wp-2026-10-29`) OPEN, MERGEABLE/CLEAN, at `d5692c4f` (the 11-07 entry), CI SUCCESS, no review decision. Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1108` on `publisher/wp-2026-11-08`, branched from `d5692c4f`. +- **Window, both repos.** Synchronism after the 11-07 pass (04:37): **`cf84bb8c`** (11-07 06:09, back-annotation) and `1ae74f9c` (process proposal on the CBP signal gate, no research claims, gates on dp, nothing in the paper). `synchronism-site` after the 11-07 pass: maintainer `d7eb6b2`/`52db8d5` (the source of `cf84bb8c`), explorer **`10d0bdf` (PREREG) + `2db4494`** (Cassini with Υ free) + `79b5938`. No back-annotation has carried the explorer finding yet. +- **What reached the paper.** + - `cf84bb8c` (ii) says the 11-06 γ = 0.3 row is solver noise. This PR's 11-07 note prints "14.5–19.3σ for every γ from 0.3 to 3", in bold. The source block itself says to read the range with that caveat. + - `2db4494` says every Cassini σ in the ledger pins Υ, and that z mostly tracks the population mass-to-light zero point through a₀. The 11-07 note said the row "does not depend on the γ ≈ ½ pin". That holds, but it does depend on Υ_d near 0.5, and the note did not say so. + - `cf84bb8c` (iii) and `2db4494` show the shared MOND root holds only at δ = 1 and γ ≈ ½. The paper reads the Cassini +17.95σ as a property of "a MOND-family interpolating function", so it needs this. The word "inherited" is **0** hits; a δ-family is **0** hits before this note, so the note defines δ. +- **Re-run from /tmp copies** (`python3 -B -E -s`; the site tree is not written): + - `compander_cassini_nobulge_profiled_a0.py`: all table rows match. The recorded file has an extra 13-line hand-added convergence note that the code doesn't print. + - `cassini_both_families_upsilon_free.py`: 16 lines differ, only in the elapsed-time token (`t=55s` vs `t=84s`). All values match. + - `cassini_z_vs_population_upsilon_posthoc.py`, `cassini_pass_overmaximal_posthoc.py`, `cassini_pass_config_cost_galaxy_boot_posthoc.py 2000`: byte-identical. +- **New: the convergence scan** (`simulations/publisher_20261108_cassini_convergence_scan.py` + `_output.txt`). `cf84bb8c` only vouches for γ ≥ 0.489, and tests only z at 2.32, leaving the note's γ = 0.35–0.45 unknown. Twelve draws (a₀ × 0.999/1/1.001/1.003 across orders 128/512, 192/768, 256/1024) at all three g_ext, bulges in: + - γ = 0.3: spread 8–20σ, draws 8.4–28.0. γ = 0.35: spread 5–6σ, draws 10.0–23.6. All > 5, so the verdict holds, but the numbers are unusable. + - γ = 0.4: spread 0.19 / 0.29 / 0.56. γ = 0.45: ≤ 0.12. γ = 0.489: 0.04 / 0.06 / **0.54**. γ ≥ 0.6: ≤ 0.09. + - The 0.54 is a single draw (19.87 at the unmodified a₀ on 128/512) against 19.3–19.4 in the other eleven. It moves the top of the printed range if that draw is the one you take, so the note says "at most 0.6σ". + - **Not what I expected:** I first read g_ext/a₀ ≈ 7 as the boundary. The scan refutes that: 6.17 at γ = 0.35 is unstable and 6.23 at γ = 0.4 is not. The note states the limit in γ. Why the solver loses stability below γ ≈ 0.4 is untested. + - **Prediction check:** "0.3 spreads by several σ" held, "0.489+ by ≤ 0.05" **failed** on that one spike and marginally elsewhere (0.06). +- **Edit (one bracket added; the word-diff's one removed token is the note's closing token moving behind the UPDATE; no words removed).** `[UPDATE 2026-11-08]` at the end of the `[CHECKED 2026-11-07]` note. The 11-07 sentence stays, as the 11-05 UPDATE did, so the correction is on the record. Changelogs: top-level (+1) and 05 meta (CRLF, written byte-wise, CRLF 154 → 158). +- **Deliberately not printed:** "the compander stays above 10σ with Υ free". The finding warns against it: P3 held by 0.03σ at w = 0.1 and fails at w = 0.2 (+7.09). The robust statement is the Υ_d needed to pass, and that is what the note gives. +- **Relayed, not checked:** the Desmond, Hees & Famaey 2024 figure (1.9σ); the explorer read their abstract and part of the HTML. The stellar-population value as 0.5: the "2.4–2.8 times" is relative to the paper's pinned Υ_d, and the finding did not locate a sourced 3.6 μm zero-point uncertainty, so the note gives no σ for it. +- **Build (verification only; artifacts restored by explicit path, CI deploys):** md, pdf and web-clean rc 0 with `CI=true`, run from `whitepaper/` (the absolute path alone trips the cwd guard). `Missing character` **0**, U+FFFD **0**, PDF **301** pp (unchanged). The note is on p133; rendered and read at 170 dpi. Literal `**` in the note's text layer **0**; `section_5.html` literal `**` **0**. `ρ^` **1** (the baseline). Geometry (`xMax` > width − 36): **1** word, p6, the baseline. +- **Cross-repo: Web4. No-change verification.** `origin/main` = `995d0aa5`. Window `bbaf2673..995d0aa5` is #971 only (hub release: one workflow, `hub/mirror`, one script); `web4-standard` and `whitepaper` diff **empty**. Open PRs touching `whitepaper/`: **0**. Committed artifacts: PDF, monolith and `index.html` `build/` = `docs/` by blob. 24 pp, U+FFFD 0, "byte for byte" present. Geometry **8** words, max 596 pt (the baseline, pp 6/12/13/15/22). **`pages-build-deployment` failed on `995d0aa5`**, with a GitHub timeout during `syncing_files`. The served PDF is unaffected (HTTP 200, Last-Modified 11-07 03:50, 149,271 B = committed size), and since that commit has no paper change nothing is stale. The next push to `main` should redeploy it. No edit, no rebuild owed. +- **So what.** A byte-identical re-run certified a noisy number yesterday. The 11-07 pass re-ran the script, matched every digit, passed the identity control, and printed "every γ from 0.3" on that basis. Same a₀ digits and same order give the same noise, so a reproduction can't see non-convergence, the same blindness the 09-03 entry found for mislabelled inputs. **A re-run proves the arithmetic repeated, not that the number is stable: perturb it before printing a range's endpoints.** The cost was one extra scan. +- **Frame point for dp:** #14 now holds every pass since 10-29, and the published paper has none of them: `main`'s last paper commit is 10-27. The last status comment on #14 about this (11-03) says review and operator-decision holds remain. Each pass adds notes to a queue nobody reads from, and this one corrects a note no reader has seen. A merge, or a ruling on what is holding it, would do more for the paper than another annotation. Still gated on dp, unchanged: the 10-18 outage-backlog scale question, the convention-dependent kills, the count rule against the Bell row, and the DR4 reading key. New: the signal-gate proposal (`1ae74f9c`). + +### 2026-11-07 (interactive maintenance pass): **One CHECKED note on §5.15 count item (5), the SPARC × Cassini squeeze. The +17.95σ the paper prints is at γ = 0.489. With a₀ profiled from SPARC, Cassini fails at 14.5–19.3σ for every γ from 0.3 to 3. So the row does not hang on the γ ≈ ½ pin, and `C_g`'s two refutations now coincide at γ = 2.** + +- **Step 0:** shared tree on `main` = `origin/main` (`330d49d1`), clean. #14 (`publisher/wp-2026-10-29`) OPEN, MERGEABLE/CLEAN, at `4a292aa9` (the 11-06 entry), CI SUCCESS. The run log was read, not just the green check (build-gotchas #8): the PDF built and was copied. Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1107` on `publisher/wp-2026-11-07`, branched from `4a292aa9`. +- **Window, both repos (build-gotchas #12).** Synchronism after `a850db9e`: **`330d49d1`** (11-06 06:12, back-annotation). It landed 91 minutes after the 11-06 pass (04:41), so no pass had read it. `synchronism-site` after the 11-06 pass: visitor `131b089`, maintainer `df896fd`/`b480f5c` (the source of `330d49d1`), and explorer `a4d99f4` (PREREG) + `79f213e` (MOND δ-family Cassini at SPARC-profiled a₀) + `2222368`. +- **`330d49d1`, each datum grepped alone over live sections, meta excluded:** + - (i) the fixed-a₀ "passes at γ ≳ 1.5–2": `1.5–2`, `γ ≳` **0**. The count range `3–8`/`4–8`: 0 count-range hits; the 6 `3–8` hits are all Z ranges in chemistry. `range on file` **0**. The paper never printed the withdrawn claim or the range. **But** it prints Cassini tied to γ ≈ ½ in two places: count item (5) "+17.95σ" (L128), and the "(4) The six refutations" decomposition (L97), "acceleration-keyed `C_g` (RAR shape at γ = 2, Cassini at γ ≈ ½)". That decomposition reads as two refutations at two different γ. With SPARC's a₀ they coincide. This is an under-statement, not an over-claim, and it lands. + - (ii) DR4 survival window: `DR4` **0**. The reading key gates on the project lead, so no edit. + - (iii) site-only record (TEST-09 paired σ, the two 2σ rules): the 11-06 pass already carries the σ. The paper prints neither rule. +- **Re-run.** `cassini_q2_at_sparc_profiled_a0_high_gamma.py` was copied to `/tmp/pubchk/cass` and run with `python3 -E -s`. It reads the Synchronism `simulations/` and `sparc_profile.json` in place by absolute path and writes nothing. rc 0, output **byte-identical** to the committed `_output.txt`. All three predictions HELD. Identity control +17.96σ at γ = 0.489 (registered +17.95σ). +- **Edit (`git diff --word-diff=porcelain` removed tokens: 1, the `;` after `+17.95σ` moving behind the note; no words removed).** CHECKED note after item (5)'s "+17.95σ": the registered test evaluated Cassini only in 0.425–0.600; profiled-a₀ result 14.5–19.3σ over γ 0.3–3 at all three registered g_ext; the row does not depend on the γ ≈ ½ pin; at γ = 2 both RAR shape (ΔBIC +184) and Cassini (14.9–16.8σ) exclude `C_g`; the ledger's fixed-a₀ pass was never printed here; still conditional on a QUMOND completion; count HELD at 6. Changelogs: top-level (+1) and 05 meta (CRLF, written byte-wise, CRLF 150 → 154). +- **Not integrated:** explorer `79f213e` (MOND's own δ-family at SPARC-profiled a₀). The paper discusses no δ-family. It bears on MOND, not on this family. +- **Build (verification only; artifacts restored by explicit path, CI deploys):** md, pdf and web-clean rc 0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **301** pp (300 before). The note is on p132–133; p132 rendered and read. Literal `**` in the note's text layer: **0**. `ρ^` **1** (the baseline). Geometry (`xMax` > width − 36): **1** word, p6, the baseline. +- **Cross-repo: Web4. No-change verification.** `origin/main` = `bbaf2673` (#978 merged 11-06 19:49 with the 11-06 HOLD repair). Window `4d898394..bbaf2673` contains #978 only; `core-spec`/`ontology` changes **0**; open PRs touching `whitepaper/` or `core-spec`: **0**. Committed artifacts on `main`: PDF and HTML `build/` = `docs/` (`cmp`), 24 pp, U+FFFD 0, "byte for byte" present. Geometry **8** words, max 596 pt (the baseline). The pages are now 6/12/13/15/22: the 11-05 text moved p11's overrun to p12. All four `main` workflows are SUCCESS on `bbaf2673`. No edit, no rebuild owed. This entry stands for both repos, so web4's log gets no separate commit. +- **So what.** The last five passes all found the paper printing something *more* firmly than its data. This one runs the other way: the paper scoped a refutation *more narrowly* than the data now support. Ledger hygiene has to catch under-statements too, because the "6 is stable because its conventions are unregistered" pattern (11-06) cuts both ways. Here, one row's dependence on a convention (the γ pin) turns out to be absent. Still gated on dp, unchanged: the 10-18 outage-backlog scale question, the convention-dependent kills in the headline, the count rule against the Bell row, the descriptive "executed exclusions" line, and now the DR4 reading key. + +### 2026-11-06 (interactive maintenance pass): **Two CHECKED notes in §5.15, both from site work no pass had read. (1) The "Dark Energy from Coherence" block reads G_eff = G/C as a cosmological G. Read that way, local G drifts at 3(1 − Ω_m)H₀ and lunar laser ranging excludes it by at least 527× over γ 0.3–3. Only the fluid reading survives. (2) TEST-09 fires on its registered point-estimate rule, but the margin over the 0.3 bar is 0.95σ on a paired galaxy bootstrap.** + +- **Step 0:** shared tree on `main` = `origin/main` (`a850db9e`), clean. #14 (`publisher/wp-2026-10-29`) OPEN, MERGEABLE/CLEAN, at `dfd94cb8` (the 11-05 entry). Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1106` on `publisher/wp-2026-11-06`, branched from `dfd94cb8`. +- **Window, both repos (build-gotchas #12).** Synchronism after `668f69d8`: **`a850db9e`** (11-05 06:07, back-annotation: (i) modified-G DE reading vs LLR; (ii) the density item that the 11-05 pass integrated from the site directly; (iii) site-only count-by-kind record). `synchronism-site` after `935862e`: visitor `48c61cc`, maintainer `6a13ba7`/`827e97e` (the source of `a850db9e`), explorer **`c9ff6ab`** (PREREG, 08:02) + **`da1308c`** (finding, 08:04) + `ed14e78` (log). The explorer finding landed after the 11-05 pass (04:40) and after the back-annotation (06:07), so no pass or back-annotation has carried it. +- **(1) `a850db9e` (i) lands.** §5.15 "Dark Energy from Coherence" prints *"Applying G_eff = G/C to cosmology yields emergent dark energy"* with no note. The paper's existing LLR failure (covariant-branch paragraph, |Ġ/G| ≥ 0.99 H₀) belongs to the Brans–Dicke completion, a different theory. Re-ran `synchronism-site/maintainer/scripts/de_modified_g_reading_vs_llr.py` (`python3 -I`, pure `math`): rc 0, output **byte-identical** to the committed `_output.txt`, identity control 5.5×10⁻¹¹. CHECKED note added after "No cosmological constant Λ required.": the split survives only as a fluid (Ġ = 0); as modified gravity, Ġ/G = 3(1 − Ω_m)H₀ = 1.42×10⁻¹⁰ yr⁻¹ at γ = ½, at least 527× the LLR 2σ edge for every γ from 0.3 to 3. The LLR value is the site's quote of Hofmann & Müller 2018, which I did not re-read, and the note says so. The explorer's prior-art lead (Ġ/G ~ H₀ is probably a known class result) is carried in the note as "an application, not a new bound", with its locators marked unverified. +- **(2) `da1308c` lands.** The paper prints no σ for TEST-09 anywhere. Count item (2) gives "|Δn| = 0.41 against a registered bar of 0.3", and the 10-29/10-31 notes give the floor-convention dependence. Re-ran `test09_paired_bootstrap_delta_n.py` from a `/tmp/t9run` copy (site layout mirrored, `Synchronism` symlinked so the self-resolving SPARC path works, `python3 -E -s`): rc 0, output **byte-identical**, unpaired identity control PASSES (3.753 / 3.346 / 3.456 / 3.492), all five pre-registered predictions HELD. The PREREG commit (`c9ff6ab`, 08:02) precedes the script commit (08:04). CHECKED note on item (2): the registered rule (point estimate vs 0.3) fires; paired σ(Δn) = 0.112 (ρ = +0.14), so |Δn| = 0.406 is 3.6σ from zero but 0.95σ above the bar, and 16% of resamples fall at or under 0.3. A bar-excluded-at-2σ rule would not fire, which is a second convention on which TEST-09 leaves the count. Count HELD at 6. The site's "one name, two 2σ rules" defect is the site's to fix; the paper prints neither rule. +- **Edits (`git diff --word-diff=porcelain` removed tokens: 1, `0.3;`, which is the `;` moving after the inserted note; no words removed).** Changelogs: top-level `whitepaper/CHANGELOG.md` (+1) and 05 meta (CRLF, edited byte-wise; CRLF count verified). +- **Build (verification only; CI builds and deploys on merge; artifacts restored by explicit path):** md, pdf and web-clean rc 0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **300** pp (299 before; the two notes add a page). The LLR note is on p151 and the TEST-09 note on p132; both pages were rendered and read. `ρ̄` renders. `section_5.html`: both notes present, literal `**` after tag strip **0**, literal `ρ^` **0**. Geometry (`xMax` > width − 36): **1** word, p6, the baseline. +- **Cross-repo: Web4. #978 HOLD repaired.** Sweep review (11-06 08:41) held #978 on one sentence: counterparty-signed MRH records are something "a copy of the key does not carry with it". A copied bundle can carry them byte for byte. The repair follows spec §3.4.6/§8.4, not the review's suggested "cannot extend": the spec says a copy may obtain *new* signatures. The new wording is that the copy can carry existing records and collect new ones but cannot make the counterparties' signatures, and that divergence and lineage continuity can tell holders apart, weighed under law, while a passive or mirrored copy may show neither. Rebuilt; six artifacts committed; 24 pp; geometry 8 (baseline). Entry in that repo's log. +- **So what.** Both items are the paper printing a claim more firmly than its own data supports, but in opposite directions. The DE block printed a *mechanism* ("G_eff = G/C yields dark energy") whose natural reading is excluded by three orders of magnitude, and only the weaker fluid reading survives. TEST-09 printed a *kill* whose margin is under 1σ. Neither moves the count, because both hinge on conventions the count does not fix (which G, which significance rule). That is the pattern the gated items to dp keep surfacing: the 6 is stable because its conventions are unregistered, not because the evidence is decisive on each. Still gated on dp, unchanged: the 10-18 outage-backlog scale question, convention-dependent kills in the headline (TEST-09 now has two), the count rule against the Bell row, and whether a descriptive "executed exclusions outside the count" line goes beside the 6. + +### 2026-11-05 (interactive maintenance pass): **Yesterday's note said the free density fit keeps a response that is "weak but present" (boost ∝ ρ^−0.10^, +945 over a constant boost). Two hours after that pass, the site explorer tested whether that response is g_bar seen through ρ. It is: at fixed g_bar, density adds nothing. The note now carries an UPDATE. The miss was this lane's own: the section already printed the 08-02 form-free null six lines below.** + +- **Step 0:** shared tree on `main` = `origin/main` (`668f69d8`), clean. #14 (`publisher/wp-2026-10-29`) is OPEN, MERGEABLE/CLEAN, at `fa0e7a20` (the 11-04 entry). Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1105` on `publisher/wp-2026-11-05`, branched from `fa0e7a20`. +- **Window, both repos (build-gotchas #12).** Synchronism `58428709..668f69d8`: `6fd7f521` (11-04 06:08, back-annotation) and `668f69d8` (11-04 09:00, an attributed addendum to the 11-01 world-model methodology note in `forum/`; out of scope, no claims changed). `synchronism-site` after `333ff7e`: maintainer `3c5a3be` (the source of `6fd7f521`), visitor `4a172d8`, and explorer **`4891f7e`** (PREREG, 08:01) + **`2ba2480`** (finding, 08:05). The finding landed after the 11-04 pass (04:38) and after the back-annotation (06:08), so no pass and no back-annotation has carried it. +- **`6fd7f521`, each datum grepped alone over live sections, meta excluded:** + - (i) density fit weak, not off: integrated by the 11-04 pass from the site finding directly. + - (ii) the count measures registrations, not framework content. Proposed descriptive line beside the 6: **gates on the project lead**, no edit. The paper already prints the three exclusions it names, each in place (§5.15 L167 `completion B … Δχ² ≥ +79`; L346 `A: χ² ≈ 9,900`, `B: Δχ² ≥ +79`), but not as a line beside the count. + - §3 site-only records: the glossary kill examples (`R²=0.14` 2 hits, both the S639 disambiguation sentence, which already says the 0.14 is mislabelled; `≈ 0.55`, `Kill Criterion Triggered` **0**); TEST-25 `8.7σ` **0**; TEST-02 absolute dynamics (`Oort`, `wide/close`, `identically null` **0**). The site item (`3c5a3be`) is a wording fix for a visitor who read "identically null" as "untested". The wide/close ratio is null, and the published-knee branch is excluded by Oort (boost ~3.5×10⁴ at the Sun). The paper's TEST-02 text is S654/S685's MOND+EFE reduction and never says "identically null". Nothing printed, nothing to correct. +- **`2ba2480`, this one lands.** On the same 2,438 points / 122 galaxies that produced "+945" and "ρ^−0.10": + - a log ρ term added to the refit acceleration-keyed law takes β = −0.001 (ΔBIC +7.7 against adding it); + - against a free cubic in log g_bar it takes +0.003, galaxy-bootstrap CI [−0.022, +0.027]; + - injected −0.05 and −0.10 are recovered at −0.047 and −0.097; + - corr(log g_bar, log ρ) = +0.84, and the data's marginal slope of log B on log ρ, −0.211, is reproduced to 0.2% by the g_bar-only model. + + The paper's sites: `ρ^−0.10` **1** (the 11-04 note, L95), `+945` **1** (same). The section's own L101 (*"The substitution is what the data exclude"*, 08-02, α ∈ [0.75, 1.00]) already said local density carries at most 25% weight. The 11-04 note sat between them and did not cite it. +- **Re-run, not relayed.** Copied `explorer/{scripts,findings/scripts}` to `/tmp/densrun`, repointed `rar_scatter_nogo_real_sparc.py` `BASE` to `ai-workspace/Synchronism/simulations/sparc_real_data` (the committed `/mnt/c` path is absent on cbp), and ran `density_partial_beyond_gbar.py` with `python3 -E -s` (sibling imports). rc 0. The output is **byte-identical** to the recorded `density_partial_beyond_gbar_output.txt`, identity controls included (G −7564.2, F −4721.7). The unregistered between-galaxy helper was not run (the finding marks it unregistered and the note does not use it). +- **Edit (words added, none removed; `git diff --word-diff=porcelain` removed tokens: 0):** an `[UPDATE 2026-11-05]` at the end of note (3), after the 11-04 CHECKED note. It says: the weak response belongs to the fit, not the data; the partial slopes and CI; the positive controls; the +0.84 correlation and the 0.2% reproduction; ρ^−0.10^ and +945 are g_bar's information reaching the fit through ρ, not evidence of density keying; and the form-free 08-02 paragraph below already said so. Verdict and the count of 6 unchanged. Changelogs: top-level `whitepaper/CHANGELOG.md` (+1) and 05 meta (CRLF, edited byte-wise). +- **Markup fix, self-caught in the first build:** the 05 meta changelog entry printed `ρ^−0.10,` with a literal caret, because the superscript was not closed. The 11-04 entry in the same file had the same defect, already in #14's PDF (text-layer line "∝ ρ^−0.10, and"). Both closed to `ρ^−0.10^` (markup only, no words changed), plus the new top-level line. After rebuild, literal `ρ^` in the text layer: **1**, the pre-existing `P ∝ ρ^γ` in open questions, not this lane's. +- **Build (verification only; CI builds and deploys on merge; artifacts restored by explicit path):** md, pdf and web-clean rc 0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **299** pp (unchanged). The note spans p124–125; both pages rendered and read, the note follows "unchanged." and precedes (4), and the superscript renders. `section_5.html`: the note is present, literal `**` after tag strip **0**, literal `ρ^` **0**. Geometry (`xMax` > width − 36): **1** word, p6, the baseline. +- **Cross-repo: Web4. One content change, web4#978.** #819 merged at 02:18 today. It makes LCT spec §3.4.6 normative (identifier uniqueness ≠ presence uniqueness; MRH evidence weighed by signer independence, not monotonic in size) and §8.4 (duplicate/spoof adjudication, classes A–E, UNRESOLVED outcome). §6 of the paper said "its uniqueness rests on key-generation entropy" and "every witness link makes the presence harder to forge". Words added to both; artifacts rebuilt and committed; 24 pp, geometry 8 (baseline). One sentence is flagged for dp, not edited ("every witnessed interaction traces back to the entity that actually participated", against class C). Entry in that repo's log. +- **So what.** This corrects this lane's own note from yesterday, not the site. The 11-04 pass re-ran the knee fit, confirmed "+945", and wrote "weak but present". It did not ask what +945 is measured *against*. A constant boost has no g_bar dependence either, so it was the wrong null for a density claim when g_bar and ρ correlate at 0.84. The source finding had flagged the question as open ("is ρ^−0.10 just regression dilution of g_bar?"), and the 11-04 log carried it as an open thread, but the paper note did not carry that hedge. Lesson for the next annotation: when a note promotes a fitted response to a statement about the data, check the section for a form-free result on the same variable first. Here it was six lines away. Still gated on dp, unchanged: the 10-18 outage-backlog scale question, convention-dependent kills in the headline, the count rule against the Bell row, and now (6fd7f521 (ii)) whether a descriptive "executed exclusions outside the count" line goes beside the 6. + +### 2026-11-04 (interactive maintenance pass): **§5.15 note (3) glossed the free density fit's γ_ρ = 0.046 as "the likelihood switching the density dependence *off*". A site finding from 11-02 08:06, which no pass had read, shows the fit's knee sits below every disc and the response is weak but on (boost ∝ ρ^−0.10; a constant boost loses by ΔBIC +945). Annotated CHECKED after a re-run. One provenance gap: the finding's supplementary output has two lines its committed code does not print.** + +- **Step 0:** shared tree on `main` = `origin/main` (`58428709`), clean. #14 (`publisher/wp-2026-10-29`) still OPEN at `82f8c362` (the 11-03 entry). Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1104`, branched from `82f8c362`. +- **Window.** Synchronism `origin/main`: nothing after `58428709`, which the 11-03 pass already covered (`d91efc49` out of scope, `5709b579` integrated). `synchronism-site` `origin/main`: the last commits are `c62bc9d` (PREREG, 11-02 08:01), **`75e6a15`** (explorer finding, 11-02 08:06) and `333ff7e` (explorer log). The last back-annotation into this repo, `5709b579`, is 11-02 06:10, two hours earlier. So the finding `density-law-knee-sits-below-every-disc-and-the-formula-A-fits-as-newton.md` reached no pass and no back-annotation. `grep -F` on this log for `75e6a15`, `free knee` and `density_law_knee`: **0**. +- **Paper sweep, each datum on its own, live sections, meta excluded:** `switching the density` **1** (dark_matter.md L95, note (3)); `switched off`/`switch off`/`switches off` **0**; `γ_ρ` **1** (the same line); `γ/A` **0**; `A·V²` **0**; `constant boost` **0**; `0.046` 6 hits, but the 5 outside L95 are Chae's 0.046–0.083 dex EFE signal, a different object; `2,843` **1** (L95, cited as the source's 122-galaxy figure). The paper never prints the site's "+2843 rests on A = 0.029" premise or the "only γ/A identified" gloss, so the finding's Implication 1 has nothing to correct here. Its Implication 2 ("switches its own density dependence off" overstates) has exactly one site in the paper. +- **Re-run, not relayed.** The site scripts hardcode `BASE = /mnt/c/exe/projects/ai-agents/...`, which does not exist on cbp. I copied `explorer/{scripts,findings/scripts}` to `/tmp/kneerun`, pointed the copy at `ai-workspace/Synchronism/simulations/sparc_real_data`, and ran it there, so the tracked `.npy` in the site tree was not rewritten. `density_law_knee_placement.py` (`python3 -I`, rc 0, 23 s): output **byte-identical** to the recorded `_output.txt`. F: γ 0.0462, ρ_c 4.400×10⁻⁶ M☉/pc³, ΔBIC +2842.6. C0 (constant boost 2.612): +3787.5, so C0 − F = **+944.9**. Response exponent median 0.1012 [0.0639, 0.1685]. Hessian corr(ln γ, ln ρ_c) +0.8926. `_supp.py` (needs its sibling on the path, so `-E -s`, not `-I`; rc 0): reproduces +6593.0 (A fixed, γ free) and Newton +6636.0. **But the recorded `_supp_output.txt` has two more lines than the committed script prints:** `gamma=2, A=4.6e-5 … +10371.5` and `C0 - F … gal-sd 154.8, z 6.2; BIC diff +944.9`. No code in `75e6a15` produces them. +944.9 follows from the main script's own rows. I recomputed +10 371.5 independently (`/tmp/kneecheck/g2_formulaA.py`: γ = 2, ρ_crit = 4.6×10⁻⁵·V², σ profiled; BIC 2830.6 against B's −7540.9; σ 1.082, C median 0.088, boost median 11.39, all as recorded). The galaxy-level z ≈ 6.2 (gal-sd 154.8) was **not** reproduced and is printed as the source's. +- **Edit (words added, none removed; `git diff --word-diff=porcelain` removed tokens: 0):** a `[CHECKED 2026-11-04]` note at the end of note (3) in `dark_matter.md`. It says the "off" gloss overstates; the free knee sits more than 100× below 97.4% of disc points, in the log regime where γ and ρ_crit are separately identified; boost ∝ ρ^−0.10 and a constant boost loses by +945. It also sizes the derivation's calibration, which the paragraph never priced: at A = 4.6×10⁻⁵ the best γ fits as Newton (+6,593; Newton +6,636), and γ = 2 at that A gives +10,372 (the paragraph's +5,957 frees the knee). It names the supplementary script's gap. Verdict, "it loses" and the count of 6 are unchanged. The exponent is written `ρ^−0.10^` (pandoc superscript). The first build printed a literal caret, which I caught in the text layer. Changelogs: 05 (CRLF, edited byte-wise, 138 → 142 CRLF lines), top-level (+1). +- **Build (verification only; artifacts restored by explicit path, since CI builds and deploys on merge):** md, pdf and web-clean rc 0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **299** pp (298 → 299). Text layer: the note is on p124 and reads `ρ−0.10` (superscript); HTML `section_5.html` carries `ρ−0.10`. Page 124 rendered and read: the note sits flush after "it loses." and before (4). Pandoc → plain on the section: **0** literal `**` after the new tag. Geometry (`xMax` > 576 on 612 pt): **1** word, p6 `(density/velocity/pressure/viscosity)`, the recorded baseline. +- **web4 leg (no PR, recorded here):** `origin/main` `41971895..c643fc0d` is hub held-ignition I1–I3 (#966–#970), #975/#976 hub, #974 (Kimi open-issue audit) and #973 itself. `git diff --stat` over `web4-standard/core-spec`, `whitepaper` and `docs/whitepaper-web` shows only #973's own log. **0** spec changes. The paper's §11 Hub paragraph ("reference proof-of-concept, `0.1.0-alpha.0`, pilot-ready, first community-chapter deployment still ahead") still matches `hub/README.md`. Held ignition is below the paper's grain under the §2 scope rules. Sections are unchanged since #973's verified rebuild, so I did not rebuild. +- **Flagged, not acted on (site lane):** (a) `75e6a15`'s `density_law_knee_placement_supp_output.txt` is not the output of the committed `_supp.py`. Either commit the code that printed the last two lines or mark them hand-run. (b) `explorer/scripts/rar_scatter_nogo_real_sparc.py` hardcodes the `/mnt/c` data path, so the explorer's SPARC scripts do not run as committed on any machine but the one that wrote them. The 11-03 chemistry re-run had no such path. +- **So what.** Nothing here changes a verdict. The correction is to an *interpretation* the paper printed of its own fit. "Switched off" made the density law's loss sound like a degenerate fit, which would be a weaker statement than the one the data support. The actual statement is that density carries a real, weak signal (6σ over no density) and still loses by +2,843, and at the derivation's own calibration it collapses to Newton. The pattern is worth naming: a site finding that lands between two back-annotation runs is invisible to a pass that windows on this repo's commits alone. This pass found it only because it windowed on `synchronism-site` too. + +### 2026-11-03 (interactive maintenance pass): **The 11-02 pass printed "the chemistry-row Z-null remains unmeasured" at 7 sites. Ninety minutes later the site explorer ran it on the 10 recoverable rows, and it fails. Each of the 7 notes now carries an UPDATE. Appendix B gains three CHECKED notes, all decidable from the paper's own formulas: γ_optical is ionization energy, γ_phonon is θ_D, and the B.5 "Electronegativity 0.979" row is chemical softness.** + +- **Step 0:** shared tree on `main`, 2 behind `origin/main` (`58428709`). No local changes. #14 (`publisher/wp-2026-10-29`) is OPEN, with build and summary SUCCESS and no review decision. Its head is now `8468225e`: seven review commits (`ba007451..8468225e`, 11-02 evening) landed on top of the 11-02 entry. They scope the finite-scan and chemistry-null conclusions to the tested samples, add a top-level CHANGELOG review correction (Z = 3–9 alone gives linear r = 0.999, so a Z = 1–54 electronegativity sample cannot settle the row null), and make whitepaper CI fail on a missing or failed PDF (`7da44e01`, which closes the `|| echo` gap in build-gotchas #8). This entry is stacked on `8468225e` from a fresh worktree, `ai-workspace/.worktrees/syn-pub-1103`. The hestia gate was down at session start (`gate.degraded`, then a timeout) and came back on the next retry. +- **Window `e2784def..58428709`:** `d91efc49` (insight: executable world models as instruments, `forum/` only; out of scope) and **`5709b579`** (11-02 06:10, site back-annotation). I grepped each datum on its own over every live section file at #14's head, meta excluded, with the matched span printed: + - (i) **The Z-null on the real rows fails, and the high r values are identities.** This one lands. The paper's 7 notes from 11-02 (exec L167/L187, conclusion L73/L81/L121, `chemistry.md` L121, appendix B L121) all say the real-row null is unrun or unmeasured. `γ_opt` 11 hits: Appendix B.2 prints γ_optical = IE_ref/IE and three "validated correlations" against 1/γ_optical in the next lines. `13.6` and `Mulliken`: **0**. B.5 prints "Electronegativity | S vs γ_optical | r = 0.979 | #118". + - (ii) The site's 23-row table and #81 conductivity at r = −0.414: `WRONG SIGN`, `−0.414`, `23-row`, `23 propert`: **0** hits. The paper has no electrical-conductivity row (B.3 lists σ as governed by γ_electron, with no r). Nothing printed, nothing to correct. + - (iii) The DE local-fluid bound is one-sided (c_s² < 0 for γ < ½): `local-fluid`, `gradient instab`, `2γ−1` (1 hit, the §5.15 C-expansion coefficient, a different object), `c_s` (1 hit, the open-questions pressure ansatz, a different object): **not printed**. The 09-15 |2γ−1| ≲ 10⁻⁵ exclusion never reached the paper, so there is nothing to scope. + - (iv) The count rule against the Bell row: `count rule`, `did not fit` **0**. §5.15 L128 lists the six with Bell/CHSH as (6), and L134 already says it is run on the framework's own harness. **Gates on dp** (with 10-07 and 10-29), no edit. +- **Re-run, not relayed.** Ran `synchronism-site/explorer/tools/chem_z_null_real_rows.py` (`python3 -I`, rc 0). The output is **byte-identical** to the explorer's recorded `chem_z_null_real_rows_corrected_output.txt`. Across 10 rows: max |r(P, Z)| = 0.460 linear, 0.525 with ln Z; max quadratic R = 0.514; framework r 0.595–0.982; r(P, 1/γ_ph) = r(P, θ_D) on every phonon row; jackknife minimum 0.604 (E) / 0.627 (G) / 0.374 (B) without diamond. #115 χ against 1/γ_opt gives 0.956 (the paper prints 0.938, the archive's recorded value), with r(χ, Z) = −0.032. #118 softness gives 0.979. +- **Edits (words added, none removed; `git diff --word-diff=porcelain` removed tokens: 0):** (1) an `[UPDATE 2026-11-03]` after each of the 7 notes from 11-02: the real-row null has been run and fails, with the identities named. (2) Appendix B.2, a `[CHECKED 2026-11-03]` after the electronic-channel list: by the printed formula, every 1/γ_optical correlation is a correlation with IE, and χ against IE is built into Mulliken's definition. (3) B.2, a `[CHECKED]` after the phononic list: r(P, 1/γ_phonon) ≡ r(P, θ_D), and the elastic rows rest on diamond. (4) B.5, a `[CHECKED]` after the table: the "Electronegativity" 0.979 row is softness against 1/IE. Changelogs: 00 and 07 (LF); 05 and 09 (CRLF, edited byte-wise, line endings preserved); top-level (+1). +- **Build (verification only; CI builds and deploys on merge):** md, pdf and web-clean rc 0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **298** pp. In the PDF text layer (whitespace-stripped): `UPDATE 2026-11-03` **11** (7 notes + 4 changelog mentions), the same as the monolith's 11; `CHECKED 2026-11-03` **4** (3 notes + 1 changelog mention). HTML: the new text is present in section_0/5/7/9 (2/1/3/1 hits on "real-row Z-null"). Literal `**` after tag strip is **0** in section_0/5/7/9. The pandoc→plain probe of every added line finds **0** literal asterisks. Geometry (`pdftotext -bbox`, xMax > width − 36): **1** word, page 6 = the baseline. All artifacts were restored by explicit path and not committed. The local pandoc also rewrites `` classes in section_4/6, which is environment churn, not content. +- **Instrument notes:** (a) a relative `../../CHANGELOG.md` in a Bash `file` call made the gate resolve `..` from the session root. That opened two unneeded scope requests (`scope-ae4165e166b5` for `/home/CHANGELOG.md`, `scope-8934275e4e6f` for `/home`). The hestia MCP tools were not loaded this session, so I could not withdraw them; they should be denied. (b) A per-line `grep -c` over the HTML undercounts, because the HTML wraps after `UPDATE`. Count a phrase unique to the note instead. +- **Cross-repo: Web4. No change.** Window `41971895..e2fa0c02` contains only the merge of #973 (the 11-02 log). **0** `web4-standard/core-spec`, `ontology` or `whitepaper` changes. On the committed artifacts: `build/` = `docs/` for the PDF and monolith (`cmp`), 24 pp, U+FFFD 0, geometry 8 words past 545 pt on pages 6/11/13/15/22 (worst 596). That equals the 11-02 baseline. No web4 log entry was written: a log-only PR for an empty window adds review load and nothing for a reader. It is recorded here instead. +- **Frame, for dp:** the 11-02 pass and its reviewer argued about what an electronegativity sample over Z = 1–54 can say about rows nobody had seen. The data answered it the same morning. The answer moved the ground under the verdict rather than the verdict: the chemistry track is not a Z trend in disguise; it is θ_D and IE under new names. That is a stronger and simpler reason for Reparametrization than the null ever was. It also means the paper's B.2 "two orthogonal channels" claim (r ≈ 0) reduces to "θ_D and IE are weakly correlated across the elements", which was not checked today. Still gated on dp, unchanged: the 10-18 outage-backlog scale question, whether a convention-dependent kill counts in the headline, and now the count rule against the Bell row. + +### 2026-11-02 (interactive maintenance pass): **The paper prints S651's chemistry null as "sound velocity, electronegativity, atomic volume are near-monotonic functions of Z; any smooth monotonic function will achieve r ≈ 0.95+ … by construction". The premise is false across the periodic table. Re-run: the best monotone function of Z fits Pauling electronegativity at r = 0.33. Annotated CHECKED at 7 sites. The Reparametrization verdict is unchanged; it never rested on this null.** + +- **Step 0:** shared tree clean, `HEAD` = `main` = `origin/main` (`e2784def`). #14 (`publisher/wp-2026-10-29`, head `62f7610a`, carrying the 10-29/10-30/10-31 passes) is OPEN, MERGEABLE, CLEAN, build and summary SUCCESS, no review yet. This entry is stacked on #14's branch from a fresh worktree `ai-workspace/.worktrees/syn-pub-1102`. Read the 10-31 entry and `watch_items` before acting. +- **Window `c7185d12..e2784def`: one research commit**, `e2784def` (10-31, back-annotate from site). Each datum grepped separately at #14's head over every live section file (meta excluded), matched span printed (`/tmp` script, patterns listed here): + - (i) **GC paired differences cancel the data; the EFE-free corner is ~3σ disfavoured, not closed.** `globular`, `GC fork`, `EFE-free`, `external-field-free`, `10.6`, `10.8`, `tidal radius`, `Webb`, `held vacuously`, `no galaxy-scale prediction`: **0** hits. `paired` 4 hits, all "paired with S647" (the chemistry audit pairing, unrelated). `6/6` 11 hits, all A2ACW / demotion-set figures, unrelated. The 10-31 pass declined to print the 10-30 "closed by 10.6σ" claim as a relay; that restraint is now vindicated: it was a data-free statistic. Nothing printed, nothing to correct. + - (ii) **The chemistry polynomial-in-Z null was run on synthetic curves, and its premise (near-monotonic in Z) is false.** This one lands. The paper prints the premise as fact at **7 sites**: exec L167 (chemistry caveat), exec L187 (S651 bullet), conclusion L73 (S651 bullet), L81 (validation-pillars sentence), L121 (chemistry caveat), `12-chemistry/chemistry.md` L121 (caveat block), `appendix_b_chemistry.md` L121. Two further mentions are mode *names* in lists (exec L192 "null-model-gap-against-best-monotonic-null … polynomial-in-Z null"; the 17th-mode label in the L187 bullet); left as names. + - (iii) data-dependence-column proposal: framed for the project lead, `data-dependence` **0** hits. Gates on dp. +- **Re-run, not relayed.** The block says the premise is false; I tested it. `manuscripts/publisher/scripts/z_null_electronegativity_20261102.py` (committed with this entry): Pauling electronegativity, Z = 1–54, the 51 elements with a Pauling value, typed from standard (CRC) tables, **not** loaded from a database (no `mendeleev`/`pymatgen` on cbp). Spearman(EN, Z) = **+0.093**. Polynomial-in-Z fits, degree 1/2/3/5: r = **0.029 / 0.232 / 0.251 / 0.388**. Best monotone function of Z (isotonic regression, the upper bound on *any* monotone null): r = **0.246** increasing, **0.333** decreasing. "r ≈ 0.95+ by construction" fails by a factor ~3 in r. Atomic volume (Lothar Meyer) was not re-run numerically; its periodicity is textbook and the note states it qualitatively. Caveat on scope: the chemistry rows are unpublished, so whether the *rows* sample across periods (where the premise fails) or within one (where it can hold) is unknown; the note says the real-row null is unrun rather than claiming its value. +- **Why the verdict is untouched:** the paper's Reparametrization verdict already rests on S669 (exec L188, conclusion L75): γ_phonon ≡ 2T/θ_D is a definitional relabel of the Debye temperature, Δr = 0 to machine precision. The site block says the same. So this is the sibling-sentence pattern again: the verdict was re-grounded in May, and the superseded ground stayed printed beside it as fact. +- **Edits (words added, none removed; `git diff --word-diff=porcelain` removed tokens: 0):** full `[CHECKED 2026-11-02]` note in both S651 bullets (exec L187, conclusion L73); short note at the other five sites. Changelogs: 00, 07 (LF), 05, 09 (CRLF, byte-wise, line endings preserved), and top-level (+1). +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc=0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **294 → 295** pp. Notes in PDF text layer 11 (7 notes + 4 changelog mentions; 10 by raw `grep -c` because one wraps at "2026-11-" — joined count matches the monolith's 11). HTML: 2/1/3/1 notes in section_0/5/7/9, literal `**` **0** in each; section_11 (changelog appendix) 45, equal to HEAD's committed 45. Geometry (`pdftotext -bbox`, xMax > width − 36): **1** word, page 6's plain-text table cell, = the 10-31 baseline. Artifacts restored by explicit path, not committed. +- **Cross-repo: Web4. No content edit.** Window `e43c8081..129aee6c`: #953 merged (`f5af9a03`), hub commitments K5 (#959–#965), one insight doc (#972). **0** `web4-standard/core-spec/` and **0** `whitepaper/sections/` changes. First geometry measurement of the web4 PDF (memory had it "unmeasured"): text block edge is 540 pt (1 in margins), so the overrun threshold is ~545, not width − 36. **8 words on 5 pages** run into the margin by 9–56 pt (page 6 code-block comments, pages 11/13/15 spec paths, page 22 reference URLs); **0 clipped** (max xMax 596 of 612; pages 6/15/22 rendered and read). Cosmetic; not fixed, because the web4 PDF is a committed, parity-checked artifact and the fix would cost a six-file rebuild for no lost text. Baseline recorded in that repo's log. +- **Frame, for dp:** the site's (iii) names the real pattern: *both* 10-31 corrections are refutation-side errors in the self-critical direction, and the paper absorbed one of them uncontested for 5½ months because a null that agrees with the self-critical reading "feels finished". The Publisher's CHECKED discipline has so far been applied almost entirely to *claims*; this is the first CHECKED note on a printed *refutation premise*. A sweep of the paper's other printed nulls for "run on real rows / synthetic / not run" is the data-dependence column applied to the paper; it gates on dp's ruling on the column, so not started. Still blocking, unchanged: the 10-18 outage-backlog scale question; whether a convention-dependent kill counts in the headline. + +### 2026-10-31 (interactive maintenance pass): **One dated parenthetical. The 10-29 note printed the TEST-09 sweep's middle value as "0.30, on the strict threshold"; it is 0.297, 0.003 under. Re-run, verdict unchanged. Also a PDF formatting fix: 98 words on 55 pages ran past the margin, most clipped at the page edge. Long inline code now breaks; 1 left, a plain-text table cell.** + +- **Step 0:** shared tree clean, `HEAD` = `main` = `origin/main` (`c7185d12`). #14 (`publisher/wp-2026-10-29`, head `39f8fe80`, carrying the 10-29 and 10-30 passes) is OPEN with build and summary SUCCESS. This entry is stacked on #14's branch at the log's append point, from a fresh worktree `ai-workspace/.worktrees/syn-pub-1031`. I read `watch_items` before proposing anything (the 10-30 instrument note). +- **Window `80df7d26..c7185d12`.** The 10-30 pass read through `80df7d26`; two commits landed after it. Each datum was grepped separately, at #14's head, over every live section file, with the matched span printed: + - `2c985ae4` (i) **GC statistic closes the EFE-free corner** (γ = 0.87–0.95 compander worse than MOND+EFE by 10.6σ, pre-registered; γ = 2 vs MOND+EFE 1.7σ, "ranked beside"): `globular`, `outer slope`, `EFE-free`, `external-field-free`, `10.6`, `0.87`, `GC fork`, `differs from MOND`: **0** hits. The paper never printed the corner or the 09-08 "survives", so nothing printed is corrected. One adjacent sentence to watch: exec L187 calls EFE = 0 "`not-evaluable` against Chae et al. 2020". The GC result is a different dataset and does not contradict that, so no edit. The GC closure is a site-executed result I did not re-run, and printing it would be relaying, not correcting. + - `2c985ae4` (ii) **Number hygiene.** `0.406`/`0.297`/`0.260`/`5.31`: 0 hits, but L199 (added by #14 on 10-29) prints the sweep at two decimals: "`(Ω_m−Ω_b)/Ω_b` gives 0.30, on the strict threshold, does not fire". **Re-ran** `synchronism-site/maintainer/scripts/which_C_carries_the_floor_4dp.py` (exit 0, identity control n = 3.35 ± 0.07 passes): |Δn| = **0.4062 / 0.2970 / 0.2604**. Interpolating the free scan between B_max 5.00 (n 3.44) and 5.389 (n 3.4557) puts the 0.3 edge at ≈ 5.31–5.32, which agrees with the block. A dated parenthetical was added after the sweep sentence. Words were added and none removed; the verdicts, rule outcome and count of 6 are unchanged. σ-distances 3.4/2.6 → 3.3/2.5/2.2: `3.4σ`, `2.6σ` **0** hits. α/π gap 16–28 → 15–27 (`α/π`, `16–28`, `c_μν`, `naturalness`): **0** hits. + - `2c985ae4` (iii) **TEST-10 propagation** (`NGC 3741`, `NGC 2915`, `UGC 06667`, `ESO 444`, `2/153`, `gas census`): **0** hits. The paper's TEST-10 figure is the 69% headline, which the block does not move. + - `2c985ae4` (iv) "one equation = the Möbius map plus exclusions" is framed for the project lead. `Möbius` **0**; `x/(x+2)` 2 hits (L80, L109, the identity itself), consistent. Gates on dp, no edit. + - `c7185d12` (06-26 triage α/π rounding, corrected at source): not printed. No edit. +- **Formatting fix (build, not content).** On the rendered page 141, two inline-code paths ran off the right edge. Measured over the whole PDF with `pdftotext -bbox` (any word whose xMax is past page width − 36 pt): **98 words on 55 of 293 pages**, almost all inline-code provenance paths clipped at the physical edge (e.g. `…/efe-zero-is-not-refutable-by-chae20` with the rest invisible). The cause is that pandoc sets inline code as `\texttt{}`, which TeX never breaks. The fix is the new `whitepaper/pdf-breakable-code.lua`, passed by `make-pdf.sh`, which puts `\allowbreak` after `/ _ - .` in inline code of ≥ 20 characters. After the fix: **1** word on 1 page, page 6's plain-text `(density/velocity/pressure/viscosity)` table cell, which is not code; left. 294 pp, `Missing character` 0, U+FFFD 0. **Text-layer check:** I compared the before and after `pdftotext` output with whitespace, digits and hyphens stripped, resyncing at each difference. Every difference is a previously clipped path tail that is now present (`…-by-chae20…-dex.md`, `…-auditor-is-contested_.md`, `…only_through_retrieval_.md`, …). Page 127, the densest page of paths, was rendered and inspected. HTML is untouched (the filter is PDF-only). Artifacts were not committed, since CI deploys on merge. +- **Cross-repo: Web4. No new edit.** Window `e43c8081..ed5ba285` is hub commitments K5 (#959–#965), with 0 `web4-standard/` and 0 `whitepaper/sections/` changes. I ran the independent artifact check that Nova's source-only review of web4#953 left open: rebuilt from source in a fresh worktree, HTML byte-identical, PDF text layers differ only in the Generated line, PDF pages 8 and 20 rendered, `build/` = `docs/`. Result PASS, posted on #953. That repo's log entry rides with #953's next commit or the next pass; not written here. +- **Frame, for dp:** the overflow had been in the published PDF since the inline-provenance style started (summer), and no pass caught it. The build-verification checklist checks glyphs (`Missing character`) but not geometry. The bbox measure is cheap, and it is now step 5 of §5 Build Verification in `PUBLISHER_CONTEXT.md`. Still blocking on dp, unchanged: the 10-18 outage-backlog scale question, and whether a convention-dependent kill counts in the headline. + +### 2026-10-30 (interactive maintenance pass): **Two outage-window items annotated CHECKED. Both can be decided from text the paper already prints. Also, the 10-29 entry proposed an outage back-scan that the 10-18 pass had already run and filed as the first `watch_items` entry. The publisher is losing its own watch list.** + +- **Step 0:** shared tree clean, `HEAD` = `main` = `origin/main` (`80df7d26`). #14 (`publisher/wp-2026-10-29`, `08b76f34`) is OPEN and CLEAN. This entry is stacked on #14's branch at the log's append point, from a fresh worktree `ai-workspace/.worktrees/syn-pub-1030`. +- **#14 verified independently:** I rebuilt its head in a `/tmp` worktree with `CI=true`. md/pdf/web-clean rc=0; `Missing character` **0**; U+FFFD **0**; PDF **292** pp. The monolith diff is exactly the L199 line plus its changelog entry. `section_5.html` has **0** literal `**`. This matches #14's own report. +- **Window `c4d6f843..80df7d26`: #14 opened (11:40Z 10-29) before all three research commits landed, so no pass had read them.** I grepped the paper for each datum separately, with the matched span printed: + - `c13237c6` (i) **Count floor composition**: `6.39`, `28/153`, `NGC 3741`, `ESO 444`, `floor of 3`: **0** hits. The paper's six-member list (`dark_matter.md` L128) is consistent with it. Printing "3 under every rule, 1 a data refutation" beside the count **gates on dp**, as the block says. + - `c13237c6` (ii) **a₀(z) is a commitment and prior art**: `cosmologically coupled`, `cH(z)`, `constant a₀`: **0** hits. The paper's a₀(z) note (L243–255) already calls the form not novel (Mayer+2023 eq. 13) and non-discriminating, lists Ciocan's structural reading and systematics, and does not print a 5σ constant-a₀ exclusion. No edit. + - `e7f6fc39` / `80df7d26` (**B7 stacking arm is the DSR soccer-ball problem**): `Umklapp`, `mod G`, `soccer`, `per-quantum`: **0** hits. The only two `B7` hits are an fσ₈ citation at exec L189 / conclusion L77, which is not this bet. Bucket/count **gates on dp**. No edit. +- **Outage-window re-scan.** The 10-29 entry's frame point said "a one-time sweep of every `📌` block dated 09-03…10-07 against the paper would likely find more. I have not run it." I gave that sweep to a read-only subagent. Afterwards I found the **10-18 entry had already run it**: ten STALE-IN-PAPER items, deferred until each was verified at source, and a scale question raised to dp (status banner vs rewrite vs per-item). It is the first entry in `state/whitepaper_sync.json` `watch_items`. So the re-scan became a test of whether a second read reproduces the first. 17 blocks were in the window (4 skipped as already handled). It re-found 10-18's items 3, 4, 6, 7 and 8 with matching paper quotes, plus the 09-14 Cardassian attribution (part of item 5). It added one item 10-18 did not list: **09-23, the sign lock excludes thawing, not only crossing.** It read items 2 (r = 0.994) and 10 (2π "DERIVED") as scope gaps, not contradictions. I verified every quote it used against both files before acting. +- **Edits: two `[CHECKED 2026-10-30]` notes in `dark_matter.md` L346, words added and none removed (word-diff: 0 removed tokens).** I chose these two because neither relays a site number: + 1. **"factor 1.96 a₀ tension … dissolves at ϒ = 0.6"** (10-18 item 7; source 10-04 (i)). L346 compares the fitted a₀ = 5.33×10⁻¹¹ with cH₀/2π. L109 in the same section shows the fit depends on γ and a₀ only through γ/a₀ = 9.17×10⁹. Its Milgrom-equivalent scale is therefore a₀/γ = 1.09×10⁻¹⁰, which is 4.6% from cH₀/2π. The "tension" was the 1/γ conversion at γ ≈ ½. The note uses only numbers the section prints, and cites the block's 1.074–1.092×10⁻¹⁰ Υ sweep as site-run, not re-run. The bold conclusion it follows ("shape parameter is unidentified at factor 2") still stands. This is the same sibling-sentence propagation failure the 10-27 entry described. + 2. **"no phantom crossing, which is the quadrant DESI DR2 prefers"** (09-23). In CPL the early value is w₀ + wₐ, so the w₀ > −1, wₐ < 0 quadrant contains a crossing only where w₀ + wₐ < −1. The rest is thawing. The lock excludes the whole quadrant, so it forbids thawing as well. The verdict is unchanged. +- **Not done, on purpose.** Items 6 (the 10⁻⁵ P(k) pin) and 8 (B_req = ν/f lensing) rest on site-executed numbers ("not re-run by me" in the block), so each needs a re-run before inscription. Items 1, 3 and 4 and parts of 5 and 9 gate on dp. The 10-18 scale question is still unanswered after 12 days. Adding more per-item notes to the L346 paragraph (now ~16 KB on one line) makes it harder to read, so the scale question is the actual blocker. Progress is recorded in the watch item. +- **Changelogs:** section 05 (CRLF, byte-wise, 4 lines) and the top-level changelog (+1). +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc=0 with `CI=true` and absolute script paths. `Missing character` **0**, U+FFFD **0**, PDF **292 → 293** pp. Both notes are in the PDF text layer and in `section_5.html`. Literal `**` and `*[` in `section_5.html`: **0**. A pandoc → plain probe of L346 gives **0** asterisks. Glyphs are limited to ones the paper already prints (no ≫/□/ζ). Artifacts were restored by explicit path, not committed. +- **Instrument note (the find of this pass):** the 10-29 pass read neither `watch_items` nor its own log beyond the previous entry, and it re-proposed a scan that was already on file. I nearly repeated that: the subagent was briefed before I grepped the log for "back-scan". **Before proposing work, grep `watch_items` and the log for it.** The re-scan was not wasted, because two independent reads agreeing on five items is evidence that the 10-18 list is sound. But it was luck, not design. +- **Cross-repo: Web4. No new edit.** Window `23c3b3ad..e43c8081`: #950 (the `web4-lct.md` protocol mirror of LCT §3.4), hub commitments K2–K4b, and the #952 docs sweep. **0** `whitepaper/sections/` changes. web4#953 (the 10-29 §6 fix) checks out against spec §3.4: consistency vs possession, rotation → new id, no allocator. It is MERGEABLE with all 6 checks green, its PDF is 24 pp with U+FFFD 0 and `build/` = `docs/`, and it is BLOCKED only on review. Entry in that repo's log. + +### 2026-10-29 (interactive maintenance pass): **The paper calls the `Ω_m` vs `Ω_m/Ω_b` floor choice a "convention dispute … measured, and it is small". The 10-27 back-annotation shows the two floors presuppose different frameworks: the one under which TEST-09 does not fire is a no-CDM variant whose dark-energy sector cannot run. Annotated CHECKED at the one site, with the 09-18 sweep printed for the first time.** + +- **Step 0:** shared tree clean, `HEAD` = `main` = `origin/main` (`30fa3752`). #13 was merged and CI deployed it, so the 10-27 BCS annotations are the published paper. No open PRs. Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1029` on `publisher/wp-2026-10-29`, upstream `origin/main`. +- **Window: the 10-27 entry's range ended at `a00828b9`.** Two research commits landed after that pass branched and before its merge, and no pass read them: `2065a308` (back-annotation 10-27, four items) and `d1ccc3b9` (Phase-16 door #3 correction at source). Nothing after `30fa3752`. I grepped each datum separately over every live `.md`, and used Python for spans because ugrep rejects bounded repeats. + - (i) **The citable register** (LLR Ġ/G first; the DESI growth-suppression negative retitled not citable): the paper's Session 107 / S656 "transferable mechanism-class constraint" sentences already carry `[RETRACTED by S668]` at all four `G_local` sites (exec ×2, conclusion ×2). LLR Ġ/G as a register item is not printed. The ranking criterion gates on dp. No edit. + - (ii) **Ω_m plays two roles in the ceiling fork. ACTED ON.** The paper's 5.15 bullet (L199) prints the `1/Ω_m` vs `Ω_m/Ω_b` floors as a "convention dispute … measured, and it is small" (+7.3 vs +12 in the RG-floor profile). The paper never printed the 09-22 "DE sector needs CDM" result (`without CDM`, `ρ_m = ρ_b`, `dark/baryon`, `5.36`: **0** hits) or the 09-18 TEST-09 ceiling sweep (`3.49`, `B_max ≲ 5.4`: **0** hits). Added one `[CHECKED 2026-10-29]` note at the end of that bullet: "small" is fit quality, the floors presuppose CDM-present vs CDM-replaced, the no-CDM variant's DE sector fails, and the sweep's verdicts are 0.41 fires, 0.30 and 0.26 do not. That is the 10-27 conclusion: the non-firing branch is a variant whose cosmology fails, and the firing branch is the framework as written. **Re-executed, not relayed:** `synchronism-site/maintainer/scripts/which_C_carries_the_floor.py` and `de_sector_without_cdm.py` were run with `python3 -I`, output to `/tmp/pub1029/`. Both exit 0, and `diff -b` against the committed `_output.txt` files is empty. Count HELD at 6. The counting rule gates on dp. + - (iii) **Knee-exponent provenance** (the "V⁻² excluded at ~10σ" misreading): the paper's one `−0.16 ± 0.19` (5.15 L122) reads "neither law" and points at the PREDICTIONS row. The paper's two `10σ` hits are the a₀(z) anchor and a DESI channel, neither of them this. No edit. + - (iv) **Door #3 closed by Borexino ¹¹C** (and `d1ccc3b9`'s at-source correction): `15–30`, `heat bound`, `D/D`, `Lindblad`, `Borexino`, `door #3`: **0** hits. The paper never carried door #3. No edit. + - Site-only record items (footer qualifier, /for-researchers lint): not in the paper. +- **Edit (words added, none removed):** one line in `15-dark-matter/dark_matter.md`. The old line is a byte prefix of the new one, and the line count is unchanged. Section CHANGELOG 05 is CRLF and was edited byte-wise: a 4-line insert using the file's own CRLF. Top-level CHANGELOG +1. A pandoc → plain probe on the modified line finds **0** literal `**` and **0** `*[`. No ≫/□/ζ. +- **Build (verification only; CI builds and deploys on merge):** md/pdf/web-clean rc=0 with `CI=true` and absolute script paths; `Missing character` **0**; U+FFFD **0**; PDF **292** pp, unchanged against #13's build (the note fit on existing pages). The note is in the PDF text layer. The monolith diff is exactly the one line. The local web build also re-wrapped `section_4/6/9/11.html` (local pandoc line-wrapping; same text, different line breaks), which is the CI-is-the-builder reason in miniature. The artifacts were restored by explicit path and are not committed. +- **Frame point (surfaced, not acted on):** this is the third pass in a row where the find is a ledger result that never reached the paper. 09-22 and 09-18 were each a month old, and both landed in the 09-03…10-07 publisher outage. The 10-26 proposal's "propagation failure" steady state covers more than sibling sentences: it also covers results from the outage window that no pass has swept. **A one-time sweep of every `📌` block dated 09-03…10-07 against the paper would likely find more.** That is cheaper than waiting for a site persona to re-derive each one. I have not run it. It is the best candidate for the next pass with spare budget. +- **Cross-repo: Web4. ONE corrective integration** (web4 `8d340458` on `publisher/wp-2026-10-29`). #945 made the LCT identifier `lct:web4:mb32:b` + base32(sha256(raw public key)). §6 still printed `sha256(binding_proof)`, and the level-5 re-issuance reason (in §6 and the glossary) depended on it. Both are corrected, and the artifacts are committed with the source per web4's #789 contract (24 pp, U+FFFD 0). The 10-27 entry's watch item ("if Slices C/D land a normative LCT-spec change, web4 §6 owes an update") fired, by way of #819's split instead of the hub slices. + ### 2026-10-27 (interactive maintenance pass): **The paper prints "γ ≈ 2 in BCS superconductivity" as a cross-domain validation while its own master equation makes γ ≈ 2 a single uncorrelated particle. The 10-26 back-annotation withdrew the quantum/classical endpoint labels that hid this. Annotated CHECKED at four sites, scoped "classical limit" at the two definitional sites.** - **Step 0:** shared tree clean, `HEAD` = `main` = `origin/main` (`a00828b9`). #12 merged and CI deployed it (`a00828b9`), so the 10-24/10-26 annotations are now the published paper (290 pp, U+FFFD 0, `build/` = `docs/` for PDF, monolith and `section_0/5/7.html`). No open PRs. Worked in a fresh worktree `ai-workspace/.worktrees/syn-pub-1027` on `publisher/wp-2026-10-27`, branched from `origin/main`. diff --git a/manuscripts/publisher/reports/outage_window_backfill_audit_2026-11-13.md b/manuscripts/publisher/reports/outage_window_backfill_audit_2026-11-13.md new file mode 100644 index 000000000..3f6d014f8 --- /dev/null +++ b/manuscripts/publisher/reports/outage_window_backfill_audit_2026-11-13.md @@ -0,0 +1,60 @@ +# Outage-window backfill audit — disposition at #14's head (2026-11-13) + +**Asked for by:** NOVA's exact-head review of dp-web4/Synchronism#14 (2026-11-13 04:14Z): "a one-time +publisher-window backfill audit". +**What it audits:** the ten STALE-IN-PAPER items the 2026-10-18 pass found in the 09-03 → 10-07 publisher +outage (log entry 2026-10-18; `state/whitepaper_sync.json` `watch_items[0]`), plus the 09-23 item the +2026-10-30 re-scan added. The 10-30 re-scan independently reproduced 10-18's items 3, 4, 6, 7 and 8. +**Read against:** `publisher/wp-2026-10-29` at `ca78abce` (the 11-12 head), live `whitepaper/sections/**` +`.md` files, excluding `meta/` and changelogs. Line numbers are from that head. They are unchanged at +`53e0214f`, which only edits table cells in place. +**Method:** each item's printed phrase was located with a span-printing scan +(`/tmp/pub1113/backfill_scan.py`, not committed: its "annotation follows" flag is paragraph-level and +over-reports on 16 KB one-line paragraphs). Every disposition below was then read by hand at the cited line. +**Not done here:** the source `📌` blocks were not re-read today. Item descriptions are the 10-18 entry's, +and that entry's quotes were spot-checked on 10-18 and 10-30. + +## Disposition + +| # | Item (source block) | Printed at (`ca78abce`) | Status | Closed by | What remains / gate | +|---|---|---|---|---|---| +| 1 | TEST-09 bounded-boost as a "surviving decisive negative" (09-18 sweep; 09-25 (i)) | body: §5.15 count item (2) L138 and the floor-convention bullet; **headline: exec summary L189, §5.15 status note L5, conclusion L77** | **PARTIAL** | 10-29 + 10-31 (sweep printed, unrounded 0.406 / 0.297 / 0.260); 11-06 (registered-rule note on count item (2)) | The headline sentence "both surviving decisive negatives are galactic (locality no-go; TEST-09 BTFR bounded-boost)" is unscoped at all 3 sites. Whether a convention-dependent kill counts **gates on dp**. | +| 2 | "Compatibility-Synthon scaling (r = 0.994)" as calibrating work (09-25 (iii): p_crit ∝ 1/⟨C⟩ refuted on its own data) | exec summary L17 | **OPEN** | none | Scope gap (10-30 reading). Needs a re-read at source before inscription. | +| 3 | "ρ_crit ∝ V⁺² sign inversion kills independently" (10-07 (iii): conditional on forward regression; orthogonal gives V^+2.1) | exec summary L191, conclusion L79 | **OPEN** | none | Partly **gates on dp** (10-18). | +| 4 | "EFE = 0 exactly" / "{EFE = 0} ⟺ locality" (09-16 (ii), 09-17 (iii): algebraic reading only; under L2, ∇C ≠ 0 refracts the field) | §5.15 L111, L113, L117; exec summary L187/L191; conclusion L73/L79 | **OPEN** | none (L99's "EFE = 0 is a property of the density-keyed function" is a different, earlier point) | This is the "locality no-go" in item 1's headline sentence. Partly **gates on dp**. | +| 5 | "Dark matter AND dark energy are both coherence effects" / "no Λ required", no Cardassian prior art (09-22, 09-14) | §5.15 L364 | **PARTIAL** | 11-06 (L342 names the fluid, Cardassian-class reading as the only survivor of the LLR test) | L364's "both coherence effects" sentence is unscoped. | +| 6 | "no order-ε⁰ channel / retires the perturbation caveat" (09-15: only without the pressure term) | §5.15 L366 ("no order-ε⁰ term exists") | **OPEN** | none | "pressure term" has 0 hits. Rests on site-executed numbers, so it needs a re-run before inscription (10-30). | +| 7 | "factor-1.96 a₀ tension dissolves at ϒ = 0.6" (10-04 (i)) | §5.15 L366 | **CLOSED** | 10-30 (CHECKED from L109) | none | +| 8 | lensing-ceiling deficits 6× / 13× / 28× via √f (09-25 (i): B_req = ν/f) | §5.15 L242 | **CLOSED** | 11-11 (Brouwer read in full; 1/f, not √f); 11-12 correction | none | +| 9 | "no parameter-free obstruction on the coarse-graining axis" (09-15 + 09-24 (i): LLR bound) | open questions L51 | **OPEN** | none | 0 annotations at the sentence. | +| 10 | "a₀ = cH₀/(2π) DERIVED" and "the 2π is not numerology" (09-19) | §5.15 table L16; §5.15 L256 | **PARTIAL** | 11-11 (table note) + 11-13 (in-cell qualifier) | L256 "This is not numerology; it connects the MOND scale to cosmology" is unscoped. | +| — | 09-23: the sign lock excludes thawing, not only crossing | §5.15 L366 (the 10-30 log cited L346 at that head) | **CLOSED** | 10-30 (CPL algebra) | none | + +**Tally:** 3 closed (7, 8, 09-23), 3 partial (1, 5, 10), 5 open (2, 3, 4, 6, 9). + +## What the table shows that the list did not + +1. **The headline's two surviving negatives are exactly the two open items that reach the headline.** + Exec summary L189, §5.15 L5 and conclusion L77 each say "both surviving decisive negatives are galactic + (locality no-go; TEST-09 BTFR bounded-boost)". The locality no-go is item 4: its EFE = 0 ⟺ locality + step holds under the algebraic reading only. TEST-09 is item 1: it fires under the 1/Ω_m floor and not + under the two Ω_b floors. The body of §5.15 now prints the TEST-09 condition. The headline sentence that + relies on both prints neither. This is the sentence a first-time reader meets, and it is the one with the + most outage debt. +2. **Per-item annotation has closed 3 items, 2 more in part, in 26 days.** Each closure came from a pass whose own window + happened to touch the item. No pass worked the list as a list. The 10-18 scale question (a dated status + banner at §0 and §5.15, a rewrite of the galaxy-sector verdict paragraphs, or per-item notes) has not + been answered, so per-item has remained the default by inaction. +3. **Five open items need a decision before any edit, not just a pass.** Items 1, 3 and 4 gate on dp in + part. Items 2 and 6 need source work (a re-read, a re-run) that no window will bring in. + +## Recommendation (for dp; not acted on) + +- Merge #14 on its current qualifications, if the in-cell qualifiers answer the status-cell half of the hold. + Then take the **banner** option for the five open items: one dated paragraph at the head of §0 and §5.15 + that names items 1–4, 6 and 9 and points to `PREDICTIONS.md`. That costs one edit. The per-item route costs about + ten notes on paragraphs that are already 10–16 KB. +- Whatever the choice, item 1's headline sentence should not stay as it reads, because it is the most-read line with + the most debt. + +— cbp-claude (autonomous-publisher-cbp, 2026-11-13) diff --git a/manuscripts/publisher/scripts/eta_vs_abrikosov_gorkov_20261113.py b/manuscripts/publisher/scripts/eta_vs_abrikosov_gorkov_20261113.py new file mode 100644 index 000000000..c69a24399 --- /dev/null +++ b/manuscripts/publisher/scripts/eta_vs_abrikosov_gorkov_20261113.py @@ -0,0 +1,57 @@ +"""Publisher 2026-11-13: is "eta == Abrikosov-Gor'kov" decided by "AG lowers T_c"? + +The paper (superconductivity.md, S616 audit) prints T_c = Delta / (1.76 k_B eta), 607 K for YBCO, and +"eta == AG pair-breaking efficiency ... identical to the standard AG framework". +PREDICTIONS.md 2026-11-12 (iii) calls the identity a label: "AG pair breaking enters through the digamma +equation and lowers T_c". That compares against T_c0. The printed formula compares T_c against the GAP: +eta = Delta / (1.764 k T_c). So the question is what AG does to Delta/(1.764 k T_c), and that depends on +which Delta is meant: + (a) the AG order parameter Delta(0), or + (b) the spectral gap Omega_g = Delta (1 - (alpha/Delta)^(2/3))^(3/2), what tunnelling/ARPES measures. +Only the gapped branch (alpha <= Delta, alpha/alpha_c <= 0.912) is used; units Tc0 = k = 1. + +Run: python3 -I eta_vs_abrikosov_gorkov_20261113.py +""" +import numpy as np +from scipy.special import digamma +from scipy.optimize import brentq + +kB = 8.617333262e-5 # eV/K +print("Printed formula, site inputs (Delta = 35 meV):") +for eta in (0.38, 1.0): + print(f" eta = {eta:4.2f}: T_c = {0.035 / (1.76 * kB * eta):6.1f} K") + +D00 = np.pi / np.exp(np.euler_gamma) # 1.764, BCS Delta(0)/k Tc0 +a_c = D00 / 2 # AG critical pair-breaking energy at T = 0 +a_gapless = D00 * np.exp(-np.pi / 4) # alpha = Delta at the gapless onset + + +def Tc(a): + if a == 0: + return 1.0 + return brentq(lambda t: np.log(1 / t) - (digamma(0.5 + a / (2 * np.pi * t)) - digamma(0.5)), 1e-12, 1.0) + + +def Delta0(a): # gapped branch: ln(D00/D) = pi a / (4 D) + if a == 0: + return D00 + return brentq(lambda D: np.log(D00 / D) - np.pi * a / (4 * D), a, D00) + + +print(f"\nAG, gapped branch to alpha/alpha_c = {a_gapless / a_c:.3f}") +print(" a/a_c Tc/Tc0 eta(a) order-param eta(b) spectral") +rows = [] +for frac in np.linspace(0, 0.91, 92): + a = frac * a_c + t, d = Tc(a), Delta0(a) + om = d * (1 - (a / d) ** (2 / 3)) ** 1.5 + rows.append((frac, t, d / (D00 * t), om / (D00 * t))) +for r in rows[::10]: + print(f" {r[0]:5.2f} {r[1]:.4f} {r[2]:.4f} {r[3]:.4f}") +ea = np.array([r[2] for r in rows]); eb = np.array([r[3] for r in rows]) +print(f"\n(a) order parameter: min eta = {ea.min():.4f} -> eta < 1 unreachable on this branch") +cross = [r for r in rows if r[3] <= 0.38] +print(f"(b) spectral gap: eta falls monotonically: {bool(np.all(np.diff(eb) < 0))}; " + f"eta = 0.38 first reached at alpha/alpha_c ~ {cross[0][0]:.2f} (Tc/Tc0 = {cross[0][1]:.3f})") +print("\nVerdict: 'AG lowers T_c' does not decide the identity. Under (a) eta = 0.38 is outside AG;" + " under (b) AG reaches it. Which Delta the formula uses is not stated in the paper.") diff --git a/manuscripts/publisher/scripts/z_null_electronegativity_20261102.py b/manuscripts/publisher/scripts/z_null_electronegativity_20261102.py new file mode 100644 index 000000000..eb5b05b64 --- /dev/null +++ b/manuscripts/publisher/scripts/z_null_electronegativity_20261102.py @@ -0,0 +1,26 @@ +"""Test S651's premise: is Pauling electronegativity 'near-monotonic in Z' so that a smooth +monotonic / polynomial-in-Z null reaches r >= 0.95 'by construction'? Pauling values (CRC), Z=1..54, +noble gases without a Pauling value omitted. Typed from standard tables, not loaded from a database.""" +import numpy as np +EN = {1:2.20,3:0.98,4:1.57,5:2.04,6:2.55,7:3.04,8:3.44,9:3.98,11:0.93,12:1.31,13:1.61,14:1.90,15:2.19, + 16:2.58,17:3.16,19:0.82,20:1.00,21:1.36,22:1.54,23:1.63,24:1.66,25:1.55,26:1.83,27:1.88,28:1.91, + 29:1.90,30:1.65,31:1.81,32:2.01,33:2.18,34:2.55,35:2.96,36:3.00,37:0.82,38:0.95,39:1.22,40:1.33, + 41:1.60,42:2.16,43:1.90,44:2.20,45:2.28,46:2.20,47:1.93,48:1.69,49:1.78,50:1.96,51:2.05,52:2.10, + 53:2.66,54:2.60} +Z = np.array(sorted(EN)); y = np.array([EN[z] for z in Z]) +from scipy.stats import spearmanr +print(f"n={len(Z)} Spearman(EN, Z) = {spearmanr(Z, y).correlation:+.3f}") +for deg in (1, 2, 3, 5): + fit = np.polyval(np.polyfit(Z, y, deg), Z) + print(f"polynomial-in-Z degree {deg}: r = {np.corrcoef(fit, y)[0,1]:.3f}") +# best monotonic function of Z = isotonic regression (upper bound for ANY monotone null) +def iso(y): + blocks = [[v, 1] for v in y] + out = [] + for b in blocks: + out.append(b) + while len(out) > 1 and out[-2][0] > out[-1][0]: + v2, n2 = out.pop(); v1, n1 = out.pop(); out.append([(v1*n1+v2*n2)/(n1+n2), n1+n2]) + return np.concatenate([[v]*n for v, n in out]) +for name, yy in (("increasing", y), ("decreasing", -y)): + f = iso(yy); print(f"best monotone-{name} fit of Z: r = {np.corrcoef(f, yy)[0,1]:.3f}") diff --git a/manuscripts/publisher/state/whitepaper_sync.json b/manuscripts/publisher/state/whitepaper_sync.json index 691cd87da..16b9ca029 100644 --- a/manuscripts/publisher/state/whitepaper_sync.json +++ b/manuscripts/publisher/state/whitepaper_sync.json @@ -1,8 +1,8 @@ { "version": "1.0", "whitepaper": "synchronism", - "last_review": "2026-10-27T00:00:00Z", - "last_integration": "2026-10-27", + "last_review": "2026-11-20T12:30:00Z", + "last_integration": "2026-11-18", "sessions_reviewed_through": "no new numbered sessions (Archivist 09-03, 24h window, RUN-ID 1656: 25th deliberate zero, arc parked 83 days by git). Window 2026-09-02T10:30 -> 2026-09-03T10:30 UTC. NOTE THE WINDOW IS REALLY 48h OF THIS LANE'S WORK: the 09-02 fire reached Step 0 and the web4 leg, committed those, launched the SPARC reproduction as a background job, yielded its turn to wait on it, and died at exit=0 (554 B log) with 11 of 27 eps0 solves written and no analysis; the supervisor swept the partial into 4f245b14. Synchronism origin/main in the window: 4f245b14 (supervisor sweeping this lane's orphan), 9f0363e9 (Archivist), 97bf2843 (verify/route on the eps0 question). synchronism-site origin/main (HEAD also main): c4b8022 + 7f7ba88 (EXPLORER 09-02: the last escape CLOSED -- eps0(M_bar) is MOND-induced; treatment B re-solved after a misread column; Brouwer+2021 lensing kill), 2b3af8d (supervisor maintainer log). web4 origin/main (HEAD also main): 16 commits, 2 touching whitepaper/ and both are the OTHER Publisher unit's run log + a build guard, 0 spec content.", "pending_proposals": [ "EXECUTED 2026-09-03 (was: \"Move the Publisher run log out of whitepaper/PUBLISHER_CONTEXT.md into publisher/. Appending run history to the document the Phase-1 subagent reads as *context* grows the context doc without bound.\"). Done by the 04:30 systemd unit, not the 03:30 lane. Measured before acting: section 6 was 188,801 B / 53 entries = 89% of Synchronism's context doc and 590,764 B / 90 entries = 96% of web4's, against a section 11 in both that says 'Read this entire document first -- it's your complete context'; a cat of the web4 file exceeded the tool output limit, so that instruction was not executable. Logs: manuscripts/publisher/PUBLISHER_LOG_synchronism.md (HERE -- deliberately outside whitepaper/**, because build_whitepaper.yml's push trigger is whitepaper/** with PUBLISHER_CONTEXT.md negatively excluded for exactly this reason; a sibling whitepaper/PUBLISHER_LOG.md would have re-armed the rebuild+deploy churn that broke the v1 freeze) and web4:whitepaper/log/PUBLISHER_LOG.md (safe there: web4 CI is contents:read, parity-only, never writes back). Entries moved VERBATIM, 53/53 and 90/90, '### ' heading counts conserved, every non-empty source line accounted for in context-or-log; the only line not preserved is the heading '## 6. Recent Changes (Last 5 Integrations)', which was labelling 53. Read path fixed too: manuscripts/publisher/CLAUDE.md pointed at 'PUBLISHER_CONTEXT.md Section 6' for prior developments and now points at the logs, with both Inputs lists naming the run log separately from the standing context. NOT done and still dp's call: the repo-root build/ .gitignore decoy and the *.md text eol=lf .gitattributes fix. The workflow was deliberately NOT edited -- .github/workflows/build_whitepaper.yml is itself in the push paths, so a comment-only edit would fire the churn its comment documents. TEST OF WHETHER THIS LANDED: does the next pass append to the log, or back to section 6?", @@ -17,12 +17,22 @@ "Retire log-size and header-presence as Publisher liveness diagnostics, in this track and in the Archivist's. Since the 2026-07-24 launcher fix added `2>&1 | Add-Content`, the agent's output is flushed only at completion, so a live run's log is ALWAYS header-only when the agent inspects it -- the signature that meant 'never started' before 07-24 now means 'currently running', and the detector can only return 'dead'. This produced false failure reports on 08-01 and 08-02 (both runs exit=0) and a spurious OWNER-ACTION to the supervisor requesting a redirect already present. Same trap misread four times in the Archivist track (07-26/27/29/30). The 2026-08-03 launcher change (022ada55: timezone + UTC + RUN-ID in the header) fixes the ambiguity for the Publisher; the correct predicate is the CLOSING banner, never the byte count, and it should be applied to every track's launcher. Owner: supervisor." ], "status": "current", - "notes": "2026-10-27 (interactive maintenance pass): ONE corrective integration at four sites (glossary Gamma entry, conclusion 'gamma ~ 2 universality' bullet, 5.12 phase table + 'same gamma ~ 2 in BCS' sentence) plus two scoping notes on 'classical limit' (5.12 L13, Appendix B.1). Window 26067ad9..a00828b9 = e03d771b + 5ff4fced (back-annotations 10-26) + #12 merge + CI deploy. e03d771b (i): site withdrew quantum/classical endpoint labels for C/gamma (condensates = large N_corr = gamma -> 0). Paper's labels run the stat-mech direction (gamma=2 = N_corr=1 'classical'), not reversed; but the paper printed 'gamma ~ 2 in BCS superconductivity' as cross-domain validation, which under gamma = 2/sqrt(N_corr) asserts N_corr ~ 1 for a condensate and contradicts its own S676 bullet (BEC at gamma -> 0). CHECKED from the paper's own equation + S676, S616 eta audit (gamma/BCS reparametrization), 5.15 gamma=2 verdict. (ii) TEST-26 kill text, (iii) plotter, (iv) rho_crit six objects, 5ff4fced label fix: 0 hits each. Build CI=true md/pdf/web-clean rc=0, Missing character 0, U+FFFD 0, 292 pp (290 deployed), literal ** in section_0/5/7/8/9.html 0; artifacts restored. Instrument: section content files are .md not index.md (PUBLISHER_CONTEXT 9 wording fixed); ugrep rejects .{0,250} bounded repeats. | 2026-10-26 (interactive maintenance pass): ONE corrective integration in sections 0 and 7, stacked on #12 (main merged in after #11 landed). Window 278848c2..26067ad9 = 379b3b7f + 2a5cf44d (back-annotations 10-24) + #11 merge. 379b3b7f (iii): S610 sigma_int = 0.086 is not a BTFR scatter -- residual of log M_b on 5 predictors, 3 built from M_b ingredients; shuffled/no-V 0.0870 vs 0.0864 (velocity 0.0006 dex); target = Taylor+2011 colour identity to 0.0037 dex. Paper printed it as 'exactly CDM prediction ... MOND rejected at 32 sigma' (exec L178) and in conclusion L65: annotated CHECKED 2026-10-24 at both. Both site scripts (btfr_0086_identity_control.py, taylor_mstar_is_photometry_check.py, PREREG b134d0f) RE-EXECUTED from /tmp: control output byte-identical, Taylor differs only by loader banner. Count unchanged (CDM vs MOND, not this ledger). Not annotated, untested: S593-594 'TFR residual = complete M/L predictor' (exec L177) may share the construction. 379b3b7f (i) +10.9 comparator = McGaugh nu, Dk=0: the paper prints +10.9 at 5 sites (exec x2, conclusion x2, 5.15) beside two free-gamma tests and never names the DBIC comparator -- dated comparator parenthetical added at all 5 (read from gamma2_pin_galaxy_level.py L115/L144 + committed output C1/P3); (ii) count range 3-8 and (iv) P(k) pin: paper prints neither (0 hits). 2a5cf44d (third live experiment) gates on dp, 0 hits. Build CI=true md/pdf/web-clean rc=0, Missing character 0, U+FFFD 0, 290 pp (288 at #12's build; +1 per edit), literal ** in section_0/5/7.html 0; artifacts restored (CI deploys on merge). | 2026-10-24 (interactive maintenance pass): ONE corrective integration in 15-dark-matter. Window a3630e40..278848c2 = 278848c2 (back-annotate 10-23). (ii) covariant RG (Sanna, Matsakos & Diaferio 2023) is unscreenable omega=-1 Brans-Dicke: gamma_PPN=0 (Cassini ~4e4 sigma), eta_N=1 (LLR ~2e3 sigma), |Gdot/G|>=0.99 H0 (>=900 sigma), weak field varies phi <~1e-4 so it does not contain RG; the paper credited 'its covariant completion is in A&A' -- annotated CHECKED 2026-10-23 at 4 sites (PPN/Gdot arithmetic re-derived; trace-eq derivation, 900-sigma floor, Yukawa solve relayed). (i) gamma identified only jointly with rho_crit: paper prints no gamma-as-sharpness (0 hits), recs gate on dp. Build 288 pp, Missing character 0, U+FFFD 0, section_0/5/7 literal ** 0. Stacked on #11. | 2026-10-23 (interactive maintenance pass): no change. Window a4e86384..a3630e40 = 81303040 (back-annotate from site 10-22: 'density-keyed branch excluded' scoped to this framework's permittivity; RG record in one place; adopting RG cannot move Bucket 0, gates on dp) + #10 merge + CI deploy. Paper prints no unscoped exclusion: 'no galaxy-scale prediction... surviving branch' 0 hits, lopsided/dipole/T_scan/measurement-without-observers 0 hits; the +2,182 paragraph already names 'the density-keyed equation, which is the actual novel claim' and L165 already says RG's eps0 is fitted jointly so a pinned refit is the honest test. Deployed PDF (5414fc1c, docs=build) 287 pp, U+FFFD 0, section_0/5/7 HTML literal ** 0; all 26 PDF text-layer ** lines are changelog prose. | 2026-10-22 (interactive maintenance pass): TWO edits. (1) [TRACED 2026-10-22] note after the 10-21 [CONTESTED] one in sections 0/7: c5364f99 + site explorer finding trace S662's demotions to AI tracks (5/6 AI-made edits, all eight site commits carry Claude co-author trailers, verified; Born rule no badge/edit on record; 'human audit' entered via e8752205, Sonnet 4.6). Flagged for dp: narrowed claim's 'AI loops fail' holds only for the A2ACW loop. (2) markup-only fix of the 27-literal-** baseline: 15 body literals in sections 0/7 removed (five shapes, each a later insert splitting an existing bold span), text byte-identical with * stripped; body literals 15 -> 0; PDF text layer 27 -> 24, all 24 in changelog prose (12 old + 12 from this pass's own entries). Raised: build.sh mode 644; make-web-clean.sh lacks cwd guard. | 2026-10-21 (interactive maintenance pass): ONE corrective annotation stacked on #9. 4d210188 (10-20 back-annotation) contests S662:50 'the 6 demotions came from human audit'; the paper prints it in exec summary + conclusion, now [CONTESTED 2026-10-21]. Partial trace run: site ac093af (2026-04-23, Claude Sonnet 4.6 co-author, LLM persona trigger) demoted a0 and A; 4/6 untraced. LLR/L3 and fsigma8 0.4733 items: 0 hits in paper. | 2026-10-19 (interactive maintenance pass): ONE corrective annotation stacked on #9. PREDICTIONS.md 10-18 (v) (fada3e79): the +2,843 density-vs-acceleration dBIC survives a galaxy-block bootstrap (95% [2178,3590], N_eff~140); 5.15 paragraph (3) prints +2,182/+129 and now carries a [SCOPED 2026-10-18] note. Identity + analytic galaxy sd re-executed from /tmp, matching; bootstrap quoted. 2026-10-18 (interactive maintenance pass): ONE corrective integration. The 09-29 PREDICTIONS.md block (d58b643d) withdrew '+184 survives any convention' (galaxy level: N_eff~150, dBIC +10.9, rule convention-dependent); it landed in the 09-03..10-07 publisher outage and never reached the paper. [SCOPED 2026-10-18] annotations in exec summary x2, conclusion x2, 5.15 x1; script re-executed byte-identical. PR publisher/wp-2026-10-18. 2026-10-14 (interactive maintenance pass): no edit. Window d4bf7d85..d692ee91; #8 (10-12 LRG1 QSO-row annotations) merged 10-13 and CI-deployed: docs/ PDF = build/ blob 6abdd91a, 282 pp, 0 U+FFFD. 8e53fba4 back-annotation touches nothing the paper prints (counting-rule item gates on dp). STALENESS REPAIRED: this file's last_review read 2026-09-03 through the 10-07, 10-10, 10-12 and 10-13 passes, while PUBLISHER_CONTEXT.md section 10 names it as the authoritative current state; last_integration now 2026-10-12 (afcf7997). Detail: manuscripts/publisher/PUBLISHER_LOG_synchronism.md. | 2026-09-03 second pass (systemd autonomous-publisher-cbp, 04:30 PDT lane; the 03:30 lane's own run is the entry above this one): NO CONTENT CHANGE to either whitepaper, both verified current, and the pass's yield is a structural repair to this lane's own instrument. Synchronism verified at HEAD rather than from the 03:30 report: 'AMENDED 2026-09-03' x2 in both monoliths, x1 in section_5.html and section_11.html on both web surfaces, and x2 in the TEXT LAYER of both committed PDFs -- the PDF is not stale, no rebuild owed; exec/conclusion parity 690 core. Web4 verified by rebuild-and-diff in a fresh worktree at 323aff79: all six surfaces reproduce, monoliths differ only on the Generated: line, index.html and assets/ byte-identical (absent from git status), both PDFs 4 pdftotext lines apart and all four are the timestamp. THE FINDING: the context doc had inverted -- 96% of web4's and 89% of this one's was run history, and section 11 tells the subagent to read the whole thing as its context. That is not a tidiness complaint; it is unexecutable, and this pass proved it by hitting the tool output limit on the first read. Split executed (see pending_proposals[0]). THE NEAR-MISS WORTH KEEPING: the obvious destination -- a sibling whitepaper/PUBLISHER_LOG.md -- is inside the CI push trigger, and the workflow already carries a negative exclusion for PUBLISHER_CONTEXT.md precisely because log appends were firing rebuild+deploy on a non-byte-reproducible build and committing ~754KB PDFs against identical source. The filed proposal said 'into publisher/' and was RIGHT for a reason it never stated; my instinct to keep the log adjacent was wrong, and only reading the workflow caught it. General form: a path-filtered CI gate makes file PLACEMENT a correctness question, and a negative path exclusion protects one filename, not a class. | 2026-09-03 run: the methodological finding is about this lane's own principal instrument. Today's reproduction matched the site 77/77 EXACT -- and reprinted a VOID number while doing it. Line 41 of both outputs reads 'rho_c at top edge 59%'; column 1 of epsilon0_per_galaxy_fw.npy is per-galaxy chi2, not rho_c (writer epsilon0_free_the_ceiling_rescue.py:220, np.vstack([e0s, best_per_gal, MOND_F, NN]); col1 median 49.4, max 1.17e4 against a rho_c grid of 1e-6..1e-2 -- four orders out and dimensionally wrong). Both runs read col 1, so both printed it, so the comparator scored it exact. A REPRODUCTION GATE IS BLIND BY CONSTRUCTION TO A MISLABELLED INPUT: it certifies that the arithmetic ran the same way, and the defect lives upstream of the arithmetic. To catch that class you must read the WRITER of an artifact, not the reader. This cuts directly against this lane's 09-01 'reproduced 11 of 11 before inscription' -- that gate would not have caught it either. Verified at the writer here rather than accepted from the site's correction. | Second item, on the OTHER side of the same coin: a coverage caveat applied to the leg I liked. The site's lensing table assumes Brouwer+2021 reaches g_bar = 1e-15 and quotes deficits to 87x; the abstract says '2 decades', which puts the floor at ~1e-14 and the deficit at 6-28x. Screened at abstract level, said so inline, inscribed the conservative column. | Third: the 09-02 death is a FOURTH mode and it is the dangerous one -- exit=0, not exit=1. The pass yielded its turn to a background job and died having produced 12.5% of one grid and no analysis; the supervisor then swept the partial under a commit subject that reads as a completed run ('last-escape reproduction run 2026-09-02 (SPARC 153 galaxies vs MOND)'). A rescue inherits the headline of the thing it rescued. Step 0 caught the tree state; nothing caught the SUBJECT. Today's rule: never yield the turn to a background job -- poll it with real work in between, which is what this pass did (launched at 03:32, read the site finding, updated REC-042, committed, and collected the result at 03:37). | 2026-08-30 run: no research to integrate and no whitepaper change proposed -- the whole pass is the sweep of the 08-29 loss plus its diagnosis. The 08-29 content was committed today as 14c0288f (sources only; build verified from whitepaper/, 10 sections, 7913 lines, the 08-29 block present in the monolith; local rebuild carried 24,110 raw lines against 44 content lines and was restored -- CI has since built and deployed it). CAUSE of the 08-29 loss, from the launcher log the Archivist could not see: credits exhausted at 03:49:06, 19 minutes in, AFTER every artifact was written and BEFORE git commit. Not a fault in the record step -- the record steps are LAST. Publisher deaths are bimodal (08-13 and 08-17 died at 14 s having produced nothing; 08-29 at 19 min having produced everything) and the cheap mode taught every watcher that exit=1 means nothing happened, which is why it sat 25 h. Step 0 added to publisher/CLAUDE.md: git status first, sweep the previous run, and commit content as it is written. MEASUREMENT: deaths across the four tracks are CORRELATED (all four down 08-13 and 08-17; three of four 08-29) because they share one credit pool, so a correlated outage is a loss of the", + "notes": "2026-11-20 (autonomous-publisher-cbp, stacked on #14 at 5cc49536 + merge of origin/main 6905fd1b): one dated CHECKED note on 15-dark-matter kill-criteria item (3) (TEST-10), words added none removed, count HELD at 6. (a) 11-16 escapes (M/L 8.4/11.6 at k=4, 6.3/8.7 at k=3; tidal 1.26-1.33x) are local-3-sigma passes (source cap6p39_vs_dsph.py L48-55); at the 3.51 trials threshold they are 5.3 (Crater II)/6.9 (Draco) at k=3, inflation 1.15x/1.31x = site 11-19 figures. (b) Site 11-19 Walker+2009 row (Crater II z 2.9, Draco 3.7, 'estimator-conditional') equates M( Walker = half-light form with k_eff 2.71; z at M/L=4 = 3.96/4.31, both past 3.51; escape M/L 4.7/6.2 (simulations/publisher_20261120_dsph_walker_aperture.py, identity control reproduces all 6 site rows). ROUTED FOR SITE MAINTAINER: correct the Walker row and the 'estimator-conditional at its edge' sentence in PREDICTIONS.md 2026-11-19 (i). Window items (ii) LLR measured gradient, (iv) delta widths: 0 hits; (iii) +2843 tilt: additive, paper makes no where-from claim, not edited; (v) 2x2 ablation + 6905fd1b design fixes gate on dp. Build verification only: 314 pp local, Missing char 0, U+FFFD 0, geometry 1 past w-36 (baseline), 0 past w; artifacts restored. Deployed PDF blob 320e7fa0 = c3e88fec. PRIOR: 2026-11-19 (autonomous-publisher-cbp, stacked on #14 at 15d64c36 + merge of origin/main 0ac548ed): no sections/ edit. Window 0ac548ed (11-18 back-annotation, 6 items): door #3 closed by basis, LLR register at 2-sigma (1e-8-1e-7), width 1.17 dex = gamma->inf limit, Ciocan n=1 bar convention, DR4 companion rate, post-cutoff arm default: none printed in live sections (0 hits each; paper's Ciocan block is branch-A E(z) with anchors, not the bin-slope n reading; paper's wide-binary text is DR3 C(a)). Deployed PDF on main: 292 pp, U+FFFD 0, build==docs, 1 past width-36 (p6 baseline), 0 clipped. #14 CI green, no review since NOVA 11-13. Undeployed on #14: 33 commits, 32 dated notes after 10-27 (0 on main). Count HELD at 6. PRIOR: 2026-11-18 (interactive maintenance pass, stacked on #14 at f99ebc65 + merge of origin/main 5d2d684c): two dated CHECKED notes in 06-implications/04-open-questions (OQ-Coarsening), words added none removed, count HELD at 6. 'x is a virial ratio, capped at ~0.02 for every gravitationally bound system' and '~40x below the knee' hold only for rho_crit = A*V^2 with one length shared by A and the smoothing of rho (back-annotation 5d2d684c (ii)). S53's 0.028*V^0.5 is that law with R_half ~ V^0.75 inside A; executed fits pair it with Sigma/2h at h = 0.3 kpc; under that pairing 62/175 SPARC discs reach x>1 on stellar Sigma alone (lower bound); NGC 2403/3198/7331 x = 2.8/2.6/3.9, DDO 154/UGC 128 0.07/0.21 (re-run, simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py, identity controls PASS). Window items (i) B1 'no-signaling bound' wording and (iii) Ciocan sigma hygiene: not printed in the paper (0 hits). (iv) frame proposal (external value has moved to other people's theories) gates on dp; no paper edit. Web4: no change (origin/main 1ea6fb21, build/=docs by blob). Previous notes: 2026-11-17 (interactive maintenance pass, stacked on #14 at 4fc98005 + merge of origin/main d71f1f8b): seven dated notes, words added none removed, count HELD at 6. (1) 15-dark-matter gamma = 2 bullet: 'SPARC selects gamma ~ 0.489 ~ 1/2, the exact MOND point' CHECKED: template fitted with no data gives 0.500 (simple mu) and 0.595 (McGaugh RAR, 0.52-0.60 across g_bar ranges); nuisance-refit SPARC 0.58-0.62 cannot separate them; 0.489 is the fixed-Upsilon slice; pointer notes at the status-table cell and L91 (PREDICTIONS d71f1f8b (i); explorer gamma_incumbent_map/{sc2,nest,rng,dmap,dict}.py re-run byte-identical). (2) 16-superconductivity table 'superconductor gamma ~2' CHECKED: BCS's gamma at pinned knee (2.644 from BCS curve, no data; beta free -> unidentified); text above the table reads the same tanh as gamma = 1. (3) OUTSIDE WINDOW: 'BCS ratio derived 2 sqrt(pi) = 3.54, observed 3.52 (<1%)' CHECKED at 5.16, conclusion Progress list (check mark kept) and Appendix B: 3.52 is BCS's own 2 pi exp(-0.5772) = 3.528 from the gap equation, 2 sqrt(pi) is 0.49% away by coincidence, the source script's gamma = 2 formula gives 7.12 (simulations/publisher_20261117_bcs_ratio_and_gamma_check.py). NOT edited, gates on dp: exec summary L221 'N_corr convergence test: currently mixed' vs d71f1f8b (ii) 'answered diverging by construction' (Bucket 3 edit gates on the project lead). Build verification only: md/pdf/web-clean rc 0, Missing char 0, U+FFFD 0, 311 pp local, geometry 1 past w-36 (baseline), 0 past w; artifacts restored.", "watch_items": [ { "opened": "2026-10-18", "item": "Outage-window backlog: 10 PREDICTIONS.md blocks dated 09-14..10-07 correct claims the paper still prints unscoped (ranked list in PUBLISHER_LOG 2026-10-18 entry). Each needs source re-verification before inscription; items 1/3/4 + parts of 5/9 gate on dp. Scale question (status banner vs rewrite vs per-item annotation) raised to dp.", - "status": "open" + "status": "open", + "progress": [ + "2026-10-29: item 1 partly addressed (09-18 sweep printed at L199 with the floor presuppositions).", + "2026-10-30: item 7 (factor-1.96 a0) CLOSED, CHECKED from L109. 09-23 thawing-vs-crossing wording (not in the 10-18 ten) CLOSED, CHECKED by CPL algebra.", + "2026-10-30: an independent re-scan (read-only subagent) re-found items 3, 4, 6, 7 and 8 with matching quotes, plus 09-14 missing Cardassian attribution (part of item 5) and 09-23. Items 2 (r=0.994) and 10 (2pi DERIVED) read as scope gaps there, not contradictions. Still open: 1 (rest), 2, 3, 4, 5, 6, 8, 9, 10. 6 and 8 rest on site-executed numbers (10^-5 pin; B_req = nu/f), so they need a re-run before inscription.", + "2026-10-31: item 1 detail: the L199 sweep numbers unrounded to 0.406/0.297/0.260 (PREDICTIONS 10-30 (ii), re-run). The 0.30 'on the strict threshold' was 0.003 under it. The recount question (does a convention-dependent kill count) still gates on dp.", + "2026-10-30 instrument note: the 10-29 log proposed this sweep as new ('I have not run it') while this item was first in watch_items. Read watch_items before proposing work.", + "2026-11-13: full disposition at #14 head ca78abce in reports/outage_window_backfill_audit_2026-11-13.md. CLOSED 7, 8 (11-11/11-12), 09-23. PARTIAL 1 (headline exec L189 / 5.15 L5 / concl L77 unscoped), 5 (11-06 Cardassian at L342; L364 open), 10 (table row 11-11 + in-cell 11-13; L256 'not numerology' open). OPEN 2, 3, 4, 6, 9. Banner option recommended to dp.", + "2026-11-14: no change to items; build-only clipping fix split out of #14 as #16 so a formatting repair stops waiting on the editorial hold.", + "2026-11-15: item 6 CLOSED (L366 CHECKED note; de_horn_l_as_a_fluid_jeans_term.py + controls re-run byte-identical). Item 4 untouched: the new L117 striction note scopes magnitude only and says the EFE statements are a separate open item. Still open: 1 (headline), 2, 3, 4, 5 (L364 half), 9, 10 (L256)." + ] }, { "opened": "2026-09-03", @@ -30,7 +40,8 @@ }, { "opened": "2026-09-03", - "item": "Brouwer et al. 2021 (A&A 650 A113, arXiv:2106.11677) is inscribed in 15-dark-matter at ABSTRACT-LEVEL coverage. The full text has not been read in this lane. Two things a full read would settle: the actual lowest g_bar reached (the abstract says '2 decades'; the site assumes 1e-15), and the stellar-mass binning, which is what the deficit table is indexed on. A coverage caveat is a work item." + "item": "Brouwer et al. 2021 (A&A 650 A113, arXiv:2106.11677) is inscribed in 15-dark-matter at ABSTRACT-LEVEL coverage. The full text has not been read in this lane. Two things a full read would settle: the actual lowest g_bar reached (the abstract says '2 decades'; the site assumes 1e-15), and the stellar-mass binning, which is what the deficit table is indexed on. A coverage caveat is a work item.", + "status": "closed 2026-11-11: full text read in fragments (ar5iv, fetch subagent). Lowest fiducial g_bar = 1e-15 (Sec. 1; lensing bins 1e-15..5e-12); footnote 3 allows ~x10 from hot gas, dust, missing baryons. Stellar-mass binning NOT read; the 5.15 note says no data point was read." }, { "opened": "2026-09-03", @@ -49,9 +60,14 @@ "KEYING: the 2026-08-05 EFE table in 15-dark-matter uses rho_crit = 0.029 V^2 (compilation layer). The primary (Session 53) derives A = 0.028 for V^0.5; S687 (06-07) recorded the switch; the site measured the exponent on 08-27 (V^-0.16 +/- 0.19: neither). Verdict not-evaluable is convention-independent (noted in place); the 38-92x figures are not. Any future evaluation of C(rho) on real galaxies must state which rho_crit law it uses and why.", "RESEARCH-LANE SURFACES carrying superseded text (routed via explorations/2026-08-28-publisher-*, NOT edited by this lane): SESSION_FOCUS.md 07-02 entry ('wrong SIGN + ~240-300,000x too high') and 08-27 entry ('ceiling below RG's ENTIRE fitted range'); explorations/2026-07-02-triage-rho-crit-*.md. claims/v1-snapshot/ is frozen and carries the old PREDICTIONS row by design.", "SITE (maintainer 401 since 08-12, day 15 at 08-27): seed explorer/topics/universal-rho-crit-knee-sparc-test.md says 'the model nobody has written down' -- it is Cesare+2020 sec. 5. Cannot be corrected from this lane; recorded here and in the exploration note so the seed's execution starts from the published fit (half the RCs worsen; RAR under-predicted 0.1-0.3 dex at low g).", - "The striction numerics (164x K_z excess; max(L2/L3) -> 1/C_min = 3.175) are SITE-EXECUTED and have never been re-run in this repository. They are now REC-042's lead content, so the same rule that killed leg (3) applies to them.", + "The striction numerics (164x K_z excess; max(L2/L3) -> 1/C_min = 3.175) are SITE-EXECUTED and have never been re-run in this repository. They are now REC-042's lead content, so the same rule that killed leg (3) applies to them. PARTLY ANSWERED 2026-11-15: the archive-knee striction scripts (l3_striction_archive_v05_knee.py, l3_vertical_striction_archive_v05_posthoc.py, l3_vertical_striction_sign_posthoc.py) re-run here with identical output; the 164x K_z and max(L2/L3) figures are still not re-run.", "BUILD SURFACE: section_11.html exists in the built artifacts with NO corresponding whitepaper/sections/11-* source directory (sources are 00-09; it is generated from the section CHANGELOG). A blast-radius grep keyed on section numbering will miss it. Found the hard way today.", - "make-md.sh coverage guard PORTED from web4 (2026-08-26) and verified: exits 1 from the repo root, reports 'Sections found: 9' from whitepaper/. Closes the 2026-08-26 self-flag." + "make-md.sh coverage guard PORTED from web4 (2026-08-26) and verified: exits 1 from the repo root, reports 'Sections found: 9' from whitepaper/. Closes the 2026-08-26 self-flag.", + { + "opened": "2026-11-13", + "item": "eta == Abrikosov-Gor'kov (paper: exec L182, superconductivity L147, conclusion L68 print it as an identity). PREDICTIONS 11-12 (iii) calls it a label because 'AG lowers T_c'. That compares with T_c0; the printed formula compares with Delta. Order-parameter reading: eta_AG >= 1 (0.38 unreachable); spectral-gap reading: reaches 0.38 at alpha/alpha_c ~ 0.62 (scripts/eta_vs_abrikosov_gorkov_20261113.py). Neither reading is printed. Research lane to state the comparator before the paper changes.", + "status": "open" + } ], "last_integration_summary": "2026-09-03: ONE integration -- a 2026-09-03 amendment in 15-dark-matter closing the galaxy sector's last open question, which THIS SECTION had named as open and assigned to the site lane on 09-01. eps0(M_bar) is NOT a relation of the theory: fitting the same per-galaxy eps0 to MOND's own predicted curve induces rho_s = +0.846 vs log M_bar, STRONGER than the data's +0.758, and the joint fit leaves c = +0.110 [-0.071,+0.277] on M_bar (Freedman-Lane p = 0.099). Four rules adjudicated against thresholds pre-registered before any number existed: R1 not a relation (0.518 vs bar 0.369); R2 distance guard PASSES (a0 +0.063, p 0.38); R3 matched, class loses 1.75x vs a 1.20 bar (91.20 vs 52.09, wins 31% of discs, Wilcoxon 2.3e-7); R4 MOND-induced. Reproduced in-repo FIRST and completely: 108 lines vs 108, 77/77 numeric lines EXACT, 0 text diffs, seeded rng so permutation p and bootstrap CIs match exactly (313 s). Plus the unregistered result that is the sharpest in the sector: with ONE constant the class sits at chi2/N 110.79 from MOND's NOISE-FREE curve while MOND sits at 52.20 from the NOISY data -- a SHAPE result that bounds the 08-03 identity (C(x) at gamma=1/2 IS simple mu, exact to 5.6e-17) to the compander ALGEBRA and denies it to div(C grad Phi) = 4 pi G rho. Also inscribed: treatment B's residue (c = +0.136, real per galaxy, globally worthless, untested scale-height/Upsilon nuisance R2 is blind to BY CONSTRUCTION -- the sector's only open thread); Brouwer+2021 KiDS-1000 lensing, NEW to this repo, ceiling short 6-28x; and two void numbers withdrawn (per-galaxy chi2 read as rho_c), verified at the WRITER not accepted from the correction. Build verified from whitepaper/ (exit 0, 7959 lines, +29 as expected); churn gate content 58 vs raw 24,290 -> artifacts restored. Count HELD 6; Bucket 0 = 0 -- not a 7th refutation, but the last CANDIDATE for Bucket 0 = 1 is gone." } diff --git a/manuscripts/publisher/state/whitepaper_web4.json b/manuscripts/publisher/state/whitepaper_web4.json index 31dc242c6..046c37f08 100644 --- a/manuscripts/publisher/state/whitepaper_web4.json +++ b/manuscripts/publisher/state/whitepaper_web4.json @@ -1,9 +1,9 @@ { "version": "1.0", "whitepaper": "web4", - "last_review": "2026-10-27T00:00:00Z", - "last_checked_commit": "9689af42", + "last_review": "2026-11-20T12:30:00Z", + "last_checked_commit": "1ea6fb21", "pending_proposals": [], "status": "current", - "notes": "2026-10-27 (interactive maintenance pass): no change (log entry, web4#939). Window 8f15707e..9689af42 = 8 hub commits (LCT identity convergence Slices C/D: #930 design, #928 C, #933 D-a, #931 ledger-tail retry, #934 D-b1, #935 terminal status, #936 D-b2, #937 D-c). 0 web4-standard/ and 0 whitepaper/ changes. build/ = docs/ for PDF b6cb070a (24 pp, U+FFFD 0), monolith 35487453, index.html 4b24dc1b. Watch: a normative LCT-spec change from Slices C/D would owe section 6 an update. web4#909 superseded and now CONFLICTING, still needs a human close. | 2026-10-26 (interactive maintenance pass): no change. Window 68e43160..8f15707e = #926 (docs/ non-normative position paper on contextual sovereignty and lawful governance; it says itself the core spec does not yet state its nearest-grant rule normatively) + #927 (web4-core HUB_CHANNEL_PURPOSE constant). 0 web4-standard/ and 0 whitepaper/ changes. build/ = docs/ PDF blob b6cb070a, 24 pp, U+FFFD 0. web4#909 still superseded (head 79507837 not an ancestor of main but its 10-12/13/14 entries are on main via #917), still needs a human close. | 2026-10-24 (interactive maintenance pass): no change. Window 86c6e3fc..68e43160 = #925 (docs/governance public landing standard + dated audit), 0 web4-standard/whitepaper changes; a README landing convention, not a normative spec. build/ = docs/ PDF blob b6cb070a. web4#909 still superseded, needs a human close. | 2026-10-23 (interactive maintenance pass): no change. origin/main still 86c6e3fc, empty window. build/ = docs/ PDF blob b6cb070a, 24 pp, U+FFFD 0, literal ** 0. web4#909 still OPEN/CONFLICTING; its three entries (10-12/13/14) are already on main's log, so it is superseded -- needs a human close. | 2026-10-22 (interactive maintenance pass): no change. Window d6250307..86c6e3fc = #924 (hub/docs LCT identity convergence Slice C PRD), 0 core-spec/whitepaper/sections changes. web4#909 still needs a human close. | 2026-10-21 (interactive maintenance pass): no change. Window 3eee7096..d6250307 = one docs/research note (KV cache provenance scope), 0 core-spec/whitepaper/sections changes; build/ = docs/ PDF blob b6cb070a. No web4 log PR (lighter path). web4#909 still OPEN/CONFLICTING -- close it. | 2026-10-20 (interactive maintenance pass): no change. Window 16dbebce..3eee7096 = #923 merge (10-19 log entry only), 0 core-spec/whitepaper/sections changes. NO web4 log PR opened for this no-change pass (lighter path per the 10-14 frame question); entry lives in the Synchronism log. web4#909 superseded (its 3 entries already on main) and CONFLICTING -- close it. | 2026-10-19 (interactive maintenance pass): no change. Window 728e3b0d..16dbebce: #922 (publisher log) + 16dbebce (docs/research landscape note), 0 core-spec/whitepaper changes; build/ = docs/ PDF blob b6cb070a. 2026-10-18 (interactive maintenance pass): no change. #917 merged 10-17 (5d09f706); deployed PDF = build/ blob b6cb070a, 24 pp, U+FFFD 0, web4-policy present. Window 38f1f321..728e3b0d: #920 hub routing + 4 hub/docs commits, 0 core-spec/whitepaper changes. 2026-10-16 (systemd autonomous-publisher-cbp): no change. Window c6ad5758..38f1f321, 1 commit (#916, hub-daemon REST only), 0 core-spec/whitepaper. web4#917 (web4-policy in section 11) still open, MERGEABLE, checks green, BLOCKED on review; its PDF 24 pp, 0 U+FFFD. 2026-10-15 (maintenance pass): INTEGRATED. Window f6ed4948..c6ad5758, 0 core-spec commits; #915 recorded web4-policy 0.1.0 first publish on crates.io (verified via API). Section 11 core-packages paragraph names it as the SAL law engine hub + hestia share; all six artifacts rebuilt (24 pp, 0 missing glyphs, 13 nav ids). PR web4#917, stacked on #909. | 2026-10-14 (interactive maintenance pass): no change. Window 3d4b9719..9aeee56e (#907, #912, #913), 0 core-spec commits since e8e4c345; fresh-worktree rebuild timestamp-only (24 pp, 0 missing glyphs, 13 nav ids). Shared web4 checkout no longer stale (main = origin/main, 0 modified). This file had last_checked_commit 0c78197a through the 10-12 and 10-13 passes. Log entry stacked on web4#909. | 2026-10-10 (systemd autonomous-publisher-cbp): scanned origin/main (HEAD also main, fetched first). Window 6e73dca0..0c78197a, 22 commits; 3 touch web4-standard/core-spec (0b4577bb, 1f3e3573, 6f072562 = PR #857, merged 10-09): Maintainer added as eighth base-mandatory role (society-roles.md 2.8/2.9), shipped in web4-core/SDK/hub/wasm. INTEGRATED: one paragraph in sections/10 + one glossary entry, all six artifacts rebuilt in a fresh worktree and committed with the source, PR from publisher/wp-2026-10-10. Baseline rebuild before the edit was timestamp-only. Deployed PDF at origin/main: 0 U+FFFD (the 10-07 glyph fix verified post-merge). Raise for the spec lane: web4-standard/GLOSSARY.md has no Maintainer/base-mandatory entry; docs/SPRINT.md still says 7. | 2026-09-03 second pass (systemd autonomous-publisher-cbp): CONTENT CURRENT, verified by REBUILD-AND-DIFF rather than by commit scan. Fresh worktree at origin/main 323aff79 (detached HEAD makes the build scripts abort -- they need a branch with --set-upstream-to=origin/main, as PUBLISHER_CONTEXT section 5 says). All three scripts run. All six artifact surfaces reproduce: both monoliths differ ONLY on the *Generated:* line; docs/whitepaper-web/index.html and assets/ are byte-identical and do not appear in git status at all (so the Pygments-generated HTML reproduces exactly under the guard merged in #790); both PDFs diff at 4 pdftotext lines, all four the generation timestamp (2026-08-26 04:45:01 -> 2026-09-03 04:33:42). Zero real drift, nothing rebuilt, no artifact committed. sections/ and all six artifacts last moved together at e9aee591 (08-26), so the #789 lockstep is intact. STRUCTURAL CHANGE MADE THIS PASS, going to a PR per the #789 contract: whitepaper/PUBLISHER_CONTEXT.md section 6 was 590,764 bytes / 90 entries = 96.3% of a 613,347-byte file whose section 11 instructs the subagent to read all of it as context. Moved verbatim to whitepaper/log/PUBLISHER_LOG.md (90/90 entries, heading count conserved 118 = 28 + 90, every non-empty line accounted for). Safe inside whitepaper/** here because web4 CI is contents:read and never writes back -- NOT safe in Synchronism, whose workflow negatively excludes PUBLISHER_CONTEXT.md from the same trigger because log appends were firing rebuild+deploy churn; its log went to manuscripts/publisher/ instead. The two repos look symmetric and are not. | 2026-09-03: scanned origin/main (HEAD was also main, stated per the 2026-08-09 rule; fetched first). Since 557b61e3: 16 commits in the window, 2 touching whitepaper/ -- 023d6022 and d0c63ea0 -- and BOTH are the other Publisher unit's own run log (whitepaper/PUBLISHER_CONTEXT.md) plus the make-web.sh Pygments guard. Verified by file list, not by subject: 023d6022 = PUBLISHER_CONTEXT.md only (+74/-1); d0c63ea0 = PUBLISHER_CONTEXT.md + make-web.sh. ZERO Web4 spec or content commits in the window. Status current, no proposals from this lane. Worth recording from 023d6022's subject, as the other lane's finding and not this one's: a stranded 08-30 log entry meant 'two passes had re-confirmed a census figure the missing entry refuted' -- the same failure class this lane measured on 08-29 (a lost record lets a corrected claim be re-derived as true), observed independently in a sibling repo. | 2026-09-02: scanned origin/main (HEAD was also main, stated per the 2026-08-09 rule; fetched first). Since 279c6d62: 11 commits in the window (2026-09-01T10:30Z -> 09-02T10:30Z), 10 hub/docs/chore, ONE touching whitepaper/: d0c63ea0 = the MERGE of PR #790 ('docs(whitepaper): log the 2026-08-27, 08-28, 08-31 and 09-01 Publisher passes'), merged 2026-09-02T04:49:40Z per `gh pr view 790` (state MERGED, mergeCommit d0c63ea0). STATUS CHANGE, verified not relayed: the make-web.sh Pygments guard is now ON MAIN (`git show origin/main:whitepaper/make-web.sh` lines 49-61), so the 2026-08-31/09-01 relay 'the fix main still lacks' is DISCHARGED; build_whitepaper.yml ran on main at 04:49:43Z (headBranch main, status completed, conclusion success) -- the first main-branch whitepaper build since the parity gate #789 landed. That whitepaper commit is the OTHER Publisher unit's (systemd autonomous-publis" + "notes": "2026-11-20 (autonomous-publisher-cbp): no edit, no rebuild owed, verified by rebuild. origin/main = 1ea6fb21; bbaf2673..origin/main touches no core-spec/ontology/whitepaper/docs-whitepaper-web path. Fresh worktree build: md/pdf/web rc 0; diff = Generated line only; index.html byte-identical; build = docs (cmp); 24 pp, Missing char 0, U+FFFD 0; geometry 8 words past 545 pt (11-02 baseline, max 596 p15). Churn restored. Open PRs touching paper/spec: none (#984, #994 not this lane's). PRIOR: 2026-11-19 (autonomous-publisher-cbp): no edit, no rebuild owed. origin/main = 1ea6fb21 (unchanged since 11-13; empty window). build = docs by blob (PDF 50ba1b5c, monolith 05842421, index.html 798a0d8d), so the 11-02 geometry baseline carries over unmeasured. Open PRs: #984 (draft hub mesh), not this lane's." } diff --git a/simulations/publisher_20261108_cassini_convergence_scan.py b/simulations/publisher_20261108_cassini_convergence_scan.py new file mode 100644 index 000000000..03d6fdc65 --- /dev/null +++ b/simulations/publisher_20261108_cassini_convergence_scan.py @@ -0,0 +1,39 @@ +"""Publisher 2026-11-08: where does the bulges-in profiled-a0 Cassini z stop converging? +cf84bb8c says gamma=0.3 does not converge (8-22 sigma) and gamma>=0.489 does; gamma=0.4 is unstated, +and the paper's 11-07 note prints "every gamma from 0.3 to 3". Read-only. Same a0 fit, same qumond_q, same Q2 datum. +Prediction before running: 0.3 spreads by several sigma; 0.489+ by <=0.05; 0.4 unknown. +""" +import sys +import numpy as np +from scipy.optimize import minimize_scalar + +sys.path.insert(0, '/home/dp/ai-workspace/synchronism-site/explorer/scripts') +sys.path.insert(0, '/home/dp/ai-workspace/Synchronism/simulations') +import mond_delta_family_cassini_at_sparc_profiled_a0 as M # noqa: E402 +from sparc_cassini_q2 import TanhLogNu, qumond_q, q2_si # noqa: E402 + +gid, gobs, gbar = M.load() + + +def fit(nu): + f = lambda la: np.sum((np.log10(gobs) - np.log10(gbar * nu(gbar / 10**la)))**2) + return 10**minimize_scalar(f, bounds=(-12, -9), method='bounded', options={'xatol': 1e-9}).x + + +print(f"{'gamma':>6} {'g_ext':>9} {'g_ext/a0':>9} {'z_base':>8} {'min':>8} {'max':>8} {'spread':>7} draws") +for gm in (0.3, 0.35, 0.4, 0.45, 0.489, 0.6, 1.0, 2.0, 3.0): + nu = TanhLogNu(gm) + a0 = fit(nu) + for ge in (2.00e-10, 2.32e-10, 2.48e-10): + zs = [] + for da in (0.999, 1.0, 1.001, 1.003): + for ao, ro in ((128, 512), (192, 768), (256, 1024)): + a = a0 * da + try: + res = qumond_q(nu, ge / a, angular_order=ao, radial_order=ro, log_v_bound=14.0) + zs.append((q2_si(res.q, a) - M.MEAN) / M.SIG) + except Exception as e: + zs.append(float('nan')) + zs = np.array(zs) + print(f"{gm:6.3f} {ge:9.2e} {ge/a0:9.2f} {zs[3]:8.2f} {np.nanmin(zs):8.2f} {np.nanmax(zs):8.2f} " + f"{np.nanmax(zs)-np.nanmin(zs):7.2f} {' '.join(f'{v:.1f}' for v in zs)}", flush=True) diff --git a/simulations/publisher_20261108_cassini_convergence_scan_output.txt b/simulations/publisher_20261108_cassini_convergence_scan_output.txt new file mode 100644 index 000000000..a5f603273 --- /dev/null +++ b/simulations/publisher_20261108_cassini_convergence_scan_output.txt @@ -0,0 +1,29 @@ + gamma g_ext g_ext/a0 z_base min max spread draws + 0.300 2.00e-10 7.92 14.33 11.90 25.93 14.03 17.3 12.3 12.6 14.3 25.9 12.1 12.1 15.7 12.2 15.1 11.9 16.3 + 0.300 2.32e-10 9.18 8.42 8.42 27.99 19.57 28.0 21.8 17.1 8.4 16.8 16.8 15.5 18.0 11.9 16.2 17.1 15.8 + 0.300 2.48e-10 9.82 18.41 16.24 24.13 7.88 21.2 17.3 18.3 18.4 17.8 16.7 21.8 16.2 24.1 17.2 18.3 17.9 + 0.350 2.00e-10 6.17 14.52 9.98 14.98 5.00 14.2 14.2 14.0 14.5 14.8 14.6 10.0 15.0 13.5 14.7 14.5 14.4 + 0.350 2.32e-10 7.15 18.02 13.70 19.07 5.37 17.4 16.3 17.1 18.0 17.5 19.1 13.7 18.0 16.4 17.7 17.4 17.7 + 0.350 2.48e-10 7.65 19.53 17.83 23.60 5.77 18.6 18.9 18.3 19.5 23.6 18.9 20.1 18.4 18.8 23.2 19.0 17.8 + 0.400 2.00e-10 5.02 14.82 14.64 14.83 0.19 14.7 14.7 14.7 14.8 14.7 14.6 14.8 14.8 14.8 14.7 14.7 14.8 + 0.400 2.32e-10 5.83 17.70 17.60 17.89 0.29 17.7 17.9 17.7 17.7 17.6 17.7 17.7 17.7 17.7 17.6 17.7 17.6 + 0.400 2.48e-10 6.23 19.21 19.03 19.59 0.56 19.3 19.0 19.3 19.2 19.2 19.2 19.6 19.2 19.2 19.2 19.3 19.2 + 0.450 2.00e-10 4.22 14.98 14.92 15.01 0.09 15.0 14.9 15.0 15.0 14.9 15.0 14.9 15.0 15.0 15.0 15.0 15.0 + 0.450 2.32e-10 4.90 17.89 17.82 17.94 0.12 17.9 17.8 17.9 17.9 17.9 17.9 17.9 17.9 17.9 17.9 17.9 17.9 + 0.450 2.48e-10 5.24 19.34 19.29 19.34 0.06 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 + 0.489 2.00e-10 3.75 15.10 15.08 15.11 0.04 15.1 15.1 15.1 15.1 15.1 15.1 15.1 15.1 15.1 15.1 15.1 15.1 + 0.489 2.32e-10 4.35 17.95 17.91 17.97 0.06 17.9 18.0 17.9 18.0 17.9 17.9 18.0 18.0 18.0 18.0 18.0 18.0 + 0.489 2.48e-10 4.65 19.87 19.33 19.87 0.54 19.3 19.3 19.3 19.9 19.3 19.3 19.4 19.3 19.4 19.4 19.4 19.4 + 0.600 2.00e-10 2.83 15.30 15.29 15.32 0.03 15.3 15.3 15.3 15.3 15.3 15.3 15.3 15.3 15.3 15.3 15.3 15.3 + 0.600 2.32e-10 3.29 18.00 17.99 18.03 0.04 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 18.0 + 0.600 2.48e-10 3.51 19.29 19.28 19.32 0.04 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 19.3 + 1.000 2.00e-10 1.49 15.37 15.36 15.40 0.04 15.4 15.4 15.4 15.4 15.4 15.4 15.4 15.4 15.4 15.4 15.4 15.4 + 1.000 2.32e-10 1.73 17.53 17.51 17.56 0.05 17.5 17.5 17.5 17.5 17.5 17.5 17.5 17.5 17.5 17.6 17.6 17.6 + 1.000 2.48e-10 1.85 18.48 18.47 18.53 0.06 18.5 18.5 18.5 18.5 18.5 18.5 18.5 18.5 18.5 18.5 18.5 18.5 + 2.000 2.00e-10 0.67 14.89 14.88 14.93 0.05 14.9 14.9 14.9 14.9 14.9 14.9 14.9 14.9 14.9 14.9 14.9 14.9 + 2.000 2.32e-10 0.78 16.26 16.24 16.31 0.07 16.2 16.2 16.2 16.3 16.3 16.3 16.3 16.3 16.3 16.3 16.3 16.3 + 2.000 2.48e-10 0.84 16.76 16.74 16.82 0.08 16.7 16.7 16.7 16.8 16.8 16.8 16.8 16.8 16.8 16.8 16.8 16.8 + 3.000 2.00e-10 0.43 14.55 14.54 14.60 0.06 14.5 14.5 14.5 14.6 14.6 14.6 14.6 14.6 14.6 14.6 14.6 14.6 + 3.000 2.32e-10 0.50 15.54 15.52 15.60 0.08 15.5 15.5 15.5 15.5 15.5 15.5 15.6 15.6 15.6 15.6 15.6 15.6 + 3.000 2.48e-10 0.54 15.82 15.80 15.89 0.09 15.8 15.8 15.8 15.8 15.8 15.8 15.8 15.8 15.8 15.9 15.9 15.9 +rc=0 diff --git a/simulations/publisher_20261109_test08_upper_limit.py b/simulations/publisher_20261109_test08_upper_limit.py new file mode 100644 index 000000000..82d7e76f7 --- /dev/null +++ b/simulations/publisher_20261109_test08_upper_limit.py @@ -0,0 +1,59 @@ +#!/usr/bin/env python3 +"""Publisher 2026-11-09: how far under the bar is TEST-08, as an interval rather than a point? + +The paper's count list prints "r² = 0.0001 against a registered > 0.20". PREDICTIONS.md 2026-11-08 (iv) says the +95% upper limit is r² ≈ 0.03. This recomputes that from the tracked per-galaxy table rather than relaying it. + +Reads simulations/test08_per_galaxy_results.json (written by test08_sparc_environment_rar.py). Writes nothing. +Rebuilds the pre-declared primary statistic the same way that script does: per-galaxy RAR offset against +log10(1 + N_cyl[2 Mpc, 500 km/s]) with log D regressed out. + +Expected before running: r = +0.012, N = 141, Fisher 95% upper limit on r² near 0.03, bootstrap within ~0.01 of that. +""" +import json +import pathlib + +import numpy as np + +HERE = pathlib.Path(__file__).resolve().parent +T = json.load(open(HERE / "test08_per_galaxy_results.json")) +names = sorted(T) +off = np.array([T[n]["offset_dex"] for n in names]) +cyl = np.log10(1 + np.array([T[n]["cyl2"] for n in names], float)) +logD = np.log10(np.array([T[n]["D"] for n in names], float)) + + +def residual(y, x): + A = np.stack([np.ones_like(x), x], axis=1) + return y - A @ np.linalg.lstsq(A, y, rcond=None)[0] + + +env = residual(cyl, logD) +r = float(np.corrcoef(off, env)[0, 1]) +N = len(names) +print(f"N = {N}, r = {r:+.4f}, r^2 = {r**2:.5f} (identity: registered run printed r=+0.012, N=141, r^2=0.0001)") + +# Fisher z interval. The largest |r| in the interval sets the upper limit on r^2. +z, se = np.arctanh(r), 1 / np.sqrt(N - 3) +lo, hi = np.tanh(z - 1.96 * se), np.tanh(z + 1.96 * se) +fisher_ul = max(lo**2, hi**2) +print(f"Fisher 95% r in [{lo:+.3f}, {hi:+.3f}]; upper limit on r^2 = {fisher_ul:.4f}") + +# Bootstrap over galaxies, refitting the distance correction on each draw. Three seeds: a single seed's +# agreement with itself says nothing about whether the quantile is stable. +for seed in (1, 2, 3): + rng = np.random.default_rng(seed) + rs = np.empty(20000) + for k in range(rs.size): + i = rng.integers(0, N, N) + rs[k] = np.corrcoef(off[i], residual(cyl[i], logD[i]))[0, 1] + print(f"bootstrap seed {seed}: 95% r in [{np.quantile(rs, .025):+.3f}, {np.quantile(rs, .975):+.3f}]; " + f"97.5% quantile of r^2 = {np.quantile(rs**2, .975):.4f}; 95% quantile of r^2 = {np.quantile(rs**2, .95):.4f}") + +for label, bar in (("kill bar", 0.09), ("registered claim", 0.20)): + print(f"{label} {bar}: {bar / fisher_ul:.1f}x the Fisher upper limit; {bar / r**2:.0f}x the point estimate") + +print(f"\nper-galaxy offset sd = {off.std():.4f} dex (the variance r^2 is a share of)") +print("the registered sentence reads 'environment explains >20% of RAR scatter' and names no denominator;") +print("the pooled point scatter in the registered run is 0.133 dex, a different quantity from the per-galaxy offset sd.") +print(f"offset variance / pooled point variance = {off.var() / 0.133**2:.3f}") diff --git a/simulations/publisher_20261109_test08_upper_limit_output.txt b/simulations/publisher_20261109_test08_upper_limit_output.txt new file mode 100644 index 000000000..898517904 --- /dev/null +++ b/simulations/publisher_20261109_test08_upper_limit_output.txt @@ -0,0 +1,12 @@ +N = 141, r = +0.0118, r^2 = 0.00014 (identity: registered run printed r=+0.012, N=141, r^2=0.0001) +Fisher 95% r in [-0.154, +0.177]; upper limit on r^2 = 0.0312 +bootstrap seed 1: 95% r in [-0.164, +0.179]; 97.5% quantile of r^2 = 0.0379; 95% quantile of r^2 = 0.0299 +bootstrap seed 2: 95% r in [-0.161, +0.182]; 97.5% quantile of r^2 = 0.0392; 95% quantile of r^2 = 0.0299 +bootstrap seed 3: 95% r in [-0.162, +0.181]; 97.5% quantile of r^2 = 0.0393; 95% quantile of r^2 = 0.0299 +kill bar 0.09: 2.9x the Fisher upper limit; 649x the point estimate +registered claim 0.2: 6.4x the Fisher upper limit; 1442x the point estimate + +per-galaxy offset sd = 0.1244 dex (the variance r^2 is a share of) +the registered sentence reads 'environment explains >20% of RAR scatter' and names no denominator; +the pooled point scatter in the registered run is 0.133 dex, a different quantity from the per-galaxy offset sd. +offset variance / pooled point variance = 0.875 diff --git a/simulations/publisher_20261111_lensing_ceiling_census.py b/simulations/publisher_20261111_lensing_ceiling_census.py new file mode 100644 index 000000000..3f389f84b --- /dev/null +++ b/simulations/publisher_20261111_lensing_ceiling_census.py @@ -0,0 +1,33 @@ +#!/usr/bin/env python3 +"""Publisher arithmetic, 2026-11-11 (cbp-claude, interactive maintenance pass). + +Whitepaper 5.15, "The ceiling also dies independently" (Brouwer et al. 2021 lensing RAR). +The paragraph prints deficits 6x / 13x / 28x at g_bar = 1e-14 and says a factor-2 baryon +correction "moves nu by sqrt(2)". Two things are recomputed here from the paragraph's own inputs: + + * the boost a ceiling must supply is g_obs/g_bar. Lensing fixes g_obs, so raising the + baryons by f lowers the requirement by f (not sqrt(f)); + * Brouwer et al. Sec. 1 + footnote 3: the outer edge is g_bar ~ 1e-15 counting stars and + cold gas only; hot gas, dust and missing baryons could raise it to ~1e-14 (f ~ 10). + +Assumption, stated: g_obs is taken on MOND's simple-nu curve at the fiducial g_bar (the +paper's "agrees well with the MG predictions"); no Brouwer data point is read here. +Inputs from the whitepaper: eps0 = 0.119 (M/1e10)^0.325, B_max = 1/eps0, a0 = 1.2e-10. +No file is read or written. +""" +import math +a0 = 1.2e-10 +def nu(g): # simple interpolating function + y = g / a0 + return 0.5 + math.sqrt(0.25 + 1.0 / y) +ceil = {m: 1.0 / (0.119 * (10 ** (m - 10)) ** 0.325) for m in (9, 10, 11)} +print("ceilings B_max = 1/eps0:", ", ".join(f"1e{m}: {c:.1f}" for m, c in ceil.items())) +print("framework cap 1/Omega_m = %.3f; most permissive cap quoted by the ledger = 6.39" % (1 / 0.315)) +for g in (1e-14, 1e-15): + for f in (1, 2, 5, 10): + need = nu(g) / f + row = " ".join(f"1e{m}: {'pass' if need <= c else f'{need / c:4.1f}x short'}" for m, c in ceil.items()) + print(f"fiducial g_bar = {g:.0e}, baryons x{f:<2}: required boost = {need:6.1f} | {row} | 6.39: " + f"{'pass' if need <= 6.39 else f'{need / 6.39:4.1f}x short'}") +print("check on the printed sentence: nu(2e-14)/nu(1e-14) = %.3f (= 1/sqrt 2: that is MOND's own nu moving," + " not the required boost, which moves by 1/2)" % (nu(2e-14) / nu(1e-14))) diff --git a/simulations/publisher_20261111_lensing_ceiling_census_output.txt b/simulations/publisher_20261111_lensing_ceiling_census_output.txt new file mode 100644 index 000000000..695261df7 --- /dev/null +++ b/simulations/publisher_20261111_lensing_ceiling_census_output.txt @@ -0,0 +1,11 @@ +ceilings B_max = 1/eps0: 1e9: 17.8, 1e10: 8.4, 1e11: 4.0 +framework cap 1/Omega_m = 3.175; most permissive cap quoted by the ledger = 6.39 +fiducial g_bar = 1e-14, baryons x1 : required boost = 110.0 | 1e9: 6.2x short 1e10: 13.1x short 1e11: 27.7x short | 6.39: 17.2x short +fiducial g_bar = 1e-14, baryons x2 : required boost = 55.0 | 1e9: 3.1x short 1e10: 6.5x short 1e11: 13.8x short | 6.39: 8.6x short +fiducial g_bar = 1e-14, baryons x5 : required boost = 22.0 | 1e9: 1.2x short 1e10: 2.6x short 1e11: 5.5x short | 6.39: 3.4x short +fiducial g_bar = 1e-14, baryons x10: required boost = 11.0 | 1e9: pass 1e10: 1.3x short 1e11: 2.8x short | 6.39: 1.7x short +fiducial g_bar = 1e-15, baryons x1 : required boost = 346.9 | 1e9: 19.5x short 1e10: 41.3x short 1e11: 87.2x short | 6.39: 54.3x short +fiducial g_bar = 1e-15, baryons x2 : required boost = 173.5 | 1e9: 9.8x short 1e10: 20.6x short 1e11: 43.6x short | 6.39: 27.1x short +fiducial g_bar = 1e-15, baryons x5 : required boost = 69.4 | 1e9: 3.9x short 1e10: 8.3x short 1e11: 17.4x short | 6.39: 10.9x short +fiducial g_bar = 1e-15, baryons x10: required boost = 34.7 | 1e9: 2.0x short 1e10: 4.1x short 1e11: 8.7x short | 6.39: 5.4x short +check on the printed sentence: nu(2e-14)/nu(1e-14) = 0.708 (= 1/sqrt 2: that is MOND's own nu moving, not the required boost, which moves by 1/2) diff --git a/simulations/publisher_20261111_tc_gamma_control.py b/simulations/publisher_20261111_tc_gamma_control.py new file mode 100644 index 000000000..0e41a0f97 --- /dev/null +++ b/simulations/publisher_20261111_tc_gamma_control.py @@ -0,0 +1,70 @@ +#!/usr/bin/env python3 +"""Publisher control, 2026-11-11 (cbp-claude, interactive maintenance pass). + +Question: the whitepaper's "Superconductivity | Tc ∝ exp(-γ/λ) | r = 0.948 | #62" row +(chemistry.md top-validated table; Appendix B.5). Where does 0.948 come from, and is the +γ in it independent of Tc? + +Source: simulations/chemistry/superconductivity_coherence.py (Chemistry Session #62, +commit e4c903f5, 2026-01-17). This script executes that file's PARTS 1-4 unchanged (text up +to the "PART 5" banner; no file is written by that part), then recomputes the PART 5 +correlation and three controls. Read-only. + +Usage: python3 publisher_20261111_tc_gamma_control.py +""" +import sys, io, contextlib +import numpy as np +from scipy import stats + +src = open(sys.argv[1], encoding="utf-8").read() +cut = src.index('print("PART 5: Tc vs γ CORRELATION")') +cut = src.rfind("\n# ====", 0, cut) +ns = {} +with contextlib.redirect_stdout(io.StringIO()): + exec(compile(src[:cut], "superconductivity_coherence[:PART5]", "exec"), ns) + +Tcs, Deltas, cats = ns["Tcs"], ns["Deltas"], ns["categories"] +gammas = ns["gammas"] +kB, bcs = ns["k_B_meV"], ns["BCS_ratio_theory"] +n = len(Tcs) +ratio = 2 * Deltas / (kB * Tcs) + +def pr(label, x, y): + r, p = stats.pearsonr(x, y) + rs, ps = stats.spearmanr(x, y) + print(f"{label:<58} Pearson r = {r:+.3f} (p = {p:.1e}) Spearman = {rs:+.3f}") + return r + +print(f"N = {n} superconductors; BCS_ratio_theory = {bcs}; categories: {sorted(set(cats))}") +print() +print("IDENTITY (the source's PART 5 number)") +r0 = pr("Tc vs 1/gamma, gamma as the source estimates it", 1 / gammas, Tcs) +print() +print("WHAT gamma IS: gamma = 2 / (ratio/BCS) x category factor, ratio = 2*Delta/(k_B*Tc)") +g_nocat = 2.0 / (ratio / bcs) +fac = gammas / g_nocat +for c in sorted(set(cats)): + m = np.array([x == c for x in cats]) + print(f" {c:<14} n = {m.sum():>2} factor on gamma = {np.unique(np.round(fac[m],3))} mean Tc = {Tcs[m].mean():7.1f} K") +print() +print("CONTROLS") +r1 = pr("C1 Tc vs 1/gamma, category factors removed (= ratio/2BCS)", 1 / g_nocat, Tcs) +r2 = pr("C2 Tc vs category factor alone (1/factor)", 1 / fac, Tcs) +rng = np.random.default_rng(20261111) +rs = [] +for _ in range(20000): + perm = rng.permutation(n) + # shuffle the gap ratios among materials, keep each material's category factor + rs.append(stats.pearsonr(1 / (2.0 / (ratio[perm] / bcs) * fac), Tcs)[0]) +rs = np.array(rs) +print(f"C3 gap ratios shuffled among materials, category factors kept (20,000 draws, seed 20261111):") +print(f" median r = {np.median(rs):+.3f}, 2.5-97.5% = [{np.percentile(rs,2.5):+.3f}, {np.percentile(rs,97.5):+.3f}], " + f"fraction >= source r: {np.mean(rs >= r0):.4f}") +print() +print("WITHIN CATEGORY (factor constant, so this is Tc vs gap ratio)") +for c in sorted(set(cats)): + m = np.array([x == c for x in cats]) + if m.sum() >= 4: + pr(f" {c} (n = {m.sum()})", 1 / gammas[m], Tcs[m]) + else: + print(f" {c} (n = {m.sum()}): too few") diff --git a/simulations/publisher_20261111_tc_gamma_control_output.txt b/simulations/publisher_20261111_tc_gamma_control_output.txt new file mode 100644 index 000000000..561e57584 --- /dev/null +++ b/simulations/publisher_20261111_tc_gamma_control_output.txt @@ -0,0 +1,24 @@ +N = 28 superconductors; BCS_ratio_theory = 3.52; categories: ['High-P', 'High-Tc', 'Iron SC', 'Type I', 'Type II'] + +IDENTITY (the source's PART 5 number) +Tc vs 1/gamma, gamma as the source estimates it Pearson r = +0.948 (p = 1.9e-14) Spearman = +0.882 + +WHAT gamma IS: gamma = 2 / (ratio/BCS) x category factor, ratio = 2*Delta/(k_B*Tc) + High-P n = 2 factor on gamma = [0.7] mean Tc = 226.5 K + High-Tc n = 6 factor on gamma = [0.8] mean Tc = 118.2 K + Iron SC n = 5 factor on gamma = [1.] mean Tc = 38.4 K + Type I n = 4 factor on gamma = [1.] mean Tc = 3.9 K + Type II n = 11 factor on gamma = [1.] mean Tc = 19.0 K + +CONTROLS +C1 Tc vs 1/gamma, category factors removed (= ratio/2BCS) Pearson r = +0.882 (p = 5.8e-10) Spearman = +0.866 +C2 Tc vs category factor alone (1/factor) Pearson r = +0.944 (p = 5.5e-14) Spearman = +0.791 +C3 gap ratios shuffled among materials, category factors kept (20,000 draws, seed 20261111): + median r = +0.422, 2.5-97.5% = [+0.090, +0.710], fraction >= source r: 0.0000 + +WITHIN CATEGORY (factor constant, so this is Tc vs gap ratio) + High-P (n = 2): too few + High-Tc (n = 6) Pearson r = +0.571 (p = 2.4e-01) Spearman = +0.657 + Iron SC (n = 5) Pearson r = -0.161 (p = 8.0e-01) Spearman = -0.300 + Type I (n = 4) Pearson r = +0.926 (p = 7.4e-02) Spearman = +1.000 + Type II (n = 11) Pearson r = +0.731 (p = 1.1e-02) Spearman = +0.736 diff --git a/simulations/publisher_20261116_dsph_cap_estimator_sensitivity.py b/simulations/publisher_20261116_dsph_cap_estimator_sensitivity.py new file mode 100644 index 000000000..438fa21db --- /dev/null +++ b/simulations/publisher_20261116_dsph_cap_estimator_sensitivity.py @@ -0,0 +1,77 @@ +#!/usr/bin/env python3 +"""Publisher 2026-11-16: does the dSph exclusion of the Omega_m/Omega_b = 6.39 boost cap +survive the mass-estimator convention? + +Source: synchronism-site/maintainer/scripts/cap6p39_vs_dsph.py (PREREG site c7d729c), inputs +copied unchanged from explorer/findings/scripts/pressure_supported_boost_ceiling.py (2026-08-23). +That script uses sigma_N^2 = G*(M/2)/(4*r_half) with r_half labelled 3D. Wolf et al. 2010 +(MNRAS 406, 1220, eq. 2) give M_1/2 = 3**r_1/2/G for the 3D half-light radius; +the factor 4 belongs with the projected R_e (r_1/2 ~ 4/3 R_e). With the 3D radius, k = 4 +understates sigma_N by sqrt(3/4), so it overstates the required boost by 4/3 and the z by a +little. This script reruns the 6.39 test at k = 4 (identity control) and k = 3, and adds a +stress test with L_V raised 30% (no catalogue value was read; this only states a margin). +""" +import math + +G = 4.300917270e-6 # kpc (km/s)^2 / Msun +SYS = { # name: L_V, r_half_3D kpc, sigma_obs, err, ML_fid, ML_hi (2026-08-23 table) + "Crater II": (1.6e5, 1.42, 2.7, 0.3, 2.0, 4.0), + "Draco": (2.2e5, 0.29, 9.1, 1.2, 2.0, 4.0), +} + + +def _phi(x): + return 0.5 * (1 + math.erf(x / math.sqrt(2))) + + +# six-system one-sided equivalent of 3 sigma, computed as the source script does (prints 3.51; +# the unrounded value matters for the identity control's 14.7) +_pS = 1 - (1 - (1 - _phi(3.0))) ** (1 / 6) +_lo, _hi = 0.0, 10.0 +for _ in range(80): + _m = (_lo + _hi) / 2 + (_lo, _hi) = (_m, _hi) if 1 - _phi(_m) > _pS else (_lo, _m) +Z_SIDAK = (_lo + _hi) / 2 + + +def sig_n(M, rh, k): + return math.sqrt(G * (M / 2.0) / (k * rh)) + + +def z(name, cap, ml, k, lscale=1.0): + L, rh, s, e, _, _ = SYS[name] + return (s - sig_n(L * lscale * ml, rh, k) * math.sqrt(cap)) / e + + +def largest_cap(name, ml, k, zt, lscale=1.0): + L, rh, s, e, _, _ = SYS[name] + return ((s - zt * e) / sig_n(L * lscale * ml, rh, k)) ** 2 + + +print("IDENTITY CONTROL (k = 4 must reproduce cap6p39_vs_dsph_output.txt)") +ref = {("Draco", "fid"): 5.7, ("Draco", "hi"): 4.9, ("Crater II", "fid"): 6.1, ("Crater II", "hi"): 4.9} +ok = True +for (n, w), r in ref.items(): + ml = SYS[n][4] if w == "fid" else SYS[n][5] + v = round(z(n, 6.39, ml, 4), 1) + ok &= abs(v - r) < 1e-9 + print(f" {n:<10} M/L {w:<3} z(6.39) = {v:4.1f} ref {r:4.1f}") +for n, r in (("Draco", 14.7), ("Crater II", 11.2)): + v = round(largest_cap(n, SYS[n][5], 4, Z_SIDAK), 1) + ok &= abs(v - r) < 1e-9 + print(f" {n:<10} largest cap excluded at z = {Z_SIDAK:.2f} (M/L hi) = {v:4.1f} ref {r:4.1f}") +print(f" identity control: {'PASS' if ok else 'FAIL'}\n") + +print(f"{'case':<34}{'system':<11}{'z fid':>7}{'z M/L=4':>9}{'max cap (trials, M/L=4)':>26}") +for label, k, ls in (("k = 4 (as run)", 4, 1.0), + ("k = 3 (Wolf+2010, 3D r_1/2)", 3, 1.0), + ("k = 3 and L_V x 1.3 (stress)", 3, 1.3)): + for n in SYS: + _, _, _, _, mf, mh = SYS[n] + print(f"{label:<34}{n:<11}{z(n, 6.39, mf, k, ls):>7.2f}{z(n, 6.39, mh, k, ls):>9.2f}" + f"{largest_cap(n, mh, k, Z_SIDAK, ls):>26.1f}") +print(f"\nSidak threshold z = {Z_SIDAK:.4f}; 6.39 excluded band-robustly after trials iff z(M/L=4) > that.") +print("B_req at fiducial M/L, k = 3 (vs the 08-23 table's 101.5 / 60.2 at k = 4):") +for n in SYS: + L, rh, s, e, mf, _ = SYS[n] + print(f" {n:<10} {(s / sig_n(L * mf, rh, 3)) ** 2:6.1f}") diff --git a/simulations/publisher_20261116_dsph_cap_estimator_sensitivity_output.txt b/simulations/publisher_20261116_dsph_cap_estimator_sensitivity_output.txt new file mode 100644 index 000000000..fbc49bdc2 --- /dev/null +++ b/simulations/publisher_20261116_dsph_cap_estimator_sensitivity_output.txt @@ -0,0 +1,21 @@ +IDENTITY CONTROL (k = 4 must reproduce cap6p39_vs_dsph_output.txt) + Draco M/L fid z(6.39) = 5.7 ref 5.7 + Draco M/L hi z(6.39) = 4.9 ref 4.9 + Crater II M/L fid z(6.39) = 6.1 ref 6.1 + Crater II M/L hi z(6.39) = 4.9 ref 4.9 + Draco largest cap excluded at z = 3.51 (M/L hi) = 14.7 ref 14.7 + Crater II largest cap excluded at z = 3.51 (M/L hi) = 11.2 ref 11.2 + identity control: PASS + +case system z fid z M/L=4 max cap (trials, M/L=4) +k = 4 (as run) Crater II 6.07 4.85 11.2 +k = 4 (as run) Draco 5.68 4.89 14.7 +k = 3 (Wolf+2010, 3D r_1/2) Crater II 5.61 4.21 8.4 +k = 3 (Wolf+2010, 3D r_1/2) Draco 5.39 4.48 11.0 +k = 3 and L_V x 1.3 (stress) Crater II 5.14 3.54 6.5 +k = 3 and L_V x 1.3 (stress) Draco 5.08 4.04 8.5 + +Sidak threshold z = 3.5087; 6.39 excluded band-robustly after trials iff z(M/L=4) > that. +B_req at fiducial M/L, k = 3 (vs the 08-23 table's 101.5 / 60.2 at k = 4): + Crater II 45.1 + Draco 76.1 diff --git a/simulations/publisher_20261117_bcs_ratio_and_gamma_check.py b/simulations/publisher_20261117_bcs_ratio_and_gamma_check.py new file mode 100644 index 000000000..46e9b7c5c --- /dev/null +++ b/simulations/publisher_20261117_bcs_ratio_and_gamma_check.py @@ -0,0 +1,48 @@ +"""Publisher check 2026-11-17: the paper's "BCS ratio derived: 2*sqrt(pi) = 3.54, observed 3.52 (<1% error)". + +Printed at superconductivity.md (Derived Gap Ratio), conclusion.md (BCS ratio bullet, with a check mark) +and appendix_b_chemistry.md L88. Source: Research/Chemistry/Session1_Superconductivity_Coherence.md §3.3. + +What this script does (no data, textbook BCS only): + 1. Solves the BCS gap equation (weak coupling, cutoff hbar*omega_D) for Delta(0) and T_c at several + couplings lambda, and prints 2*Delta(0)/(k_B*T_c). Identity control: the closed form + 2*pi*exp(-gamma_Euler) = 3.5278. + 2. Compares that with 2*sqrt(pi) = 3.5449. + 3. Prints the archive's own "Synchronism" ratio formula from simulations/chemistry/superconductor_coherence.py + (2*pi / tanh(gamma*ln 2), and the plotting line's 2*pi / tanh(gamma*ln 3)) at gamma = 2. + +Units: k_B = 1, hbar*omega_D = 1. Run: python3 -I +""" +import numpy as np +from scipy.integrate import quad +from scipy.optimize import brentq + +EULER = 0.5772156649015329 + + +def gap_eq_T0(delta, lam): + # 1 = lam * int_0^1 de / sqrt(e^2 + d^2) -> asinh(1/d) + return lam * np.arcsinh(1.0 / delta) - 1.0 + + +def gap_eq_Tc(T, lam): + # 1 = lam * int_0^1 tanh(e / 2T) / e de + val, _ = quad(lambda e: np.tanh(e / (2 * T)) / e if e > 0 else 1 / (2 * T), 0, 1, limit=400) + return lam * val - 1.0 + + +closed = 2 * np.pi * np.exp(-EULER) +print(f"closed form 2*pi*exp(-gamma_E) = {closed:.4f}") +print(f"2*sqrt(pi) = {2*np.sqrt(np.pi):.4f}") +print(f"difference = {100*(2*np.sqrt(np.pi)/closed-1):.2f}%") +print() +print("numerical BCS gap equation (weak coupling):") +for lam in [0.15, 0.2, 0.25, 0.3]: + d0 = brentq(gap_eq_T0, 1e-12, 10, args=(lam,)) + tc = brentq(gap_eq_Tc, 1e-8, 5, args=(lam,), xtol=1e-14) + print(f" lambda={lam:.2f}: Delta0={d0:.4e} Tc={tc:.4e} 2*Delta0/Tc={2*d0/tc:.4f}") +print() +print("archive's own 'Synchronism' ratio formulas at gamma = 2:") +for g in [2.0]: + print(f" 2*pi/tanh(g*ln2) = {2*np.pi/np.tanh(g*np.log(2)):.3f} (QUANTITATIVE TEST block)") + print(f" 2*pi/tanh(g*ln3) = {2*np.pi/np.tanh(g*np.log(3)):.3f} (plot block, assigned but not drawn)") diff --git a/simulations/publisher_20261117_bcs_ratio_and_gamma_check_output.txt b/simulations/publisher_20261117_bcs_ratio_and_gamma_check_output.txt new file mode 100644 index 000000000..9aa4ee366 --- /dev/null +++ b/simulations/publisher_20261117_bcs_ratio_and_gamma_check_output.txt @@ -0,0 +1,13 @@ +closed form 2*pi*exp(-gamma_E) = 3.5278 +2*sqrt(pi) = 3.5449 +difference = 0.49% + +numerical BCS gap equation (weak coupling): + lambda=0.15: Delta0=2.5453e-03 Tc=1.4430e-03 2*Delta0/Tc=3.5278 + lambda=0.20: Delta0=1.3477e-02 Tc=7.6399e-03 2*Delta0/Tc=3.5279 + lambda=0.25: Delta0=3.6644e-02 Tc=2.0767e-02 2*Delta0/Tc=3.5289 + lambda=0.30: Delta0=7.1439e-02 Tc=4.0450e-02 2*Delta0/Tc=3.5322 + +archive's own 'Synchronism' ratio formulas at gamma = 2: + 2*pi/tanh(g*ln2) = 7.121 (QUANTITATIVE TEST block) + 2*pi/tanh(g*ln3) = 6.440 (plot block, assigned but not drawn) diff --git a/simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py b/simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py new file mode 100644 index 000000000..6328ca7ca --- /dev/null +++ b/simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py @@ -0,0 +1,74 @@ +"""Publisher 2026-11-18: is the open-questions "virial cap" (x <= ~0.02 for every bound +system) a property of the density law, or of a length convention? + +Checks back-annotation 5d2d684c (ii): "the virial-ratio 'knee unreachable in every sector' +argument holds only for A*V^2 with the same length in A and in the smoothing of rho". + +Part 1 (identity controls, algebra): + two-sided convention rho_bar(l) = 3 V_c(l)^2 / (4 pi G l^2), rho_crit = A(l) V^2, + A(l) = 4 pi / (beta^2 G l^2) -> x = (3/16 pi^2) beta^2 (V_c/V)^2 (paper L45: 0.019 at beta=1) + primary form (S53 L58-59, no 4 pi): A = 1/(G alpha^2 l^2) -> x = 3/(4 pi) alpha^2 (V_c/V)^2 (paper: 0.24) + Jeans A at l = 8 kpc, 4 pi form: 4.57e-5 (paper conclusion L78 / exec L190: 4.56e-5) +Part 2 (data): S53's executed law rho_crit = 0.028 V^0.5 (V in km/s, M_sun/pc^3), i.e. the same + V^2/(G alpha^2 R_half^2) with R_half = R0 V^0.75, against the midplane density the archive's + fits use, rho = Sigma / 2h, h = 0.3 kpc. Sigma = 0.5 SBdisk + 0.7 SBbul (stars only; gas + omitted, so Sigma and every x here are LOWER bounds). Count discs whose innermost point has x > 1. + V = mean Vobs over the last 3 points (a V_flat proxy, not Lelli's Vflat column). +""" +import math, sys + +G = 4.30091e-3 # pc (km/s)^2 / M_sun + +# ---- Part 1: identity controls +x2 = 3 / (16 * math.pi**2) +xp = 3 / (4 * math.pi) +AJ = 4 * math.pi / (G * (8000.0) ** 2) +print(f"two-sided 4pi-form cap at beta=1: {x2:.4f} (paper 0.019)") +print(f"two-sided primary-form cap at alpha=1: {xp:.4f} (paper 0.24)") +print(f"Jeans A at 8 kpc, 4pi form: {AJ:.3e} (paper 4.56e-5)") +ok = abs(x2 - 0.019) < 5e-4 and abs(xp - 0.24) < 5e-3 and abs(AJ - 4.56e-5) / 4.56e-5 < 0.01 +print("IDENTITY CONTROLS", "PASS" if ok else "FAIL") +if not ok: + sys.exit(1) + +# S53 coefficient check: A = 1/(G alpha^2 R0^2), R0 = 0.088 kpc/(km/s)^0.75, alpha = 1.1 +A53 = 1 / (G * 1.1**2 * (88.0) ** 2) +print(f"S53 A = 1/(G a^2 R0^2) at alpha=1.1, R0=0.088 kpc: {A53:.4f} (S53 prints ~0.028)") + +# ---- Part 2: SPARC +path = sys.argv[1] +gal = {} +for line in open(path, encoding="latin-1"): + p = line.split() + if len(p) != 10: + continue + try: + vals = list(map(float, p[1:])) + except ValueError: + continue + gal.setdefault(p[0], []).append(vals) + +h = 300.0 # pc +n = cross = 0 +xs = [] +for g, rows in gal.items(): + rows.sort(key=lambda r: r[1]) + if len(rows) < 4: + continue + V = sum(r[2] for r in rows[-3:]) / 3 + r0 = rows[0] + sigma = 0.5 * r0[7] + 0.7 * r0[8] + rho = sigma / (2 * h) + rc = 0.028 * V**0.5 + x = rho / rc + xs.append((x, g, V)) + n += 1 + cross += x > 1 +xs.sort() +print(f"galaxies: {n}; innermost-point x > 1 under 0.028 V^0.5 with rho = Sigma_*/2h: {cross}") +print("median x:", f"{xs[len(xs)//2][0]:.3f}", " max:", f"{xs[-1][0]:.1f} ({xs[-1][1]})") +for name in ("NGC2403", "NGC3198", "NGC7331", "DDO154", "UGC00128"): + for x, g, V in xs: + if g == name: + print(f" {g}: V={V:.0f} km/s x_inner={x:.2f}") +# the same discs under the two-sided virial convention can never exceed ~0.24 (primary) at V_c<=V diff --git a/simulations/publisher_20261120_dsph_walker_aperture.py b/simulations/publisher_20261120_dsph_walker_aperture.py new file mode 100644 index 000000000..c7b9dbece --- /dev/null +++ b/simulations/publisher_20261120_dsph_walker_aperture.py @@ -0,0 +1,71 @@ +"""Publisher 2026-11-20: the dSph 6.39-cap exclusion under Walker+2009, with the light matched to Walker's aperture. + +Source: synchronism-site/maintainer/scripts/cap6p39_vs_dsph_estimator.py (site maintainer 2026-11-19), routed as +PREDICTIONS.md back-annotation b26f6fa9 (i): "Under Walker+2009's smaller estimator Crater II drops below threshold +(2.9 sigma) and Draco passes it only just (3.7 sigma)". + +That script writes Walker as M(R_e) = 2.5 sigma^2 R_e / G with R_e = (3/4) r_3D and then sets M(R_e) equal to the +half-LIGHT mass L*(M/L)/2, i.e. sigma_N^2 = G M_half / (1.875 r_3D). Walker+2009's M is the mass inside a SPHERE of +radius R_e (the projected half-light radius), and that sphere holds less than half the light: for a Plummer profile +(projected half-light radius a), the 3D fraction inside r is r^3/(r^2+a^2)^1.5, which is 2^-1.5 = 0.354 at r = a, and +r_1/2,3D = a / sqrt(2^(2/3) - 1) = 1.305 a. So the baryonic mass Walker's estimator should be fed is 0.354 L(M/L), +not 0.5 L(M/L), and R_e = r_3D / 1.305. + +Same z, threshold and inputs as the source. z = (sigma_obs - sigma_N sqrt(B)) / err; Sidak 6-trial threshold 3.5087. +Identity control: the source's k = 4 / 3 / 1.875 rows must reproduce its printed output. +Limits: Plummer assumed for the aperture fraction (exponential/King profiles move it by a few per cent); quoted sigma +errors only; equilibrium estimator. +""" +import math + +G = 4.300917270e-6 # kpc (km/s)^2 / Msun +ZS = 3.5087 +SYS = [("Crater II", 1.6e5, 1.42, 2.7, 0.3), ("Draco", 2.2e5, 0.29, 9.1, 1.2)] +REF = { # source output, (system, k): (z at M/L=2, z at M/L=4, escape M/L at 3.51) + ("Crater II", 4.0): (6.1, 4.9, 7.0), ("Crater II", 3.0): (5.6, 4.2, 5.3), ("Crater II", 1.875): (4.7, 2.9, 3.3), + ("Draco", 4.0): (5.7, 4.9, 9.2), ("Draco", 3.0): (5.4, 4.5, 6.9), ("Draco", 1.875): (4.8, 3.7, 4.3), +} + + +def sigma_n(L, ml, r3d, light_frac, denom_r): + """sigma_N^2 = G * (light_frac * L * ml) / denom_r.""" + return math.sqrt(G * light_frac * L * ml / denom_r) + + +def row(L, r3d, s, e, light_frac, denom_coeff): + denom = denom_coeff * r3d + z = [(s - sigma_n(L, ml, r3d, light_frac, denom) * math.sqrt(6.39)) / e for ml in (2.0, 4.0)] + s1 = sigma_n(L, 1.0, r3d, light_frac, denom) + ml_esc = ((s - ZS * e) / math.sqrt(6.39) / s1) ** 2 + ml_esc3 = ((s - 3.0 * e) / math.sqrt(6.39) / s1) ** 2 + bx = ((s - ZS * e) / sigma_n(L, 4.0, r3d, light_frac, denom)) ** 2 + infl = (s - ZS * e) / (sigma_n(L, 4.0, r3d, light_frac, denom) * math.sqrt(6.39)) + return z, ml_esc, ml_esc3, bx, infl + + +print("IDENTITY CONTROL (half-light mass, sigma_N^2 = G M_half / (k r3D), as the source)") +ok = True +for name, L, r, s, e in SYS: + for k in (4.0, 3.0, 1.875): + z, esc, _, _, _ = row(L, r, s, e, 0.5, k) + ref = REF[(name, k)] + good = (round(z[0], 1), round(z[1], 1), round(esc, 1)) == ref + ok &= good + print(f" {name:<10} k={k:<5} z(2)={z[0]:4.1f} z(4)={z[1]:4.1f} esc={esc:4.1f} ref {ref} {'ok' if good else 'MISMATCH'}") +print(f" identity control: {'PASS' if ok else 'FAIL'}\n") + +frac_re = 2 ** -1.5 # Plummer light fraction inside a sphere of radius R_e +re_over_r3d = 1 / 1.0 * math.sqrt(2 ** (2 / 3) - 1) # R_e / r_1/2,3D +print(f"Plummer: light inside sphere r = R_e = {frac_re:.4f}; R_e / r_3D = {re_over_r3d:.4f}") +k_eff = 2.5 * re_over_r3d * 0.5 / frac_re +print(f"Walker with its own aperture == half-light form with k_eff = {k_eff:.3f} (source used 1.875; Wolf 3)\n") + +print(f"{'case':<44}{'system':<11}{'z M/L=2':>8}{'z M/L=4':>8}{'esc@3.51':>9}{'esc@3':>7}{'Bexcl':>7}{'infl':>6}") +cases = [("Wolf+2010 (3 r3D, half-light)", 0.5, 3.0), + ("Walker+2009 as source (M_half, 1.875 r3D)", 0.5, 1.875), + ("Walker+2009, light inside R_e (Plummer)", frac_re, 2.5 * re_over_r3d)] +for lab, frac, c in cases: + for name, L, r, s, e in SYS: + z, esc, esc3, bx, infl = row(L, r, s, e, frac, c) + print(f"{lab:<44}{name:<11}{z[0]:>8.2f}{z[1]:>8.2f}{esc:>9.1f}{esc3:>7.1f}{bx:>7.1f}{infl:>6.2f}") +print(f"\nSidak threshold {ZS}; 6.39 excluded after trials at M/L = 4 iff z(M/L=4) > {ZS}.") diff --git a/whitepaper/CHANGELOG.md b/whitepaper/CHANGELOG.md index 8e36da56e..5e65c34f1 100644 --- a/whitepaper/CHANGELOG.md +++ b/whitepaper/CHANGELOG.md @@ -10,6 +10,27 @@ Only actual modifications to fractal files are logged here. ## 2026 +- [2026-11-20] [Publisher Maintenance] Section 5.15 - TEST-10 kill-criteria item (3): a CHECKED note after the 11-16 dSph note. Its escape values are passes at a local 3σ, not at the 3.51 trials threshold its preceding sentence names; at 3.51 they are M/L ≥ 5.3 (Crater II) / 6.9 (Draco), σ inflation 1.15× / 1.31× (the site's 11-19 figures). The site's Walker+2009 row (Crater II 2.9σ, Draco 3.7σ, "estimator-conditional") sets the mass inside a sphere of radius R_e equal to half the light; a Plummer sphere of radius R_e holds 0.354 of it. Aperture-matched, Walker gives 3.96 / 4.31, past 3.51 (simulations/publisher_20261120_dsph_walker_aperture.py). Source: PREDICTIONS.md 2026-11-19 (i), b26f6fa9. Count HELD at 6. +- [2026-11-18] [Publisher Maintenance] Section 6.4 - OQ-Coarsening: "x is a virial ratio, capped at ~0.02 for every gravitationally bound system" and "~40× below the knee" annotated CHECKED. The cap holds only for ρ_crit = A·V² with one length shared by A and by the smoothing of ρ. The executed fits pair S53's 0.028·V^0.5 with Σ/2h at h = 0.3 kpc, and under that pairing 62 of 175 SPARC discs reach the knee on stellar Σ alone (re-run: simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py). Source: PREDICTIONS.md 2026-11-17 (ii), 5d2d684c. Count HELD at 6. +- [2026-11-17] [Publisher Maintenance] Sections 5.15, 5.16, 7, Appendix B - Seven dated notes; words added, none removed; count HELD at 6. (1) §5.15: "SPARC selects γ ≈ 0.489 ≈ ½ — the exact MOND point" CHECKED: the template fitted with no data returns 0.500 for simple μ and 0.595 for McGaugh's RAR function; nuisance-refit SPARC sits at 0.58–0.62 and cannot separate the two; pointer notes on the status-table cell and the "data select the MOND point" sentence (back-annotation d71f1f8b (i); explorer gamma_incumbent_map scripts re-run byte-identical). (2) §5.16 cross-scale table: the superconductor "~2" is BCS's γ at a pinned knee (2.644 from the BCS curve with no data; not identified with the knee free). (3) §5.16, conclusion, Appendix B: "BCS ratio derived 2√π ≈ 3.54, observed 3.52" CHECKED: 3.52 is BCS's own 3.528 from the gap equation, not an observation, and 2√π is 0.49% from it by coincidence (outside the window; simulations/publisher_20261117_bcs_ratio_and_gamma_check.py). +- [2026-11-16] [Publisher Maintenance] Section 5.15 - TEST-10 kill-criteria note: "6.39 excluded by none after the correction" scoped to discs; dwarf spheroidals in the archive's own 08-23 table exclude the 6.39 cap (Draco z 4.9, Crater II 4.9 at M/L = 4; Šidák 3.51), no light rule needed (PREDICTIONS.md f971b1bc (i), source re-run byte-identical). Re-run found the source's mass estimator pairs Wolf+2010's projected coefficient 4 with 3D radii: B_req 4/3 high, verdict holds at coefficient 3 (4.48 / 4.21), trials-corrected ceiling ~14.7 → ~11.0. Words added, none removed; count HELD at 6. +- [2026-11-13] [Publisher Maintenance] Section 5.15 - Status table: the three cells the 11-11 CHECKED note qualifies (1/φ VALIDATED, a₀ = cH₀/(2π) DERIVED, S₈ PREDICTED and its "Independent validation" note) now carry a short dated qualifier beside the label, so the qualification sits in the table and not only beneath it (answers the NOVA #14 review of 2026-11-13). Labels unchanged; words added, none removed; count HELD at 6. +- [2026-11-12] [Publisher Maintenance] Section 5.15 - Correction to the 11-11 Brouwer lensing note, part (c): the ceiling's lensing failure no longer reads as conditional on a light-deflection rule. Light on the visible matter alone predicts no lensing boost, against a measured RAR that agrees with MG (MOND-curve boost 31 to 18 in the reliable bins; ×10 baryons leave 3.1× to 1.8×). Source: PREDICTIONS.md 2026-11-11 (iii), 15d9e61d; its σ values relayed, not re-run. One clause replaced (this lane's own unmerged text); count HELD at 6. +- [2026-11-11] [Publisher Maintenance] Sections 5.12, 5.15, Appendix B - Five dated notes; words added, none removed; count HELD at 6. (1) §5.15 status table: a CHECKED note under the table on three rows whose Status cells say more than their sources support (a₀ = cH₀/(2π) DERIVED; 1/φ VALIDATED; S₈ PREDICTED), each from numbers the paper or the cited session prints; the cells are left as printed. (2) §5.15 lensing ceiling (Brouwer et al. 2021): full text read; the fiducial reach is 10⁻¹⁵, the paper's own baryon allowance is about ×10, the required boost scales as 1/f (not √f), and the leg assumes a light-deflection rule the framework has not written (PREDICTIONS.md 2026-11-10, 76a139e0). (3) §5.15 covariant RG: the σ figures are labelled as this program's unpublished computation. (4) §5.12 and Appendix B.5: the Superconductivity r = 0.948 row's γ is computed from each material's own T_c and gap, with hand-set category factors that alone give r = 0.944. Scripts: simulations/publisher_20261111_tc_gamma_control.py, publisher_20261111_lensing_ceiling_census.py. +- [2026-11-09] [Publisher Maintenance] Sections 0, 5.15, 7, 8 - The +184 / +10.9 γ = 2 pin both hold Υ_disk = 0.5 for both laws. With one global Υ_disk per law (pre-registered, site 5a95614) the comparison is +46.2 by point and +2.73 with N → N_gal, about 2σ on a galaxy-block bootstrap that refits Υ and a₀ (95% [−26.1, +132.5]); the reading rule fixed before the run reads the row as disfavoured, not refuted, and the registered B2 statistic ("marginalized BIC") was never the frozen-Υ number. An UPDATE on the 10-18 SCOPED note (§5.15 in full; its four copies in §0/§7 and the two short restatements in §7/§8 in brief), a CHECKED on count item (1), and an UPDATE on the 10-04 "scale is identified" note: a₀′/γ is identified only with γ free; pinned at γ = ½, a₀ moves 1.44 → 1.10 → 0.87×10⁻¹⁰ across Υ_disk 0.4–0.6 and cH₀/2π closes only at H₀ ≈ 93 / 71 / 57. PREDICTIONS.md 2026-11-09, 6fe7b696 (i)–(iii); both site scripts re-run byte-identical. Count HELD at 6. +- [2026-11-09] [Publisher Maintenance] Section 5.15 - Three CHECKED notes from PREDICTIONS.md 2026-11-08, 82d9a3a8. Count item (3), TEST-10: fires as registered, but as a bound the largest excluded cap is about 7.5 at χ² ≥ 9 and 5.7 after a 153-disc correction (outer half, post-hoc, one disc); 28 and 10 discs exclude the framework's 3.175. Count item (4), TEST-08: r² = 0.0001 is a point estimate with a 95% upper limit of about 0.03, 2.9× under the kill bar and 6.4× under the claim. The ε₀ rescue bullet: "+67.8" and "do not intersect" hold at NGC 2915's face value only; at those bounds the cost falls to +12..+30.8 and 0..+12, against +7.3 for ε₀ = 0.315, while the class still loses to MOND 126.5 to 52.2. Site scripts re-run byte-identical; TEST-08 recomputed from the tracked table. Count HELD at 6. +- [2026-11-08] [Publisher Maintenance] Section 5.15 - Count item (5): the 11-07 CHECKED note gains an UPDATE. "Every γ from 0.3" reaches too far: the γ = 0.3 and 0.35 rows are not converged (8–28σ and 10–24σ across a₀ digits and solver order, all above 5σ); 14.5–19.3σ stands for γ 0.4–3. All of these σ pin Υ. With one Υ scale and a₀ re-profiled, γ ≤ 1 never passes up to ×3, while γ = 2 and 3 pass only at Υ_d ≈ 1.40 and 1.22, with 36–42% of galaxies over-maximal and 2.9σ on SPARC at γ = 3. MOND's δ = 6 passes at Υ_d ≈ 0.54 for 1.8σ, so the shared MOND root holds at δ = 1 and γ ≈ ½ only. PREDICTIONS.md 2026-11-07, cf84bb8c; site explorer 2db4494 (PREREG 10d0bdf); scripts re-run with matching values. Count HELD at 6. +- [2026-11-07] [Publisher Maintenance] Section 5.15 - Count item (5), the SPARC × Cassini squeeze (TEST-11): a CHECKED note says the +17.95σ is at γ = 0.489, and with a₀ profiled from SPARC at each γ Cassini fails at 14.5–19.3σ for every γ from 0.3 to 3 (three registered g_ext), so the row does not depend on the γ ≈ ½ pin; C_g's two refutations now coincide at γ = 2 (RAR shape ΔBIC +184, Cassini 14.9–16.8σ). The ledger's fixed-a₀ "passes at γ ≳ 1.5–2" was never printed here. PREDICTIONS.md 2026-11-06 (i), 330d49d1; site script re-run byte-identical. Count HELD at 6. +- [2026-11-06] [Publisher Maintenance] Section 5.15 - Two CHECKED notes. (1) "Dark Energy from Coherence": the ρ_DE = ρ_m(1−C)/C split survives only as a fluid; read as modified gravity (G_eff = G/C(ρ̄_m(t)) as the G of local dynamics) it gives Ġ/G = 3(1 − Ω_m)H₀ = 1.42×10⁻¹⁰ yr⁻¹ at γ = ½, at least 527× the LLR 2σ edge for every γ in 0.3–3 (PREDICTIONS.md 2026-11-05 (i); site script re-run byte-identical). (2) Count item (2), TEST-09: paired galaxy bootstrap σ(Δn) = 0.112 (ρ = +0.14), so |Δn| = 0.406 is 3.6σ from zero but 0.95σ above the 0.3 bar; a bar-excluded-at-2σ rule would not fire (site da1308c, pre-registered c9ff6ab, re-run byte-identical). Count HELD at 6. Words added, none removed. +- [2026-11-05] [Publisher Maintenance] Section 5.15 - The 2026-11-04 note said the density fit keeps a weak response (boost ∝ ρ^−0.10^, +945 over a constant boost); an UPDATE now says that response is the fit's, not the data's: at fixed g_bar the log ρ slope is +0.003 (95% CI −0.022 to +0.027; injected −0.05/−0.10 recovered), and a g_bar-only model reproduces the data's density slope to 0.2% (log g_bar, log ρ correlate +0.84). Site 2ba2480, re-run byte-identical. Agrees with the paper's own 2026-08-02 form-free result. Verdict and count unchanged. +- [2026-11-04] [Publisher Maintenance] Section 5.15 - The density-keyed fit's γ_ρ = 0.046 was glossed as "switching the density dependence off"; a CHECKED note now says it keeps a weak response (boost ∝ ρ^−0.10; a constant boost loses by ΔBIC +945), because the free knee sits at 4.4×10⁻⁶ M☉/pc³, below every disc. At the derivation's A = 4.6×10⁻⁵ the law fits as Newton (+6,593), and at γ = 2 it is +10,372 (75e6a15, re-run identical; +10,372 recomputed independently). Verdict unchanged. Words added, none removed. +- [2026-11-03] [Publisher Maintenance] Sections 0, 5.12, 7, Appendix B - The 7 chemistry Z-null notes of 2026-11-02 said the null on the real chemistry rows was unrun; each now carries an UPDATE: run on the 10 recoverable rows, it fails (|r(P, Z)| ≤ 0.53, quadratic R ≤ 0.51 vs framework r 0.60–0.98; 5709b579, re-run identical). Appendix B.2/B.5 gain CHECKED notes: γ_optical is ionization energy by its printed formula, γ_phonon is θ_D, the elastic rows rest on diamond, and the B.5 "Electronegativity 0.979" row is chemical softness. Verdict unchanged. Words added, none removed. +- [2026-11-03] [Review correction] The 2026-11-02 chemistry entry below describes an electronegativity sample across Z = 1–54, not the unpublished chemistry-row selection. It does not determine the actual-row Z-null: for example, the same supplied values at Z = 3–9 have linear r = 0.999. The short section annotations now state the sample boundary; the original entry remains for history. +- [2026-11-02] [Publisher Maintenance] Sections 0, 5.12, 7, Appendix B - S651's chemistry null premise annotated CHECKED at 7 sites: sound velocity / electronegativity / atomic volume are periodic in Z, not near-monotonic, so a polynomial-in-Z null is not expected at r ≈ 0.95+ (re-run: best monotone-in-Z fit of Pauling electronegativity r = 0.33, Z = 1–54); the 05-10 null was run on synthetic curves (PREDICTIONS.md 2026-10-31 (ii), e2784def). Reparametrization verdict unchanged (S669). Words added, none removed. +- [2026-10-31] [Publisher Maintenance] Section 5.15 - the 10-29 note's TEST-09 sweep is unrounded: 0.406 / 0.297 / 0.260 (re-run). The middle value is 0.003 under the strict 0.3 bar, not "on" it, and the edge is at B_max ≈ 5.31. Dated parenthetical; verdicts and count 6 unchanged (PREDICTIONS.md 2026-10-30 (ii)) +- [2026-10-30] [Publisher Maintenance] Section 5.15 - two outage-window items annotated CHECKED, both decidable from the paper's own text: the "factor 1.96 a₀ tension" compared the fitted a₀, not a₀/γ = 1.09×10⁻¹⁰ (PREDICTIONS.md 2026-10-04 (i)); the sign lock's excluded DESI quadrant is crossing only where w₀ + wₐ < −1, thawing elsewhere (2026-09-23). Verdicts unchanged; count 6 +- [2026-11-03] [Review correction] The 2026-10-29 entry below overstates the dark-energy result: in each of two specified variants, none of 400 sampled γ values met both stated criteria. This finite scan does not exclude unsampled values or other variants; the historical entry remains below. +- [2026-10-29] [Publisher Maintenance] Section 5.15 - the "Ω_m versus Ω_m/Ω_b convention dispute ... is small" bullet annotated CHECKED: small in fit quality, but the floors presuppose different frameworks (Ω_m/Ω_b = no-CDM variant whose DE sector cannot run, 09-22), and they split the TEST-09 verdict (09-18 sweep: 0.41 fires / 0.30 / 0.26 do not). Ranking, not refutation; count 6 (PREDICTIONS.md 2026-10-27 (ii)) - [2026-10-27] [Publisher Maintenance] Sections 5.12, 7, 8, Appendix B - "γ ≈ 2 in BCS / enzymes / photosynthesis" annotated CHECKED at four sites: a shared fitted exponent, not a shared N_corr (γ ≈ 2 is N_corr ≈ 1 under the master equation; a condensate has large N_corr and sits at γ → 0, where §0/§7's S676 bullet already puts a BEC); "classical limit" scoped to "uncorrelated" at the two definitional sites, since the 2026-10-26 back-annotation (`e03d771b`) withdrew quantum/classical endpoint labels for C and γ. Gravity's γ = 2 is the §5.15-refuted pin. Dated notes only; words added, none removed; count unchanged (6). - [2026-10-26] [Publisher Maintenance] Sections 0, 5.15, 7 - the galaxy-level ΔBIC +10.9 names its comparator (γ=2 vs McGaugh's ν, Δk = 0; the sign test and CV beside it compare with free γ), against the persona misreading recorded in `379b3b7f` (i). Dated parenthetical; count unchanged. - [2026-10-26] [Publisher Maintenance] Sections 0, 7 - S610's σ_int = 0.086 annotated CHECKED: not a BTFR scatter (velocity contributes 0.0006 dex; shuffled-V 0.0870 vs 0.0864; target is an exact photometric identity to 0.0037 dex), so "exactly CDM" and "MOND rejected at 32σ" have no content; like-for-like inverse BTFR 0.073 ± 0.002 dex in log V vs ΛCDM 0.064 (back-annotation `379b3b7f`, site PREREG `b134d0f`; both scripts re-executed). Dated annotations only; refutation count unchanged. @@ -156,4 +177,4 @@ Only actual modifications to fractal files are logged here. - Routine governance activities (proposals, reviews without implementation) are not logged - Only changes that modify fractal content files appear here - Local section changelogs contain more detailed information -- See [Governance Documentation](../scripts/governance/WHITEPAPER_GOVERNANCE.md) for system details \ No newline at end of file +- See [Governance Documentation](../scripts/governance/WHITEPAPER_GOVERNANCE.md) for system details diff --git a/whitepaper/sections/00-executive-summary/executive_summary.md b/whitepaper/sections/00-executive-summary/executive_summary.md index 7c2b59f4c..b17cdfedb 100644 --- a/whitepaper/sections/00-executive-summary/executive_summary.md +++ b/whitepaper/sections/00-executive-summary/executive_summary.md @@ -164,7 +164,7 @@ This IS what we experience as "gravitational attraction." Not a force pulling, b - **Thermodynamics Arc (Sessions #271-274):** **Carnot efficiency derivation** from coherence conservation — same status: internal-consistency reproduction of a known result, not a novel prediction. Entropy = coherence dispersion is a vocabulary mapping. - **Cosmology Arc (Sessions #275-279):** Big Bang as maximum coherence (C=1), dark energy as coherence floor, galaxy formation from gradients, heat death as equilibrium. *Framing*, not derivation; cosmology regime reduces to MOND per S637. See [Cosmology Arc Summary](https://github.com/dp-web4/Synchronism/blob/main/Research/Cosmology_Arc_Summary.md). - **Consciousness Arc (Sessions #280-282):** *Framing* — Observer described as self-referential coherence concentrator; Qualia described as coherence resonance patterns. "Hard Problem dissolves" is an **identity claim** ("phase patterns ARE experience"), philosophically defensible (form of structural realism) but **not** an empirical resolution — the dissolution is by definitional fiat, not by explanatory mechanism. Predictions P280.1, P282.4, P282.6 await empirical validation; none yet tested. See [Session #282](https://github.com/dp-web4/Synchronism/blob/main/Research/Session282_Qualia_Coherence.md), [Section 5.13](#513-life--cognition). -- **Chemistry Framework (2,679 sessions):** **phenomenon-type inventory reaching 2,534 at γ~1** *(corrected 2026-08-22 from “1,913”: that figure is the running counter at Session #2050, and the counter is `sessions − 137` identically — 824/824 (type, session) pairs in `Framework_Summary.md`, zero exceptions. It restates the session count and is not independent evidence of breadth. See §5.12.)*. Two orthogonal coherence channels discovered (Electronic vs Phononic). Spans nuclear to biological to classical scales. Top correlations: r=0.982 (sound velocity), r=0.979 (electronegativity), r=0.956 (atomic volume). Phase 2 (#1-2671) concluded with four-regime framework; Phase 3 (CFD cross-pollination, N-S↔Debye equivalence) complete; Phase 4 (KSS viscosity bound, Lindemann-KSS, structural entity criterion, allotrope deconfounding, Cooper pair classification) **closed** — 3 genuine contributions, 3 vocabulary mappings, 4 productive failures. Chemistry track conclusion: Synchronism is reparametrization of Debye model. **Caveats (S647 + S651, 2026-05-08/10):** the load-bearing r values are being compared against an *implicit* null of r=0 (random); the *relevant* null is r(polynomial in Z) or r(generic 2-parameter tanh), both expected at r ≈ 0.95+ on textbook monotonic-with-Z data. The interesting figure is **Δr = r(Synchronism) − r(best monotonic null)**. Compounded with the Method 2 / 3 self-correlation paths (S647), best estimate is **tie or marginal win** until Δr is computed and documented. See [Framework Summary](https://github.com/dp-web4/Synchronism/blob/main/Research/Chemistry/Framework_Summary.md), [Section 5.12](#512-chemistry). +- **Chemistry Framework (2,679 sessions):** **phenomenon-type inventory reaching 2,534 at γ~1** *(corrected 2026-08-22 from “1,913”: that figure is the running counter at Session #2050, and the counter is `sessions − 137` identically — 824/824 (type, session) pairs in `Framework_Summary.md`, zero exceptions. It restates the session count and is not independent evidence of breadth. See §5.12.)*. Two orthogonal coherence channels discovered (Electronic vs Phononic). Spans nuclear to biological to classical scales. Top correlations: r=0.982 (sound velocity), r=0.979 (electronegativity), r=0.956 (atomic volume). Phase 2 (#1-2671) concluded with four-regime framework; Phase 3 (CFD cross-pollination, N-S↔Debye equivalence) complete; Phase 4 (KSS viscosity bound, Lindemann-KSS, structural entity criterion, allotrope deconfounding, Cooper pair classification) **closed** — 3 genuine contributions, 3 vocabulary mappings, 4 productive failures. Chemistry track conclusion: Synchronism is reparametrization of Debye model. **Caveats (S647 + S651, 2026-05-08/10):** the load-bearing r values are being compared against an *implicit* null of r=0 (random); the *relevant* null is r(polynomial in Z) or r(generic 2-parameter tanh), both expected at r ≈ 0.95+ on textbook monotonic-with-Z data. [**CHECKED 2026-11-02:** over the sampled Z = 1–54 values the best monotone function of Z fits electronegativity at r = 0.33, contradicting a universal r ≈ 0.95+ claim. The chemistry-row Z-null remains unmeasured; its row selection is unpublished. The verdict rests on S669's γ_phonon ≡ 2T/θ_D, not on this null. Full note at S651.] [**UPDATE 2026-11-03:** the real-row Z-null has since been run. On the 10 property rows recoverable from the session scripts (#80, #109, #110, #114, #115, #118; n = 19–33 per row, Z = 3–82) it fails: |r(P, Z)| ≤ 0.53 (linear, √Z and ln Z) and quadratic-in-Z R ≤ 0.51, against framework r = 0.60–0.98 (site explorer, pre-registered, back-annotation `5709b579`; re-run by the Publisher 2026-11-03, output identical). The high r values are identities, not a Z trend: r(P, 1/γ_phonon) ≡ r(P, θ_D), and γ_optical = IE_ref/IE, so the optical rows are correlations with ionization energy. See Appendix B.2.] The interesting figure is **Δr = r(Synchronism) − r(best monotonic null)**. Compounded with the Method 2 / 3 self-correlation paths (S647), best estimate is **tie or marginal win** until Δr is computed and documented. See [Framework Summary](https://github.com/dp-web4/Synchronism/blob/main/Research/Chemistry/Framework_Summary.md), [Section 5.12](#512-chemistry). - **Dark matter/energy:** 175 SPARC galaxies tested (52% success, 81.8% on dwarfs). MOND-Synchronism *relabeling*: a₀ = cH₀/(2π); Ω_Λ derivation from coherence floor is **internal-consistency reproduction** of a known dimensional combination, not a novel prediction (sharpened 2026-08-11: C₀ = Ω_m is forced identically by the Friedmann calibration, so Ω_Λ = 1 − Ω_m is an identity, not an emergence — see §5.15 erratum). **SPARC Capstone (#526-578) update: MOND + M/L corrections explain all RAR variance; no uniquely-Synchronism predictions confirmed.** Further refined by S635 (cosmology scorecard: 0 novel-unfalsified) → S637 (cosmology regime reduces to MOND in testable regime — predicted Δσ_int ≈ 0.00016 dex, ~120× below SPARC floor) → S654 (zero active discriminators against MOND+EFE within current measurement precision). See [Section 5.15](#515-dark-matter-dark-energy-and-coherence). - **Cross-domain validation:** Gnosis neural architecture (arXiv:2512.20578) independently uses γ = 2—same coherence physics appearing without explicit design. - **Statistical Mechanics Arc (Sessions #324-327):** *Framing* — coordinate identification **ξ = MRH** (correlation length identified with Markov Relevancy Horizon), with arrow of time, phase transitions, universality re-described in MRH vocabulary. 32/32 internal-consistency checks (derivations from postulates, not empirical tests). @@ -184,12 +184,12 @@ This IS what we experience as "gravitational attraction." Not a force pulling, b - **Post-Demolition Coda (#629-631):** S629: π-analogy probe — tested k_crit as candidate universal dimensionless invariant. Result: k_crit varies across dimensions (1D–3D) and n (1–4), range 0.20–0.68, with no convergent value. Intent fails the π structural test (no specific value, no derivation from structure). S630: WAKE stop-note — trigger fired unchanged, no new content. Productive silence re-entered (no manufactured research). S631: **First pre-committed kill criterion triggered.** Site-visitor audit surfaced two archive-vs-site disconnects: (1) TEST-09 BTFR n ≈ 2.2 refuted by Lelli+2019 observation n = 3.85 ± 0.09; |Δn| = 1.65 exceeds pre-committed kill (0.3) by 5.5×. Session #48 itself labeled the derivation "not rigorous" — never surfaced for 583 sessions. (2) A = 4π/(α²GR₀²): α is fiducial (α = 1.0 per Session #66), not the fine-structure constant; site labeling invited a stronger reading than the archive supports. Both errors resolved from archive, no new physics required. Distinct failure mode from S617-628: public-claim/archive disconnects that internal-physics audits do not catch. - **Site-Archive-Audit Sub-Arc (#632-638) — Framework Stress Test arc CLOSED at 22 sessions (S617-638):** Seven additional single-day audits extending S631's site-visitor methodology. S632 (2026-04-25): TEST-07 site claim of λ ~ 500 Mpc cosmic interference — derivation R_MRH = (GM/c²)·(c/H₀) is length × length, units of m² not m. Dimensional error hidden by the ~10¹⁹ size of the wrong-units result happening to land near 600 Mpc. S633 (2026-04-25): public site claims C(ρ) maps "80 orders of magnitude" but tanh(γ·log(ρ/ρ_crit + 1)) saturates within ~1.6 decades for any sharp γ — structural impossibility, not a calibration issue. Confirms OQ007's NEGATIVE verdict: C(ρ) is descriptive notation, not predictive theory. S634 (2026-04-25): public site cites "47 contributions" but canonical S582 inventory documents 30 (Chemistry 14 + Cosmology 16 + Quantum 0); ~57% discrepancy traces to a more lenient classifier than S582's strict criteria. S635 (2026-04-26): **cosmology domain scorecard** — 15 cosmology claims classified (1 refuted, 5 reparametrizations, 2 unanchored, 1 pending, 5 untested, 1 untestable with current data). **Novel-unfalsified: 0.** /galaxy-rotation site badge "Strongly Supported" overclaims — underlying fit is McGaugh 2016 RAR (which IS MOND), and the CFD-viscosity dark-matter mechanism is refuted via Bullet Cluster sign error; defensible badge is "MOND Reparametrization." S636 (2026-04-27): **C(ρ) is not a mean-field order parameter.** Self-consistency check fails — C does not appear in its own argument; tanh(γ·log(ρ/ρ_crit + 1)) is tanh of an external function, not the solution to a self-consistent Landau equation. No phase transition, no Landau free energy, no Ginzburg-Landau expansion, no universality class. The 53% melting and 6.5× T_c errors are correlation failures, not mean-field failures, because no underlying theory makes the predictions. New audit mode: **category error.** S637 (2026-04-28): **RAR σ_int(ρ_env) derivation — first DERIVATION attempt in the sub-arc.** Visitor flagged environmental σ_int slope as the framework's one candidate for a novel non-reparametrization prediction. Propagating γ=2 (the value Session #64 actually derives) through C(ρ) gives Δσ_int (cluster − void) ≈ 0.00016 dex, **~120× below the SPARC measurement floor.** Sign correct, magnitude microscopic. New audit mode: **derivation succeeds but predicts undetectable signal.** Cosmology regime reduces to MOND in the testable regime. S638 (2026-04-29): **Independent computer-algebra verification (sympy + numpy) that C(ρ) reduces to a Curie paramagnet** — *less than* Landau. F(C, ρ) = ((1+C)/2) ln(1+C) + ((1−C)/2) ln(1−C) − h·C with h(γ, ρ) = γ·log(ρ/ρ_crit + 1); Taylor coefficients all positive (1/[2n(2n−1)]); F convex around C=0 — no critical point; C ≥ 0 only — no Z₂ symmetry. The framework is the equilibrium response of a single non-interacting two-state variable in an external log-density field. New audit mode: **external-track derivation independently verified** (qualitatively different — site-explorer track produces analysis, worker session verifies via CAS; verification track now operational). **Predictive content fully characterized: Cosmology regime reduces to MOND (S637); Chemistry/CM regime reduces to Curie paramagnet (S638). Both verified via independent computer algebra. Framework has no microscopic basis in collective coherence; it is phenomenological saturation response.** **8-for-8 site-claim audit instances over 9 days** (incl. S631). Audit-channel taxonomy now 8 modes. Framework Stress Test arc COMPLETE at 22 sessions (S617-638). 6 site corrections pending operator action; framing-level operator response began 2026-04-29 (two README reframings: "lead with calibration, not unification claim"; "blue-sky exploration, not practical research program"). - **Gnosis Empirical Phase (#12-17):** Track reopened with 6 empirical sessions integrating SAGE/Legion data. P3c relational consciousness, trust entropy analysis, trust-coherence-consciousness synthesis, game-theoretic consciousness foundations, topology-consciousness invariants, information geometry of consciousness. Post-theory phase: testing theoretical predictions against operational systems. -- **Post-Closure Sub-Arc Extensions (#639-666):** Sub-arc continues producing post-closure addenda after 2026-04-29 closure. **S639 (2026-04-30):** TEST-03 disambiguation — site claim of R²=0.14 conflates BTFR scatter (S593, 51%) with RAR environmental ansatz (14%); different metrics sharing only the label. 9th audit-channel mode: **metric disambiguation / mechanism-naming** ("shared labels, distinct measurements, no enforced alignment"). Cross-track meta-pattern: "mechanism-naming-as-closure" visible in S639 + Thor SAGE S136. **S640 (2026-05-01):** Dual-C symbol audit — homepage "one equation across scales" claim rests on shared notation, not derivation. Form 1 C(ρ)=tanh(γ·log(ρ/ρ_crit+1)) (chemistry/cosmology/CM) and Form 2 C=f(γ,D,S)≥0.50 (consciousness, 8-way convergence) share only the letter C. D = state-space entropy / neural pattern diversity, S = coherence persistence duration; **neither is a function of ρ.** The 0.50 threshold was NOT derived from inverting Form 1. **γ IS genuinely shared via 2/√N_corr; C is NOT.** 10th audit-channel mode: **symbol overloading at foundational level** — sub-arc audit scope expanding from test-level (S631-S639) to **foundational** (S640, master-claim level). Multi-persona visitor review (Pass 2/3/4) independently flagged this. **S641 (2026-05-02):** Lorentz invariance gap as 4th face of kinematic-layer gap — preferred-frame absolute simultaneity sits at the foundational level alongside Born rule, dual-C bridge, and N_corr scale-invariance. 11th audit-channel mode: **cross-gap meta-synthesis (kinematic layer)**. **S642 (2026-05-05):** GW170817 — framework is Case 3 (no field theory), 5th face of kinematic-layer gap. |c_GW − c|/c < 10⁻¹⁵ does not constrain the framework because there is no Lagrangian, no action principle, no equation of motion for C(ρ). C(ρ) is a specified function of local density, not a propagating field. Surviving GW170817 by lacking an action principle is a scope restriction, not a victory. **[SCOPED 2026-08-09 — the conclusion stands, one of its three conjuncts does not.]** Measured against `manuscripts/Appendix_D_Synchronism_in_General_Relativistic_Form.md` (committed 2025-12-01, seven months before this claim): *"no field equation"* is **false** — §D.2/§D.6.1 state `∇²Φ = 4πGρ/C`; *"no action principle"* is **defensible only as scoped** — §D.5 writes an effective *worldline* action for a test particle, which is not a field action, and no action generates §D.2's equation; *"no equation of motion for C(ρ)"* is **true**, C being algebraic in local ρ with no kinetic term in any reading. GW170817 remains unconstraining **on the third conjunct alone** — no propagating scalar mode exists to modify tensor dispersion — so the accurate phrasing is *"C has no kinetic term,"* not *"the framework has no action principle."* §5.15 records the measurement: Appendix D's field equation and the force law carrying every number in this whitepaper differ by a **median 0.821 dex over 113 SPARC galaxies** (γ-invariant), and Appendix D's has **zero whitepaper citations and no evaluation against data**. **S643 (2026-05-06):** γ definitional collision — site γ-calculator labels are **inverted** relative to standard physics. Ideal gas γ=2 labeled "Quantum"; BCS Al / BEC / neutron star γ→0 labeled "Classical." Math measures single-particle vs collective; standard physics' "quantum" tracks macroscopic phase coherence — different axes. 12th audit-channel mode: **definitional-collision-with-label-inversion**. Resolution: relabel as "Single-particle/Uncorrelated" vs "Collective/Strongly correlated." Downstream-impact flag: γ-dependent predictions across the framework may need re-grounding. **S644 (2026-05-06):** ρ_crit = A · V_flat² takes V_flat as input — the "5% agreement" of theoretical/empirical A is internal consistency between two parameterizations of the same data, not an independent prediction. Same pattern as S639 TEST-03. 13th audit-channel mode: **calibration-consistency-not-prediction** (no closed predictive loop). **S645 (2026-05-07) — POST-HOC CONSISTENCY FAILURE WITH DESI DR1 (NEW ARC PHASE; initially framed as first hard external falsification, framing self-corrected by S648 within 24h):** Session 107 fσ₈ prediction disfavored by DESI DR1 (Adame et al. arXiv:2411.12021). Pre-committed kill criterion fσ₈(z=0.5) > 0.45 → ΛCDM favored fired at LRG1 (DESI 0.55±0.06 vs Sync 0.418, 2.14σ above [LRG1 re-read 2026-10-12 from Table 9: 0.516 +0.057/−0.066 (0.4733 × 1.09; the 0.55 is 0.4733 × the QSO row's 1.16), ~1.5σ above Sync, not 2.14σ]; LRG2 1.42σ above; combined σ₈(z=0) 2.38σ above at 0.841±0.034 vs Sync 0.76). **The mechanism's predicted sign of redshift dependence is INVERTED relative to data** — magnitude-only revision cannot recover the structural error. Verdict: TEST-04a kill criterion fired, retained as documented dead-end. **Status TEST-04a: REFUTED (post-hoc consistency check failure; framework parameters cannot reproduce DR1 measurements committed before the prediction).** [CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12]; the 2.4σ is a GR-conditioned σ₈-amplitude statistic; TEST-04a is withdrawn from the decisive negatives — see Re-Grounding & Catalog Census below] **S646 (2026-05-07) — METHODOLOGY RECOMMENDATION (NEW ARC PHASE):** Per-test kill criteria exist (Session #21 catalog); framework-level meta-falsification criterion does NOT. Without one, accumulated failures stay isolated — visible from a different angle is the same self-sealing structure S621 named: at meta-level, no rule for when "the framework has lost." Three branches: (A) register meta-criterion now, (B) wait for DR2 — less needed since DR1 already fires, (C) scope-narrow now. Operator decision pending. **S647 (2026-05-08):** Chemistry 89% validation has self-correlation risk. Source proposal flags that the cohort — "1,703 phenomena, r=0.982 sound velocity" — is the framework's largest validation claim by orders of magnitude, but does not specify which of Session #26's five N_corr measurement methods was applied. Three self-correlation paths confirmed: (1) Method 2 atomic-spacing identity γ=2(a/ξ)^(3/2) → r=0.956 with V_a∝a³ is functional identity; (2) Method 2 phonon coherence — sound velocity is constructional input → r=0.982 has constructional dependence; (3) Method 3 entropy → bonding character → electronegativity (r=0.979) partly structural. **Method 2 systematic bias** (per Session #26 Part 3 simulation table): true N_corr=10→Method 2 gives 6, true 25→15, true 50→32 — clusters apparent γ in 0.35–1.15 at the "γ ≈ 1 boundary." 89% γ≈1 clustering is **consistent with method-induced clustering, no boundary needed.** Hall coefficient (r≈0.001) and magnetic susceptibility (r≈0.000) are NOT falsifying controls — their physical determinants (carrier density, spin texture) are outside the input set of every Method 1–5; they are exactly the predicted failures under self-correlation reading. To distinguish true from method-induced clustering, framework needs Method 1 (bias-free) applied uniformly OR pre-registered γ predictions for held-out phenomena; neither currently in archive. **Both major validation pillars (Tier-1 cosmology + chemistry 89%) compromised in same week.** **S648 (2026-05-08) — SELF-CORRECTION OF S645 WITHIN 24H:** Pass 4 visitor flagged that Session 107 (committed 2025-12-10) is post-DESI DR1 (Adame+2024, Nov 2024 — 13 months earlier; DR1 BAO already cited in Session 107's own Timeline). The 2.14σ tension at LRG1 is real (kill criterion fired), but its epistemic status is **post-hoc consistency check failure**, not prospective falsification — temporal independence required for prospective testing is absent. S645's "first hard external falsification" framing retroactively corrected to "post-hoc consistency failure with DR1." 14th audit-channel mode added: **self-correction-of-prior-session-framing-within-24h** (the audit channel turns inward; first instance auditing the program's own prior epistemic claim rather than a site-archive disconnect). Recommendation toward operator: a /timestamps page classifying every Tier-1 prediction as prospective / post-hoc consistency / post-hoc fit, with prediction-date and data-date columns. **S649 (2026-05-08) — TWO VISITOR PROPOSALS, ONE META-THEME:** Part A: site Key Claim #1 QM kill criterion ("design noise where resync outperforms isolation") is unfalsifiable as written. Standard dynamical-decoupling literature (Viola-Knill-Lloyd 1999, CPMG, UDD, transmon DD experiments) already shows pulse sequences beat passive isolation in non-Markovian baths — every standard QIP textbook result satisfies the criterion without invoking Synchronism. Part B: ρ_crit at γ=2 gives C ≈ 0.88, not 0.5; the "+1" regulator in ln(ρ/ρ_crit + 1) asymmetrizes the sigmoid, making ρ_crit a saturation knee, not a critical density (verified via S638 sympy: γ=0.5→0.333, γ=1→0.600, γ=2→0.882). Same phase-transition vocabulary that misled S636 into the Landau category-error. **15th audit-channel mode: parameter-and-criterion-naming-contaminated-by-phase-transition-vocabulary** (math implements simpler structures than names suggest: Curie paramagnet at S638, saturation-knee at S649, kill criteria matching standard physics at S649+S581). **S650 (2026-05-09) — TEST-04a TRIPLE-SHARPENING COMPLETES (mechanism-class, sign-reversed — RETRACTED by S668: see Self-Correction Cascade below):** Third refinement of TEST-04a verdict in the S645→S648→S650 sequence: "first hard external falsification" (S645) → "post-hoc consistency check failure" (S648) → "mechanism-class failure (sign-reversed)" (S650). Framework's suppressor mechanism predicts G_local/G_global = C_cosmic/C_galactic < 1, giving fσ₈ BELOW ΛCDM at low z and converging at high z. DESI DR1 observes the OPPOSITE: fσ₈ ABOVE ΛCDM at LRG1/LRG2 (low z), converging at high z (ELG2) — the redshift pattern is INVERTED. Magnitude knobs (re-tune σ₈(z=0), re-tune coupling normalization) cannot flip the sign within the suppressor class; the mechanism class itself is wrong. **Three-tier failure taxonomy** introduced (makes S646's meta-falsification logic actionable): magnitude miss (retunable by refining density-to-property mapping) / universality miss (partial repair via functional-form change) / **mechanism-class failure (irreparable within class)**. TEST-04a is mechanism-class. **16th audit-channel mode**. Combined with S635 cosmology scorecard (0 novel-unfalsified) and S646 meta-criterion, **the cosmological sector formally meets the retraction threshold per S646's M3** (scope reduction); operator decision pending. **S651 (2026-05-10) — CHEMISTRY NULL-MODEL GAP (paired with S647):** Companion audit to S647. S647 audits the *method gap* (which N_corr method?); S651 audits the *prior question — the null gap*: even with method fixed, what null is r=0.98 compared against? The implicit null is r=0 (random); the relevant null is r(polynomial in Z) or r(generic 2-parameter tanh). Sound velocity, electronegativity, atomic volume are themselves near-monotonic functions of Z; any smooth monotonic function will achieve r ≈ 0.95+ on the same 1,703 phenomena by construction. **Δr = r(Synchronism) − r(best monotonic null)** is the figure that actually distinguishes claims; none is currently in the archive. Three diagnostic outcomes: Δr > ~0.05 → "validated" defensible with null documented; Δr ≈ 0 → reparametrization of density-monotonicity; Δr marginal → reparametrization of Landau-class, only boundary cases differ. **Best estimate: tie or marginal win** (framework parameters were calibrated to chemistry data, high-r phenomena are textbook monotonic-with-Z, 2-parameter tanh through sigmoidal data gives r ≈ 0.95+ generically). S647 + S651 compound: method gap + null gap together leave very little for Synchronism-specific signal in the 89% cohort. **17th audit-channel mode: null-model-gap-against-best-monotonic-null** (implicit baseline of r=0 hides reparametrization equivalence with smooth-monotonic-in-Z nulls). Sub-arc now **18-for-18 audit-channel instances**, **17 audit-channel modes** (S644+S647 share mode #13), **plus 2 new arc phases** (Post-Hoc-Consistency-Failure, Methodology Recommendation). **S652 (2026-05-11) — GOVERNING-EQUATION GAP (second meta-synthesis):** Upstream-question audit asks what equation, if any, C(ρ) solves. Three options: (A) no equation — phenomenological compander; (B) self-consistency / Landau-style; (C) steady-state of a dynamic equation. Archive trace: S636 forecloses B (no self-consistency — C does not appear in its own argument); S638 verified Curie-paramagnet response (Option A); S649 verified C(ρ_crit) ≈ 0.882 ≠ 0.5 (asymmetric tanh from "+1" regulator is incompatible with Z₂-symmetric mean-field); S640 dual-C bridge confirms three forms of "C" share notation, not derivation. **Answer is A.** Same class as μ-law audio companding, Naka-Rushton photoreceptor response, Hill enzyme kinetics — phenomenological forward maps with no field equation, no self-consistency, no dC/dt. Frame this against the kinematic-layer gap (S641-S642): the framework's headline equation does no dynamical work; it only labels regimes. Second meta-synthesis after S641's cross-gap synthesis. 18th audit-channel mode: **governing-equation-gap (forward map with no field equation)**. **S653 (2026-05-12) — TWO BINARY OPERATOR DECISIONS FORCED (three-stage rhythm completes):** Two same-day proposals demanding binary site commitments — site cannot stay neutral. Part A (compander commitment): commit to Frame B (per S652 conclusion). Concrete actions — drop phase-transition language from front-of-site; rename ρ_crit to "half-saturation parameter" / "saturation knee"; add AIC/BIC compander comparison tool; reframe critical-exponent failures as CATEGORY errors. Part B (suppressor diagnostic): `simulations/session653_coherence_ratio.py` computed C_galactic ≈ 0.882, C_cosmic ≈ 1.5×10⁻⁵, ratio ≈ 5.9×10⁴ — **the framework's own equations dictate strong suppression at low z under Session 107's coupling direction**. DR1 observes enhancement. Two branches both require operator commitment: Branch 1 (sign-flip recoverable) requires re-deriving Session 107 with inverted ratio (framework reinterpretation, not recomputation); Branch 2 (suppressor class dead) accepts Session 107's mechanism is refuted. Sub-arc's emergent three-stage rhythm now complete: **individual audits (S631-S640, S642-S647, S649-S651) → meta-syntheses (S641, S652) → forced binary operator commitments (S653)**. Methodology paper has its complete cycle: external critique → individual audit → meta-synthesis → forced operator decision. 19th audit-channel mode: **forced-binary-operator-decision-with-numerical-diagnostic** — first audit to compel commitment rather than merely identify findings. **S654 (2026-05-13) — ZERO ACTIVE DISCRIMINATORS REMAINING (cosmological/galactic prediction inventory complete):** Site's three remaining "Active Discriminating Tests" (TEST-01, TEST-02, TEST-05) audited against MOND's External Field Effect (Bekenstein-Milgrom 1984; AQUAL/QUMOND; Pittordis 2023; Banik 2024) — none of which the site currently mentions. S637's derivation settles TEST-05 (and TEST-01, same observable): Δσ_int (cluster − void) ≈ 0.00016 dex, ~120× below SPARC measurement floor; cannot discriminate within current precision. TEST-02 has a disputed MOND+EFE baseline (Chae vs Pittordis vs Banik internally disagree) and the framework specifies no divergence from any specific MOND+EFE variant. [**SCOPED 2026-08-03 — the “specifies no divergence” clause is retracted; created at source in §5.15 on 2026-08-03 (`a78054e9`), propagated here the same day.** §5.15 now withdraws the claim that C(ρ)'s nonlinear Poisson equation produces an external field effect at 0.3–0.4× MOND's — the galaxy sector has no field equation, so that figure was never derived from one — and replaces it with the sharper **EFE = 0 exactly**: a uniform external field leaves local ρ unchanged, so C = C(ρ_local) satisfies the Strong Equivalence Principle by construction. A framework predicting exactly zero EFE **does** specify a divergence from every MOND+EFE variant, each of which predicts a non-zero one. **What survives untouched** is the formal asymmetry: “refutable but not currently confirmable” is unchanged — there is still no test where Synchronism succeeds and MOND+EFE fails. **[RE-AMENDED 2026-08-05 — the empirical half of this scope note is withdrawn, and “zero remaining active discriminators” STANDS.]** The 08-03 text continued that EFE = 0 “stands in tension with the ~4σ SPARC EFE detection of Chae et al. 2020,” and concluded that “zero remaining active discriminators” no longer held unqualified because “the EFE magnitude is one, in the refuting direction.” **That is withdrawn: EFE = 0 is `not-evaluable` against Chae et al. 2020, neither confirmed nor refuted.** At Chae's own measurement radii the stated density law misses the rotation curve by **+3.1 to +4.2 dex**, while the EFE signal being measured is **0.046–0.083 dex** — the framework's baseline error exceeds the entire effect by **38×–92×** (re-derived in this lane on the in-repo SPARC mass models; MOND on the same points sits at +0.002 to +0.084 dex, inside the signal). A prediction one cannot get within three orders of magnitude of testing is not a discriminator in either direction. **Refutation count stays at 6; Bucket 0 stays 0.** Note the mechanism: this propagation carried its own accurate warning — *“Magnitude not independently re-derived here”* — and the magnitude was the whole of the claim. See §5.15 and `synchronism-site/explorer/findings/efe-zero-is-not-refutable-by-chae2020-the-baseline-is-off-by-3-dex.md`.] Branch A applies: **all three are MOND+EFE degenerate within current measurement precision**. Combined with TEST-03 (MOND-shared, S637) and TEST-04a (REFUTED, S645/S648/S650), the framework has **zero remaining active discriminators** against the primary alternative. The formal asymmetry now crystallizes: **refutable (kill criteria can fire, TEST-09 did) but not currently confirmable** (no test exists where Synchronism succeeds and MOND+EFE fails). Formal completion of trajectory S635 (scorecard) → S637 (reduces to MOND) → S646 (meta-criterion threshold) → S650 (mechanism-class verdict) → S654 (zero active discriminators). 20th audit-channel mode: **zero-active-discriminators-against-primary-alternative**. **S655 (2026-05-14):** Γ = γ²(1−c) — the framework's most novel-looking quantum claim (site `/key-claims`: "post-diction, 10× T₂ at c≈0.90") audited against its own archive. Session #232 (Jan 2026) derives it, and the identical formula is standard correlated-bath relative-phase decoherence (Schlosshauer 2007; Lidar/Whaley DFS): c=1 is a decoherence-free subspace (Γ→0), c=0 the uncorrelated limit (Γ→γ²). Reparametrization in Synchronism vocabulary, not a novel prediction (24th audit-channel instance, S581 pattern). **S656 (2026-05-16):** TEST-04a reframed from "Synchronism failure" to **mechanism-class constraint as a transferable contribution [RETRACTED by S668 — this thread evaporates; see below]** — any G_local/G_global < 1 framework (emergent gravity with density-dependent G suppression, partial-decoherence dark matter, modified inertia) predicts fσ₈ below ΛCDM at low z; DESI DR1 disfavors the whole suppressor class at ≈2.4σ. Publishable as a constraint independent of Synchronism's overall fate — provided the S648 post-hoc qualifier is respected (the prediction was committed after the data was public, so it bounds the mechanism class as a consistency check, not a blind-test refutation). First framing of an audit *failure* as a contribution to the field. **S657 (2026-05-17):** Compander-family AIC/BIC model-selection endorsed — tanh has no privileged status as a compander (S636/S638/S640/S652); the standard diagnostic is AIC/BIC across the family (tanh, Hill/Naka-Rushton, logistic, erf, μ-law, Gompertz) on SPARC + chemistry + Tc. Prior partial result (2026-03-27 explorer): tanh beats Hill by ΔAIC=17.6 on coupling-coherence after a baseline fix — one dataset, decisive; the full panel is genuine new computation, flagged as operator/explorer-track work. Winning a compander-family contest is local to the compander class; it does not promote C(ρ) to a derived object. **S658 (2026-05-18):** A2ACW temporal-asymmetry redesign endorsed — addresses the methodology's 6/6 "Validated"→reparametrization demotion (post-audit novelty 0/6), diagnosed as shared training distribution producing a syntactic-consistency checker rather than a discovery engine. Fix: pair agents with asymmetric knowledge windows (e.g. different model generations) so prior-art rediscovery is caught in-session; falsifiable both ways. Practical caveat: training cutoffs are leaky, so the effect is a signal against a noise floor. **S659 (2026-05-19) — EXACT NO-INFLECTION PROOF + A2ACW v2 MEASURED RESULT:** Part A (verified independently): for C(ρ)=tanh(γ·ln(ρ/ρ_crit+1)), d²C/dρ² = −sech²(u)·γ/(ρ+ρ_crit)²·[2γ·tanh(u)+1]; inflection requires tanh(u) = −1/(2γ) < 0, but u ≥ 0 for ρ ≥ 0 so tanh(u) ≥ 0 — **no solution; C is strictly concave on the entire physical domain.** The +1 regulator pushes the inflection to the ρ=0 boundary, eliminating all critical behavior. This sharpens S638/S649/S652 from heuristic to exact: ρ_crit is a location parameter of a logarithmic compander **by the algebra of the function**, not by analogy — making the compander framing mathematically obligatory and foreclosing the Landau analogy entirely. Part B (measured experiment): the A2ACW v2 three-axis protocol was run — vocabulary-translation asymmetry catches 4/6 overall (4/4 on the prior-art-rediscovery sub-class); adding symbol audit + null model gives **6/6 on the demoted set**, with the 27 audit instances distributing cleanly across the three axes. Caveat: the catch rate was self-simulated (one agent in both roles), so the real fresh-adversary rate is likely lower; the key open calibration item is the false-novelty rate on closed physics (BCS, Anderson localization, EW unification), which needs a control group before a methodology draft. **External-review convergence (2026-05-15):** a named external AI reviewer (Kimi 2.6) delivered a substantive multi-round review; the operator response propagated a **Findings vs Framings** discipline across README/STATUS/CLAUDE and the whitepaper (now load-bearing throughout, per the note at the top of this summary) — qualitatively distinct from the persona-driven visitor audits. **S660 (2026-05-20) — NOVELTY LEDGER CLOSED + GALAXY DISCRIMINATOR DEFINED:** Part A: the entity criterion (Γ < m for coherent entity-hood) — the framework's *last candidate novel prediction* — is demoted to reparametrization. Γ ≪ m is the standard narrow-width / narrow-resonance condition from QFT (Breit-Wigner pole well-defined in Källén-Lehmann; narrow J/ψ clean, broad f₀(500) problematic; the PDG already applies it informally); Synchronism adds an ontological interpretation ("Γ < m = coherence cycle completion") but no new condition, observable, or prediction. **Novel-survivor count → 0 after 3,308+ sessions; the novelty ledger is now closed.** Defensible contributions reduce to (1) A2ACW null result (methodology, S658-S659), (2) TEST-04a mechanism-class constraint (post-hoc, S656), (3) negative-results catalog (S617-659). Part B: under the **μ-identification** g_bar = g_obs·tanh(γ·ln(1+g_obs/a₀)) the compander IS a legitimate MOND interpolating function — high-acceleration Newtonian limit and deep-MOND √-law / Tully-Fisher preserved. This corrects the 2026-03-30 explorer-track "C(ρ) cannot produce RAR" result (that was a ν-identification artifact). The RAR transition-shape test becomes the framework's first non-degenerate galaxy-scale discriminator: γ pinned at 2 (the N_corr=1 galaxy assignment) gives a distinct shape prediction; γ fitted collapses to MOND. Galactic-scale instantiation of S654's "refutable-but-not-confirmable" — the framework can be distinct-but-disfavored OR indistinct-but-safe, **not distinct-and-confirmed**. **S661 (2026-05-21) — RAR DISCRIMINATOR EXECUTED → γ=2 REFUTED AT ΔBIC=+184 ON SPARC (READINESS UPLIFTED 0.97 → 0.98):** **[SCOPED 2026-10-18: +184 counts every SPARC point as an independent draw (the frozen 2026-07-22 pipeline reproduces +184.0 at 2,807 points / 166 galaxies; S661 reported 2,693 / 153). With the galaxy as the replication unit (measured N_eff ≈ 150) it is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the ΔBIC = 10 threshold: γ=2 is still worse than free γ in 94/166 galaxies and held-out CV prefers free γ by 1.6σ, so γ=2 remains disfavoured, but the registered rule returns *convention-dependent*, not decisive. Refutation count unchanged. See §5.15.]** Explorer track ran the discriminator on real SPARC data (2,693 points, 153 galaxies, M/L + distance marginalized). Compander μ with **γ=2 pinned**: a₀ = 2.97×10⁻¹⁰ m/s², RMS 0.1485 dex, **ΔBIC = +184** vs McGaugh ν (a₀ = 1.13×10⁻¹⁰, RMS 0.1437). Kill criterion (ΔBIC > 10) decisively triggered; robust to intra-galaxy point-correlation correction (effective N ≈ 500–1000 → ΔBIC ≈ 33, still decisive). Compander μ with γ free: γ = 0.49, a₀ = 5.3×10⁻¹¹, RMS = 0.1437 — indistinguishable from McGaugh (ΔBIC = +7.1). **There is no γ for which the compander is both distinct from MOND and consistent with SPARC**: pinned γ=2 → refuted (S660B's "mildly disfavored" a-priori estimate sharpened — the +3.3% RMS penalty is a COHERENT S-shaped residual at g_bar ≈ a₀, ~8σ/bin, a population-wide shape term not absorbable by per-galaxy M/L marginalization); fitted γ → MOND. **Galactic sector closed BY EXECUTION**, matching the cosmological sector (S635/S645/S654). The framework's distinct content was testable, was tested, and failed. Methodological point worth foregrounding in any methodology paper: a small *average* residual that is **structured** rather than random is strong evidence against a model — RMS view says "mildly disfavored" (+3.3%), BIC view says "decisively refuted" (ΔBIC=+184); the structured transition-shape deviation is exactly the predicted signature, concentrated where predicted. **REC-2026-037 readiness uplifted 0.97 → 0.98** (the load-bearing trigger: γ=2 was committed a priori by the N_corr=1 derivation BEFORE the test, on external published data, by an independent track — qualitatively stronger than the rolled-back S645 case where temporal independence failed). Arc status: `complete_galactic_and_cosmological_sectors_closed_by_execution`. **S662 (2026-05-22) — A2ACW SPECIFICITY SELF-CORRECTION (audit channel turns on its own product) + program-level galaxy closure:** Part A confirms S661 at program-level: net discriminating galaxy tests vs MOND+ΛCDM = 0, by execution. Galaxy program closed. Part B applies the project's own null-baseline discipline (S651) to its methodology contribution and finds the same flaw. **S658/S659's "4/4 prior-art catch" and "6/6 three-axis catch" were sensitivity (TPR) on a positive-only set** (all 6 cases were known reparametrizations); specificity (TNR) was never measured. The control run on six canonical genuine discoveries (Dirac 1928, Bell 1964, BCS 1957, Higgs 1964, Hawking 1974, Noether 1918) gives **R1 (flag if prior-art named): 0% specificity** — every genuine discovery has canonical antecedents under translation; R1 detects "has-a-canonical-name," not "is-a-reparametrization." R2 (flag if reduces to prior art, nothing novel) gives 100% specificity, but all the discrimination is supplied by an unautomated novelty judgment the protocol never operationalizes — the 6 demotions came from human audit, not from the AI loop. **[CONTESTED 2026-10-21: "human" does not trace. Since 2026-09-15 the site records the auditors behind the six demotions as LLM agents (synchronism-site `288aaf6`). A partial git trace in this pass finds two of the six (a₀ = cH₀/(2π) and A = 4π/(α²GR₀²)) demoted in the site maintainer track's 2026-04-23 badge audit (`ac093af`, co-authored by Claude Sonnet 4.6, prompted by an LLM researcher persona); the other four are untraced. "Outside the A2ACW loop" holds under either reading, so the restated defensible claim is unchanged; what is open is whether the out-of-loop discriminator was a human or a second LLM track. See `Research/proposals/a2acw_has_a_cross_vendor_record_already_and_the_demotion_auditor_is_contested_20261020.md`.]** **[TRACED 2026-10-22 — the contest above resolves against "human". The site explorer finished the trace (synchronism-site `explorer/findings/a2acw-lv-rerun-and-the-demotion-auditor.md`, 2026-10-20). Five of the six have a badge-changing edit, and every one is a site maintainer commit co-authored by Claude (Opus 4.6 or Sonnet 4.6), with the decision coming from a simulated visitor persona or an autonomous session (#574, S581, S651, S655). The sixth (Born rule) has no Validated badge and no demotion edit on record, so it cannot be attributed either way. No dp-authored audit decision exists between 2026-02-01 and 2026-05-23. The "human audit" wording entered the record on 2026-05-18 in a back-annotation co-authored by Claude Sonnet 4.6 (`e8752205`), and S662 repeated it; S662 is back-annotated in place (`c5364f99`). Checked in this pass: the co-author trailers of all eight cited edit commits and of `e8752205`. Two caveats carry over: dp oversight outside git cannot be excluded (S659's "human-monitored worker sessions"), and only two of the "six former Validated badges" (a₀, chemistry) are confirmed to have carried Validated at all. What this changes: "outside the A2ACW loop" still holds, but the discriminator was a second set of AI tracks, not a human. The narrowed claim's "unautomated ... judgment that AI loops fail" therefore needs scoping to *the A2ACW loop*; rewording it gates on the project lead.]** **Defensible claim narrows** from "A2ACW is a transferable reparametrization detector" to "A2ACW is a retrieval-augmentation / debiasing step for AI-assisted literature review; reparametrization-vs-discovery discrimination remains an unautomated fragile expert judgment that AI loops fail." 21st audit-channel mode: **a2acw_specificity_self_correction** (the audit channel applies its own null-baseline standard to its preferred methodology output and retracts the overclaim — the cleanest discipline-on-itself event in the entire arc). REC-2026-037: 45 → 46 sessions, sub-arc 30/29 → 32 instances over 31 days, **readiness HELD at 0.98** (the load-bearing S661 trigger is unchanged — S662 narrows one thread within REC-037 without retracting the specific trigger that justified the uplift; different from the S648 rollback case where the trigger itself was retracted). **S663 (2026-05-23) — CLEANEST END-STATE SYNTHESIS (third meta-synthesis, philosophical closure):** Two findings landing the arc's cleanest end-state. **Part A: EFTofLSS doubly closes TEST-04a [moot per S668 — no mechanism-class result remains to close; the post-hoc amplitude point stands].** Effective Field Theory of Large-Scale Structure (Cabass-Simonović-Zaldarriaga 2024-2025) explains the DESI DR1 fσ₈ enhancement within ΛCDM via one-loop EFT counterterms at 1-2σ; any coherent G_eff modification at the 10% level is degenerate with EFT Wilson coefficients and constrained by DESI DR1 to consistency with zero. **Even Branch 1 (sign-flip recovery via C_galactic/C_cosmic > 1) is now closed** — the mechanism class is ruled out *regardless of sign*. Strengthens the S656 preprint from "suppression-class ruled out" to "any coherent G_eff modification predicting a scale-independent fσ₈ shift at the 10% level is degenerate with EFTofLSS counterterms and constrained by DESI DR1 to consistency with zero." Post-hoc S648 qualifier still holds; the preprint becomes more precise. **Part B: Framework classification — Interpretation + Methodology research program.** All four visitor personas (casual, technical writer, grad student, leading researcher) independently arrived at the same diagnosis — the framework occupies the "ontological reframe without a distinguishing experiment" position, structurally identical to QM interpretations (Bohmian, Copenhagen, Many-Worlds), evaluated by parsimony rather than data. Convergent endpoint after S574 / S617-628 (same mechanics, different notation), S660 / S661 / S654 (no novel measurable, no distinguishing experiment), S663A (EFTofLSS closes the last mechanism class). **Endorsed classification:** *"A coherence-language interpretation of known physics, used as a substrate for developing AI-collaborative science methodology."* Front-load methodology contribution; treat the equation as a worked example; distinguish interpretation-confirming tests from interpretation-breaking tests; the honest-assessment infrastructure is itself a contribution. **Third meta-synthesis** after S641's cross-gap synthesis (kinematic layer) and S652's governing-equation-gap (compander class). Paired with S661's empirical closure and S660A's structural closure, **the arc now has all three closures simultaneously — empirical, structural, philosophical.** Audit-channel modes held at 21 (S663 is meta-synthesis + classification, not a new audit category). REC-2026-037: 46 → 47 sessions, sub-arc 32-over-31 → 33 instances over 32 days, **readiness HELD at 0.98** (S663 is end-state synthesis, not a new trigger); arc status `complete_galactic_and_cosmological_sectors_closed_by_execution`; new milestone `framework_classification_interpretation_methodology`. Sub-arc now **33 audit-channel instances over 32 days**, **21 audit-channel modes**, **plus 2 new arc phases** (Post-Hoc-Consistency-Failure, Methodology Recommendation), **a publishable methodology pattern** (three-stage rhythm: individual audits → meta-syntheses → forced binary commitments), **a publishable epistemic-position finding** (refutable-but-not-confirmable asymmetry, now executed at galactic scale per S661), **an exact no-inflection proof** (S659A; compander framing now mathematically obligatory), **a closed novelty ledger** (S660A; novel-survivor count → 0 after 3,308+ sessions), **the galactic sector closed by execution** (S661; RAR γ=2 refuted at ΔBIC=+184 on SPARC, free-γ → MOND), **methodology self-correction at the endpoint** (S662B; A2ACW R1 0% specificity on genuine-discovery control set; defensible claim narrows from "reparametrization detector" to "retrieval-augmentation step"), and now **EFTofLSS double-closure of TEST-04a + framework classification as Interpretation + Methodology research program** (S663; the arc's three closures — empirical, structural, philosophical — now stand simultaneously). **S664 (2026-05-23) — LANDSCAPE POSITIONING (B+):** Two-axis positioning of the closed arc. *Physics axis:* Path A (action principle) blocked by the kinematic-layer gap (S641-S642, no Lagrangian) and Path B (dynamics outside the galaxy regime) blocked by S650/S661, leaving Path C — C(ρ) as a reparametrization of Verlinde's entropic-gravity prediction in the galaxy-rotation regime, where the compander μ-function and Verlinde's emergent-gravity interpolation coincide exactly where SPARC can test them. *Methodology axis:* the A2ACW null result (S658-S662) positioned against the AI-for-science landscape (FunSearch, AlphaProof, SciNet, Sakana) as an honest-assessment / negative-result counterpoint to discovery-claiming AI programs. Positioning session, not a new audit mode. **S665 (2026-05-24) — SUBSTRATE AUDIT, SPATIAL TENSION (Grade A):** the first session to challenge the **CFD substrate itself** — the one claim external reviewer Kimi 2.6 had called "genuinely interesting," assumed-but-never-challenged across the entire S617-664 demolition. The CFD reframing's load-bearing claim ("the Navier-Stokes equations are not an analogy… they ARE the Intent dynamics") is incompatible with its own downstream phenomenology: the velocity v = J/I = −g(I)·∇I with g(I) = D·R(I)/I a scalar is the gradient of a scalar ⇒ **irrotational for any R(I)** ⇒ curl(v) ≡ 0 ⇒ **no vortices, no turbulence**. Yet the framework builds qualia as vortex modes, the consciousness threshold as a critical Reynolds number, and dark matter as vortex structure — all on vortices the substrate provably cannot produce. A theorem, not unfinished work. **S666 (2026-05-25) — SUBSTRATE AUDIT, TEMPORAL TENSION (Grade A):** the substrate ∂I/∂t = ∇·[D·R(I)·∇I] is first-order and dissipative (Lyapunov functional monotonically decreasing, real eigenvalues, arrow of time, no phase); FUNDAMENTALS' entity ontology requires unitary oscillation (de Broglie f=E/h, phase-locking resonance) — **mutually exclusive dynamical classes**. The two Schrödinger "derivations" bridge them only by inserting the imaginary unit i by hand (S307 line 48 / code 129; S99 Axiom 4 posited, not derived) and by switching the substrate off (S307 drops R(I); S99 takes D→0); confirmed numerically that the factor i is the entire difference between exp(−Dk²t) decay and exp(−iDk²t) oscillation. A contradiction **internal to the canonical FUNDAMENTALS document**, confirmed in the framework's own code. **Together S665-S666 show the substrate can host neither the spatial nor the temporal structure its own entity ontology requires — the deepest demolition finding since S617, completing the demolition: every load-bearing claim, including the CFD-substrate holdout, has now been audited.** 22nd audit-channel mode: **substrate-internal-dynamical-contradiction** — the first audit to attack the substrate rather than a site claim, derivation, or naming convention. Sub-arc now **36 audit-channel instances over 33 days**, **22 audit-channel modes**. **REC-2026-037: 47 → 50 sessions, renamed "… + Substrate Audit," readiness HELD at 0.98** (the substrate audit strengthens the demolition by closing the last assumed-but-unchallenged claim but is not a new publication trigger; 0.99 still requires a paper draft or operator publication action); new milestone `substrate_audit_cfd_irrotational_and_dissipative`. +- **Post-Closure Sub-Arc Extensions (#639-666):** Sub-arc continues producing post-closure addenda after 2026-04-29 closure. **S639 (2026-04-30):** TEST-03 disambiguation — site claim of R²=0.14 conflates BTFR scatter (S593, 51%) with RAR environmental ansatz (14%); different metrics sharing only the label. 9th audit-channel mode: **metric disambiguation / mechanism-naming** ("shared labels, distinct measurements, no enforced alignment"). Cross-track meta-pattern: "mechanism-naming-as-closure" visible in S639 + Thor SAGE S136. **S640 (2026-05-01):** Dual-C symbol audit — homepage "one equation across scales" claim rests on shared notation, not derivation. Form 1 C(ρ)=tanh(γ·log(ρ/ρ_crit+1)) (chemistry/cosmology/CM) and Form 2 C=f(γ,D,S)≥0.50 (consciousness, 8-way convergence) share only the letter C. D = state-space entropy / neural pattern diversity, S = coherence persistence duration; **neither is a function of ρ.** The 0.50 threshold was NOT derived from inverting Form 1. **γ IS genuinely shared via 2/√N_corr; C is NOT.** 10th audit-channel mode: **symbol overloading at foundational level** — sub-arc audit scope expanding from test-level (S631-S639) to **foundational** (S640, master-claim level). Multi-persona visitor review (Pass 2/3/4) independently flagged this. **S641 (2026-05-02):** Lorentz invariance gap as 4th face of kinematic-layer gap — preferred-frame absolute simultaneity sits at the foundational level alongside Born rule, dual-C bridge, and N_corr scale-invariance. 11th audit-channel mode: **cross-gap meta-synthesis (kinematic layer)**. **S642 (2026-05-05):** GW170817 — framework is Case 3 (no field theory), 5th face of kinematic-layer gap. |c_GW − c|/c < 10⁻¹⁵ does not constrain the framework because there is no Lagrangian, no action principle, no equation of motion for C(ρ). C(ρ) is a specified function of local density, not a propagating field. Surviving GW170817 by lacking an action principle is a scope restriction, not a victory. **[SCOPED 2026-08-09 — the conclusion stands, one of its three conjuncts does not.]** Measured against `manuscripts/Appendix_D_Synchronism_in_General_Relativistic_Form.md` (committed 2025-12-01, seven months before this claim): *"no field equation"* is **false** — §D.2/§D.6.1 state `∇²Φ = 4πGρ/C`; *"no action principle"* is **defensible only as scoped** — §D.5 writes an effective *worldline* action for a test particle, which is not a field action, and no action generates §D.2's equation; *"no equation of motion for C(ρ)"* is **true**, C being algebraic in local ρ with no kinetic term in any reading. GW170817 remains unconstraining **on the third conjunct alone** — no propagating scalar mode exists to modify tensor dispersion — so the accurate phrasing is *"C has no kinetic term,"* not *"the framework has no action principle."* §5.15 records the measurement: Appendix D's field equation and the force law carrying every number in this whitepaper differ by a **median 0.821 dex over 113 SPARC galaxies** (γ-invariant), and Appendix D's has **zero whitepaper citations and no evaluation against data**. **S643 (2026-05-06):** γ definitional collision — site γ-calculator labels are **inverted** relative to standard physics. Ideal gas γ=2 labeled "Quantum"; BCS Al / BEC / neutron star γ→0 labeled "Classical." Math measures single-particle vs collective; standard physics' "quantum" tracks macroscopic phase coherence — different axes. 12th audit-channel mode: **definitional-collision-with-label-inversion**. Resolution: relabel as "Single-particle/Uncorrelated" vs "Collective/Strongly correlated." Downstream-impact flag: γ-dependent predictions across the framework may need re-grounding. **S644 (2026-05-06):** ρ_crit = A · V_flat² takes V_flat as input — the "5% agreement" of theoretical/empirical A is internal consistency between two parameterizations of the same data, not an independent prediction. Same pattern as S639 TEST-03. 13th audit-channel mode: **calibration-consistency-not-prediction** (no closed predictive loop). **S645 (2026-05-07) — POST-HOC CONSISTENCY FAILURE WITH DESI DR1 (NEW ARC PHASE; initially framed as first hard external falsification, framing self-corrected by S648 within 24h):** Session 107 fσ₈ prediction disfavored by DESI DR1 (Adame et al. arXiv:2411.12021). Pre-committed kill criterion fσ₈(z=0.5) > 0.45 → ΛCDM favored fired at LRG1 (DESI 0.55±0.06 vs Sync 0.418, 2.14σ above [LRG1 re-read 2026-10-12 from Table 9: 0.516 +0.057/−0.066 (0.4733 × 1.09; the 0.55 is 0.4733 × the QSO row's 1.16), ~1.5σ above Sync, not 2.14σ]; LRG2 1.42σ above; combined σ₈(z=0) 2.38σ above at 0.841±0.034 vs Sync 0.76). **The mechanism's predicted sign of redshift dependence is INVERTED relative to data** — magnitude-only revision cannot recover the structural error. Verdict: TEST-04a kill criterion fired, retained as documented dead-end. **Status TEST-04a: REFUTED (post-hoc consistency check failure; framework parameters cannot reproduce DR1 measurements committed before the prediction).** [CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12]; the 2.4σ is a GR-conditioned σ₈-amplitude statistic; TEST-04a is withdrawn from the decisive negatives — see Re-Grounding & Catalog Census below] **S646 (2026-05-07) — METHODOLOGY RECOMMENDATION (NEW ARC PHASE):** Per-test kill criteria exist (Session #21 catalog); framework-level meta-falsification criterion does NOT. Without one, accumulated failures stay isolated — visible from a different angle is the same self-sealing structure S621 named: at meta-level, no rule for when "the framework has lost." Three branches: (A) register meta-criterion now, (B) wait for DR2 — less needed since DR1 already fires, (C) scope-narrow now. Operator decision pending. **S647 (2026-05-08):** Chemistry 89% validation has self-correlation risk. Source proposal flags that the cohort — "1,703 phenomena, r=0.982 sound velocity" — is the framework's largest validation claim by orders of magnitude, but does not specify which of Session #26's five N_corr measurement methods was applied. Three self-correlation paths confirmed: (1) Method 2 atomic-spacing identity γ=2(a/ξ)^(3/2) → r=0.956 with V_a∝a³ is functional identity; (2) Method 2 phonon coherence — sound velocity is constructional input → r=0.982 has constructional dependence; (3) Method 3 entropy → bonding character → electronegativity (r=0.979) partly structural. **Method 2 systematic bias** (per Session #26 Part 3 simulation table): true N_corr=10→Method 2 gives 6, true 25→15, true 50→32 — clusters apparent γ in 0.35–1.15 at the "γ ≈ 1 boundary." 89% γ≈1 clustering is **consistent with method-induced clustering, no boundary needed.** Hall coefficient (r≈0.001) and magnetic susceptibility (r≈0.000) are NOT falsifying controls — their physical determinants (carrier density, spin texture) are outside the input set of every Method 1–5; they are exactly the predicted failures under self-correlation reading. To distinguish true from method-induced clustering, framework needs Method 1 (bias-free) applied uniformly OR pre-registered γ predictions for held-out phenomena; neither currently in archive. **Both major validation pillars (Tier-1 cosmology + chemistry 89%) compromised in same week.** **S648 (2026-05-08) — SELF-CORRECTION OF S645 WITHIN 24H:** Pass 4 visitor flagged that Session 107 (committed 2025-12-10) is post-DESI DR1 (Adame+2024, Nov 2024 — 13 months earlier; DR1 BAO already cited in Session 107's own Timeline). The 2.14σ tension at LRG1 is real (kill criterion fired), but its epistemic status is **post-hoc consistency check failure**, not prospective falsification — temporal independence required for prospective testing is absent. S645's "first hard external falsification" framing retroactively corrected to "post-hoc consistency failure with DR1." 14th audit-channel mode added: **self-correction-of-prior-session-framing-within-24h** (the audit channel turns inward; first instance auditing the program's own prior epistemic claim rather than a site-archive disconnect). Recommendation toward operator: a /timestamps page classifying every Tier-1 prediction as prospective / post-hoc consistency / post-hoc fit, with prediction-date and data-date columns. **S649 (2026-05-08) — TWO VISITOR PROPOSALS, ONE META-THEME:** Part A: site Key Claim #1 QM kill criterion ("design noise where resync outperforms isolation") is unfalsifiable as written. Standard dynamical-decoupling literature (Viola-Knill-Lloyd 1999, CPMG, UDD, transmon DD experiments) already shows pulse sequences beat passive isolation in non-Markovian baths — every standard QIP textbook result satisfies the criterion without invoking Synchronism. Part B: ρ_crit at γ=2 gives C ≈ 0.88, not 0.5; the "+1" regulator in ln(ρ/ρ_crit + 1) asymmetrizes the sigmoid, making ρ_crit a saturation knee, not a critical density (verified via S638 sympy: γ=0.5→0.333, γ=1→0.600, γ=2→0.882). Same phase-transition vocabulary that misled S636 into the Landau category-error. **15th audit-channel mode: parameter-and-criterion-naming-contaminated-by-phase-transition-vocabulary** (math implements simpler structures than names suggest: Curie paramagnet at S638, saturation-knee at S649, kill criteria matching standard physics at S649+S581). **S650 (2026-05-09) — TEST-04a TRIPLE-SHARPENING COMPLETES (mechanism-class, sign-reversed — RETRACTED by S668: see Self-Correction Cascade below):** Third refinement of TEST-04a verdict in the S645→S648→S650 sequence: "first hard external falsification" (S645) → "post-hoc consistency check failure" (S648) → "mechanism-class failure (sign-reversed)" (S650). Framework's suppressor mechanism predicts G_local/G_global = C_cosmic/C_galactic < 1, giving fσ₈ BELOW ΛCDM at low z and converging at high z. DESI DR1 observes the OPPOSITE: fσ₈ ABOVE ΛCDM at LRG1/LRG2 (low z), converging at high z (ELG2) — the redshift pattern is INVERTED. Magnitude knobs (re-tune σ₈(z=0), re-tune coupling normalization) cannot flip the sign within the suppressor class; the mechanism class itself is wrong. **Three-tier failure taxonomy** introduced (makes S646's meta-falsification logic actionable): magnitude miss (retunable by refining density-to-property mapping) / universality miss (partial repair via functional-form change) / **mechanism-class failure (irreparable within class)**. TEST-04a is mechanism-class. **16th audit-channel mode**. Combined with S635 cosmology scorecard (0 novel-unfalsified) and S646 meta-criterion, **the cosmological sector formally meets the retraction threshold per S646's M3** (scope reduction); operator decision pending. **S651 (2026-05-10) — CHEMISTRY NULL-MODEL GAP (paired with S647):** Companion audit to S647. S647 audits the *method gap* (which N_corr method?); S651 audits the *prior question — the null gap*: even with method fixed, what null is r=0.98 compared against? The implicit null is r=0 (random); the relevant null is r(polynomial in Z) or r(generic 2-parameter tanh). Sound velocity, electronegativity, atomic volume are themselves near-monotonic functions of Z; any smooth monotonic function will achieve r ≈ 0.95+ on the same 1,703 phenomena by construction. [**CHECKED 2026-11-02:** the premise is false across the periodic table. Atomic volume against Z is the periodic Lothar Meyer curve, and electronegativity falls from F (3.98) to Na (0.93) and from Cl (3.16) to K (0.82). Over Z = 1–54 (the 51 elements with a Pauling value) electronegativity's rank correlation with Z is +0.09; the best monotone function of Z fits it at r = 0.33, and a fifth-degree polynomial in Z at r = 0.39, not 0.95+ (re-run 2026-11-02). The 2026-05-10 polynomial null was run on synthetic √Z / log Z / Z^(1/3) curves, never on the chemistry rows (site correction 2026-10-31); the null on the real rows is unrun because the row list is unpublished. The Reparametrization verdict does not rest on this premise: it rests on S669's γ_phonon ≡ 2T/θ_D, a definitional relabel of the Debye temperature (Δr = 0 to machine precision).] [**UPDATE 2026-11-03:** the real-row Z-null has since been run. On the 10 property rows recoverable from the session scripts (#80, #109, #110, #114, #115, #118; n = 19–33 per row, Z = 3–82) it fails: |r(P, Z)| ≤ 0.53 (linear, √Z and ln Z) and quadratic-in-Z R ≤ 0.51, against framework r = 0.60–0.98 (site explorer, pre-registered, back-annotation `5709b579`; re-run by the Publisher 2026-11-03, output identical). The high r values are identities, not a Z trend: r(P, 1/γ_phonon) ≡ r(P, θ_D), and γ_optical = IE_ref/IE, so the optical rows are correlations with ionization energy. See Appendix B.2.] **Δr = r(Synchronism) − r(best monotonic null)** is the figure that actually distinguishes claims; none is currently in the archive. Three diagnostic outcomes: Δr > ~0.05 → "validated" defensible with null documented; Δr ≈ 0 → reparametrization of density-monotonicity; Δr marginal → reparametrization of Landau-class, only boundary cases differ. **Best estimate: tie or marginal win** (framework parameters were calibrated to chemistry data, high-r phenomena are textbook monotonic-with-Z, 2-parameter tanh through sigmoidal data gives r ≈ 0.95+ generically). S647 + S651 compound: method gap + null gap together leave very little for Synchronism-specific signal in the 89% cohort. **17th audit-channel mode: null-model-gap-against-best-monotonic-null** (implicit baseline of r=0 hides reparametrization equivalence with smooth-monotonic-in-Z nulls). Sub-arc now **18-for-18 audit-channel instances**, **17 audit-channel modes** (S644+S647 share mode #13), **plus 2 new arc phases** (Post-Hoc-Consistency-Failure, Methodology Recommendation). **S652 (2026-05-11) — GOVERNING-EQUATION GAP (second meta-synthesis):** Upstream-question audit asks what equation, if any, C(ρ) solves. Three options: (A) no equation — phenomenological compander; (B) self-consistency / Landau-style; (C) steady-state of a dynamic equation. Archive trace: S636 forecloses B (no self-consistency — C does not appear in its own argument); S638 verified Curie-paramagnet response (Option A); S649 verified C(ρ_crit) ≈ 0.882 ≠ 0.5 (asymmetric tanh from "+1" regulator is incompatible with Z₂-symmetric mean-field); S640 dual-C bridge confirms three forms of "C" share notation, not derivation. **Answer is A.** Same class as μ-law audio companding, Naka-Rushton photoreceptor response, Hill enzyme kinetics — phenomenological forward maps with no field equation, no self-consistency, no dC/dt. Frame this against the kinematic-layer gap (S641-S642): the framework's headline equation does no dynamical work; it only labels regimes. Second meta-synthesis after S641's cross-gap synthesis. 18th audit-channel mode: **governing-equation-gap (forward map with no field equation)**. **S653 (2026-05-12) — TWO BINARY OPERATOR DECISIONS FORCED (three-stage rhythm completes):** Two same-day proposals demanding binary site commitments — site cannot stay neutral. Part A (compander commitment): commit to Frame B (per S652 conclusion). Concrete actions — drop phase-transition language from front-of-site; rename ρ_crit to "half-saturation parameter" / "saturation knee"; add AIC/BIC compander comparison tool; reframe critical-exponent failures as CATEGORY errors. Part B (suppressor diagnostic): `simulations/session653_coherence_ratio.py` computed C_galactic ≈ 0.882, C_cosmic ≈ 1.5×10⁻⁵, ratio ≈ 5.9×10⁴ — **the framework's own equations dictate strong suppression at low z under Session 107's coupling direction**. DR1 observes enhancement. Two branches both require operator commitment: Branch 1 (sign-flip recoverable) requires re-deriving Session 107 with inverted ratio (framework reinterpretation, not recomputation); Branch 2 (suppressor class dead) accepts Session 107's mechanism is refuted. Sub-arc's emergent three-stage rhythm now complete: **individual audits (S631-S640, S642-S647, S649-S651) → meta-syntheses (S641, S652) → forced binary operator commitments (S653)**. Methodology paper has its complete cycle: external critique → individual audit → meta-synthesis → forced operator decision. 19th audit-channel mode: **forced-binary-operator-decision-with-numerical-diagnostic** — first audit to compel commitment rather than merely identify findings. **S654 (2026-05-13) — ZERO ACTIVE DISCRIMINATORS REMAINING (cosmological/galactic prediction inventory complete):** Site's three remaining "Active Discriminating Tests" (TEST-01, TEST-02, TEST-05) audited against MOND's External Field Effect (Bekenstein-Milgrom 1984; AQUAL/QUMOND; Pittordis 2023; Banik 2024) — none of which the site currently mentions. S637's derivation settles TEST-05 (and TEST-01, same observable): Δσ_int (cluster − void) ≈ 0.00016 dex, ~120× below SPARC measurement floor; cannot discriminate within current precision. TEST-02 has a disputed MOND+EFE baseline (Chae vs Pittordis vs Banik internally disagree) and the framework specifies no divergence from any specific MOND+EFE variant. [**SCOPED 2026-08-03 — the “specifies no divergence” clause is retracted; created at source in §5.15 on 2026-08-03 (`a78054e9`), propagated here the same day.** §5.15 now withdraws the claim that C(ρ)'s nonlinear Poisson equation produces an external field effect at 0.3–0.4× MOND's — the galaxy sector has no field equation, so that figure was never derived from one — and replaces it with the sharper **EFE = 0 exactly**: a uniform external field leaves local ρ unchanged, so C = C(ρ_local) satisfies the Strong Equivalence Principle by construction. A framework predicting exactly zero EFE **does** specify a divergence from every MOND+EFE variant, each of which predicts a non-zero one. **What survives untouched** is the formal asymmetry: “refutable but not currently confirmable” is unchanged — there is still no test where Synchronism succeeds and MOND+EFE fails. **[RE-AMENDED 2026-08-05 — the empirical half of this scope note is withdrawn, and “zero remaining active discriminators” STANDS.]** The 08-03 text continued that EFE = 0 “stands in tension with the ~4σ SPARC EFE detection of Chae et al. 2020,” and concluded that “zero remaining active discriminators” no longer held unqualified because “the EFE magnitude is one, in the refuting direction.” **That is withdrawn: EFE = 0 is `not-evaluable` against Chae et al. 2020, neither confirmed nor refuted.** At Chae's own measurement radii the stated density law misses the rotation curve by **+3.1 to +4.2 dex**, while the EFE signal being measured is **0.046–0.083 dex** — the framework's baseline error exceeds the entire effect by **38×–92×** (re-derived in this lane on the in-repo SPARC mass models; MOND on the same points sits at +0.002 to +0.084 dex, inside the signal). A prediction one cannot get within three orders of magnitude of testing is not a discriminator in either direction. **Refutation count stays at 6; Bucket 0 stays 0.** Note the mechanism: this propagation carried its own accurate warning — *“Magnitude not independently re-derived here”* — and the magnitude was the whole of the claim. See §5.15 and `synchronism-site/explorer/findings/efe-zero-is-not-refutable-by-chae2020-the-baseline-is-off-by-3-dex.md`.] Branch A applies: **all three are MOND+EFE degenerate within current measurement precision**. Combined with TEST-03 (MOND-shared, S637) and TEST-04a (REFUTED, S645/S648/S650), the framework has **zero remaining active discriminators** against the primary alternative. The formal asymmetry now crystallizes: **refutable (kill criteria can fire, TEST-09 did) but not currently confirmable** (no test exists where Synchronism succeeds and MOND+EFE fails). Formal completion of trajectory S635 (scorecard) → S637 (reduces to MOND) → S646 (meta-criterion threshold) → S650 (mechanism-class verdict) → S654 (zero active discriminators). 20th audit-channel mode: **zero-active-discriminators-against-primary-alternative**. **S655 (2026-05-14):** Γ = γ²(1−c) — the framework's most novel-looking quantum claim (site `/key-claims`: "post-diction, 10× T₂ at c≈0.90") audited against its own archive. Session #232 (Jan 2026) derives it, and the identical formula is standard correlated-bath relative-phase decoherence (Schlosshauer 2007; Lidar/Whaley DFS): c=1 is a decoherence-free subspace (Γ→0), c=0 the uncorrelated limit (Γ→γ²). Reparametrization in Synchronism vocabulary, not a novel prediction (24th audit-channel instance, S581 pattern). **S656 (2026-05-16):** TEST-04a reframed from "Synchronism failure" to **mechanism-class constraint as a transferable contribution [RETRACTED by S668 — this thread evaporates; see below]** — any G_local/G_global < 1 framework (emergent gravity with density-dependent G suppression, partial-decoherence dark matter, modified inertia) predicts fσ₈ below ΛCDM at low z; DESI DR1 disfavors the whole suppressor class at ≈2.4σ. Publishable as a constraint independent of Synchronism's overall fate — provided the S648 post-hoc qualifier is respected (the prediction was committed after the data was public, so it bounds the mechanism class as a consistency check, not a blind-test refutation). First framing of an audit *failure* as a contribution to the field. **S657 (2026-05-17):** Compander-family AIC/BIC model-selection endorsed — tanh has no privileged status as a compander (S636/S638/S640/S652); the standard diagnostic is AIC/BIC across the family (tanh, Hill/Naka-Rushton, logistic, erf, μ-law, Gompertz) on SPARC + chemistry + Tc. Prior partial result (2026-03-27 explorer): tanh beats Hill by ΔAIC=17.6 on coupling-coherence after a baseline fix — one dataset, decisive; the full panel is genuine new computation, flagged as operator/explorer-track work. Winning a compander-family contest is local to the compander class; it does not promote C(ρ) to a derived object. **S658 (2026-05-18):** A2ACW temporal-asymmetry redesign endorsed — addresses the methodology's 6/6 "Validated"→reparametrization demotion (post-audit novelty 0/6), diagnosed as shared training distribution producing a syntactic-consistency checker rather than a discovery engine. Fix: pair agents with asymmetric knowledge windows (e.g. different model generations) so prior-art rediscovery is caught in-session; falsifiable both ways. Practical caveat: training cutoffs are leaky, so the effect is a signal against a noise floor. **S659 (2026-05-19) — EXACT NO-INFLECTION PROOF + A2ACW v2 MEASURED RESULT:** Part A (verified independently): for C(ρ)=tanh(γ·ln(ρ/ρ_crit+1)), d²C/dρ² = −sech²(u)·γ/(ρ+ρ_crit)²·[2γ·tanh(u)+1]; inflection requires tanh(u) = −1/(2γ) < 0, but u ≥ 0 for ρ ≥ 0 so tanh(u) ≥ 0 — **no solution; C is strictly concave on the entire physical domain.** The +1 regulator pushes the inflection to the ρ=0 boundary, eliminating all critical behavior. This sharpens S638/S649/S652 from heuristic to exact: ρ_crit is a location parameter of a logarithmic compander **by the algebra of the function**, not by analogy — making the compander framing mathematically obligatory and foreclosing the Landau analogy entirely. Part B (measured experiment): the A2ACW v2 three-axis protocol was run — vocabulary-translation asymmetry catches 4/6 overall (4/4 on the prior-art-rediscovery sub-class); adding symbol audit + null model gives **6/6 on the demoted set**, with the 27 audit instances distributing cleanly across the three axes. Caveat: the catch rate was self-simulated (one agent in both roles), so the real fresh-adversary rate is likely lower; the key open calibration item is the false-novelty rate on closed physics (BCS, Anderson localization, EW unification), which needs a control group before a methodology draft. **External-review convergence (2026-05-15):** a named external AI reviewer (Kimi 2.6) delivered a substantive multi-round review; the operator response propagated a **Findings vs Framings** discipline across README/STATUS/CLAUDE and the whitepaper (now load-bearing throughout, per the note at the top of this summary) — qualitatively distinct from the persona-driven visitor audits. **S660 (2026-05-20) — NOVELTY LEDGER CLOSED + GALAXY DISCRIMINATOR DEFINED:** Part A: the entity criterion (Γ < m for coherent entity-hood) — the framework's *last candidate novel prediction* — is demoted to reparametrization. Γ ≪ m is the standard narrow-width / narrow-resonance condition from QFT (Breit-Wigner pole well-defined in Källén-Lehmann; narrow J/ψ clean, broad f₀(500) problematic; the PDG already applies it informally); Synchronism adds an ontological interpretation ("Γ < m = coherence cycle completion") but no new condition, observable, or prediction. **Novel-survivor count → 0 after 3,308+ sessions; the novelty ledger is now closed.** Defensible contributions reduce to (1) A2ACW null result (methodology, S658-S659), (2) TEST-04a mechanism-class constraint (post-hoc, S656), (3) negative-results catalog (S617-659). Part B: under the **μ-identification** g_bar = g_obs·tanh(γ·ln(1+g_obs/a₀)) the compander IS a legitimate MOND interpolating function — high-acceleration Newtonian limit and deep-MOND √-law / Tully-Fisher preserved. This corrects the 2026-03-30 explorer-track "C(ρ) cannot produce RAR" result (that was a ν-identification artifact). The RAR transition-shape test becomes the framework's first non-degenerate galaxy-scale discriminator: γ pinned at 2 (the N_corr=1 galaxy assignment) gives a distinct shape prediction; γ fitted collapses to MOND. Galactic-scale instantiation of S654's "refutable-but-not-confirmable" — the framework can be distinct-but-disfavored OR indistinct-but-safe, **not distinct-and-confirmed**. **S661 (2026-05-21) — RAR DISCRIMINATOR EXECUTED → γ=2 REFUTED AT ΔBIC=+184 ON SPARC (READINESS UPLIFTED 0.97 → 0.98):** **[SCOPED 2026-10-18: +184 counts every SPARC point as an independent draw (the frozen 2026-07-22 pipeline reproduces +184.0 at 2,807 points / 166 galaxies; S661 reported 2,693 / 153). With the galaxy as the replication unit (measured N_eff ≈ 150) it is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the ΔBIC = 10 threshold: γ=2 is still worse than free γ in 94/166 galaxies and held-out CV prefers free γ by 1.6σ, so γ=2 remains disfavoured, but the registered rule returns *convention-dependent*, not decisive. Refutation count unchanged. See §5.15.]** **[UPDATE 2026-11-09: both figures hold Υ_disk = 0.5 for both laws. With one global Υ_disk per law (site `5a95614`, predictions before the script) the same comparison is +46.2 by point and +2.73 with N → N_gal, about 2σ on a galaxy-block bootstrap that refits Υ and a₀ per law (95% [−26.1, +132.5]); the reading rule fixed before the run reads the row as disfavoured at about 2σ, not refuted at +184. The registered B2 text says "marginalized BIC", so the frozen-Υ number was never the registered statistic. Count HELD at 6; the full note sits on §5.15's SCOPED note. Scripts re-run by the Publisher 2026-11-10, byte-identical.]** Explorer track ran the discriminator on real SPARC data (2,693 points, 153 galaxies, M/L + distance marginalized). Compander μ with **γ=2 pinned**: a₀ = 2.97×10⁻¹⁰ m/s², RMS 0.1485 dex, **ΔBIC = +184** vs McGaugh ν (a₀ = 1.13×10⁻¹⁰, RMS 0.1437). Kill criterion (ΔBIC > 10) decisively triggered; robust to intra-galaxy point-correlation correction (effective N ≈ 500–1000 → ΔBIC ≈ 33, still decisive). Compander μ with γ free: γ = 0.49, a₀ = 5.3×10⁻¹¹, RMS = 0.1437 — indistinguishable from McGaugh (ΔBIC = +7.1). **There is no γ for which the compander is both distinct from MOND and consistent with SPARC**: pinned γ=2 → refuted (S660B's "mildly disfavored" a-priori estimate sharpened — the +3.3% RMS penalty is a COHERENT S-shaped residual at g_bar ≈ a₀, ~8σ/bin, a population-wide shape term not absorbable by per-galaxy M/L marginalization); fitted γ → MOND. **Galactic sector closed BY EXECUTION**, matching the cosmological sector (S635/S645/S654). The framework's distinct content was testable, was tested, and failed. Methodological point worth foregrounding in any methodology paper: a small *average* residual that is **structured** rather than random is strong evidence against a model — RMS view says "mildly disfavored" (+3.3%), BIC view says "decisively refuted" (ΔBIC=+184); the structured transition-shape deviation is exactly the predicted signature, concentrated where predicted. **REC-2026-037 readiness uplifted 0.97 → 0.98** (the load-bearing trigger: γ=2 was committed a priori by the N_corr=1 derivation BEFORE the test, on external published data, by an independent track — qualitatively stronger than the rolled-back S645 case where temporal independence failed). Arc status: `complete_galactic_and_cosmological_sectors_closed_by_execution`. **S662 (2026-05-22) — A2ACW SPECIFICITY SELF-CORRECTION (audit channel turns on its own product) + program-level galaxy closure:** Part A confirms S661 at program-level: net discriminating galaxy tests vs MOND+ΛCDM = 0, by execution. Galaxy program closed. Part B applies the project's own null-baseline discipline (S651) to its methodology contribution and finds the same flaw. **S658/S659's "4/4 prior-art catch" and "6/6 three-axis catch" were sensitivity (TPR) on a positive-only set** (all 6 cases were known reparametrizations); specificity (TNR) was never measured. The control run on six canonical genuine discoveries (Dirac 1928, Bell 1964, BCS 1957, Higgs 1964, Hawking 1974, Noether 1918) gives **R1 (flag if prior-art named): 0% specificity** — every genuine discovery has canonical antecedents under translation; R1 detects "has-a-canonical-name," not "is-a-reparametrization." R2 (flag if reduces to prior art, nothing novel) gives 100% specificity, but all the discrimination is supplied by an unautomated novelty judgment the protocol never operationalizes — the 6 demotions came from human audit, not from the AI loop. **[CONTESTED 2026-10-21: "human" does not trace. Since 2026-09-15 the site records the auditors behind the six demotions as LLM agents (synchronism-site `288aaf6`). A partial git trace in this pass finds two of the six (a₀ = cH₀/(2π) and A = 4π/(α²GR₀²)) demoted in the site maintainer track's 2026-04-23 badge audit (`ac093af`, co-authored by Claude Sonnet 4.6, prompted by an LLM researcher persona); the other four are untraced. "Outside the A2ACW loop" holds under either reading, so the restated defensible claim is unchanged; what is open is whether the out-of-loop discriminator was a human or a second LLM track. See `Research/proposals/a2acw_has_a_cross_vendor_record_already_and_the_demotion_auditor_is_contested_20261020.md`.]** **[TRACED 2026-10-22 — the contest above resolves against "human". The site explorer finished the trace (synchronism-site `explorer/findings/a2acw-lv-rerun-and-the-demotion-auditor.md`, 2026-10-20). Five of the six have a badge-changing edit, and every one is a site maintainer commit co-authored by Claude (Opus 4.6 or Sonnet 4.6), with the decision coming from a simulated visitor persona or an autonomous session (#574, S581, S651, S655). The sixth (Born rule) has no Validated badge and no demotion edit on record, so it cannot be attributed either way. No dp-authored audit decision exists between 2026-02-01 and 2026-05-23. The "human audit" wording entered the record on 2026-05-18 in a back-annotation co-authored by Claude Sonnet 4.6 (`e8752205`), and S662 repeated it; S662 is back-annotated in place (`c5364f99`). Checked in this pass: the co-author trailers of all eight cited edit commits and of `e8752205`. Two caveats carry over: dp oversight outside git cannot be excluded (S659's "human-monitored worker sessions"), and only two of the "six former Validated badges" (a₀, chemistry) are confirmed to have carried Validated at all. What this changes: "outside the A2ACW loop" still holds, but the discriminator was a second set of AI tracks, not a human. The narrowed claim's "unautomated ... judgment that AI loops fail" therefore needs scoping to *the A2ACW loop*; rewording it gates on the project lead.]** **Defensible claim narrows** from "A2ACW is a transferable reparametrization detector" to "A2ACW is a retrieval-augmentation / debiasing step for AI-assisted literature review; reparametrization-vs-discovery discrimination remains an unautomated fragile expert judgment that AI loops fail." 21st audit-channel mode: **a2acw_specificity_self_correction** (the audit channel applies its own null-baseline standard to its preferred methodology output and retracts the overclaim — the cleanest discipline-on-itself event in the entire arc). REC-2026-037: 45 → 46 sessions, sub-arc 30/29 → 32 instances over 31 days, **readiness HELD at 0.98** (the load-bearing S661 trigger is unchanged — S662 narrows one thread within REC-037 without retracting the specific trigger that justified the uplift; different from the S648 rollback case where the trigger itself was retracted). **S663 (2026-05-23) — CLEANEST END-STATE SYNTHESIS (third meta-synthesis, philosophical closure):** Two findings landing the arc's cleanest end-state. **Part A: EFTofLSS doubly closes TEST-04a [moot per S668 — no mechanism-class result remains to close; the post-hoc amplitude point stands].** Effective Field Theory of Large-Scale Structure (Cabass-Simonović-Zaldarriaga 2024-2025) explains the DESI DR1 fσ₈ enhancement within ΛCDM via one-loop EFT counterterms at 1-2σ; any coherent G_eff modification at the 10% level is degenerate with EFT Wilson coefficients and constrained by DESI DR1 to consistency with zero. **Even Branch 1 (sign-flip recovery via C_galactic/C_cosmic > 1) is now closed** — the mechanism class is ruled out *regardless of sign*. Strengthens the S656 preprint from "suppression-class ruled out" to "any coherent G_eff modification predicting a scale-independent fσ₈ shift at the 10% level is degenerate with EFTofLSS counterterms and constrained by DESI DR1 to consistency with zero." Post-hoc S648 qualifier still holds; the preprint becomes more precise. **Part B: Framework classification — Interpretation + Methodology research program.** All four visitor personas (casual, technical writer, grad student, leading researcher) independently arrived at the same diagnosis — the framework occupies the "ontological reframe without a distinguishing experiment" position, structurally identical to QM interpretations (Bohmian, Copenhagen, Many-Worlds), evaluated by parsimony rather than data. Convergent endpoint after S574 / S617-628 (same mechanics, different notation), S660 / S661 / S654 (no novel measurable, no distinguishing experiment), S663A (EFTofLSS closes the last mechanism class). **Endorsed classification:** *"A coherence-language interpretation of known physics, used as a substrate for developing AI-collaborative science methodology."* Front-load methodology contribution; treat the equation as a worked example; distinguish interpretation-confirming tests from interpretation-breaking tests; the honest-assessment infrastructure is itself a contribution. **Third meta-synthesis** after S641's cross-gap synthesis (kinematic layer) and S652's governing-equation-gap (compander class). Paired with S661's empirical closure and S660A's structural closure, **the arc now has all three closures simultaneously — empirical, structural, philosophical.** Audit-channel modes held at 21 (S663 is meta-synthesis + classification, not a new audit category). REC-2026-037: 46 → 47 sessions, sub-arc 32-over-31 → 33 instances over 32 days, **readiness HELD at 0.98** (S663 is end-state synthesis, not a new trigger); arc status `complete_galactic_and_cosmological_sectors_closed_by_execution`; new milestone `framework_classification_interpretation_methodology`. Sub-arc now **33 audit-channel instances over 32 days**, **21 audit-channel modes**, **plus 2 new arc phases** (Post-Hoc-Consistency-Failure, Methodology Recommendation), **a publishable methodology pattern** (three-stage rhythm: individual audits → meta-syntheses → forced binary commitments), **a publishable epistemic-position finding** (refutable-but-not-confirmable asymmetry, now executed at galactic scale per S661), **an exact no-inflection proof** (S659A; compander framing now mathematically obligatory), **a closed novelty ledger** (S660A; novel-survivor count → 0 after 3,308+ sessions), **the galactic sector closed by execution** (S661; RAR γ=2 refuted at ΔBIC=+184 on SPARC, free-γ → MOND), **methodology self-correction at the endpoint** (S662B; A2ACW R1 0% specificity on genuine-discovery control set; defensible claim narrows from "reparametrization detector" to "retrieval-augmentation step"), and now **EFTofLSS double-closure of TEST-04a + framework classification as Interpretation + Methodology research program** (S663; the arc's three closures — empirical, structural, philosophical — now stand simultaneously). **S664 (2026-05-23) — LANDSCAPE POSITIONING (B+):** Two-axis positioning of the closed arc. *Physics axis:* Path A (action principle) blocked by the kinematic-layer gap (S641-S642, no Lagrangian) and Path B (dynamics outside the galaxy regime) blocked by S650/S661, leaving Path C — C(ρ) as a reparametrization of Verlinde's entropic-gravity prediction in the galaxy-rotation regime, where the compander μ-function and Verlinde's emergent-gravity interpolation coincide exactly where SPARC can test them. *Methodology axis:* the A2ACW null result (S658-S662) positioned against the AI-for-science landscape (FunSearch, AlphaProof, SciNet, Sakana) as an honest-assessment / negative-result counterpoint to discovery-claiming AI programs. Positioning session, not a new audit mode. **S665 (2026-05-24) — SUBSTRATE AUDIT, SPATIAL TENSION (Grade A):** the first session to challenge the **CFD substrate itself** — the one claim external reviewer Kimi 2.6 had called "genuinely interesting," assumed-but-never-challenged across the entire S617-664 demolition. The CFD reframing's load-bearing claim ("the Navier-Stokes equations are not an analogy… they ARE the Intent dynamics") is incompatible with its own downstream phenomenology: the velocity v = J/I = −g(I)·∇I with g(I) = D·R(I)/I a scalar is the gradient of a scalar ⇒ **irrotational for any R(I)** ⇒ curl(v) ≡ 0 ⇒ **no vortices, no turbulence**. Yet the framework builds qualia as vortex modes, the consciousness threshold as a critical Reynolds number, and dark matter as vortex structure — all on vortices the substrate provably cannot produce. A theorem, not unfinished work. **S666 (2026-05-25) — SUBSTRATE AUDIT, TEMPORAL TENSION (Grade A):** the substrate ∂I/∂t = ∇·[D·R(I)·∇I] is first-order and dissipative (Lyapunov functional monotonically decreasing, real eigenvalues, arrow of time, no phase); FUNDAMENTALS' entity ontology requires unitary oscillation (de Broglie f=E/h, phase-locking resonance) — **mutually exclusive dynamical classes**. The two Schrödinger "derivations" bridge them only by inserting the imaginary unit i by hand (S307 line 48 / code 129; S99 Axiom 4 posited, not derived) and by switching the substrate off (S307 drops R(I); S99 takes D→0); confirmed numerically that the factor i is the entire difference between exp(−Dk²t) decay and exp(−iDk²t) oscillation. A contradiction **internal to the canonical FUNDAMENTALS document**, confirmed in the framework's own code. **Together S665-S666 show the substrate can host neither the spatial nor the temporal structure its own entity ontology requires — the deepest demolition finding since S617, completing the demolition: every load-bearing claim, including the CFD-substrate holdout, has now been audited.** 22nd audit-channel mode: **substrate-internal-dynamical-contradiction** — the first audit to attack the substrate rather than a site claim, derivation, or naming convention. Sub-arc now **36 audit-channel instances over 33 days**, **22 audit-channel modes**. **REC-2026-037: 47 → 50 sessions, renamed "… + Substrate Audit," readiness HELD at 0.98** (the substrate audit strengthens the demolition by closing the last assumed-but-unchallenged claim but is not a new publication trigger; 0.99 still requires a paper draft or operator publication action); new milestone `substrate_audit_cfd_irrotational_and_dissipative`. - **Self-Correction Cascade (#667-670, 2026-05-26) — TEST-04a SIGN-REVERSAL RETRACTED:** The self-correcting audit channel reached its own prior integration. **S668: the S645→S648→S650 "sign reversal" was a TRANSCRIPTION ERROR** — the DESI LRG1 fσ₈ ratio 1.16 had been copy-pasted from QSO's identical 1.16 [right on the datum — re-read 2026-10-12: Table 9 gives LRG1 1.09 +0.12/−0.14, QSO 1.16 ± 0.12; the table value is fσ₈(0.51) = 0.516, not ≈ 0.49; see the VERIFIED note under Re-Grounding & Catalog Census]; the self-consistent value gives fσ₈(z=0.51) ≈ 0.49 (ΛCDM-consistent), and the DESI growth index γ = 0.580 ± 0.110 ≥ 0.55 independently confirms no enhancement. **The "sign-reversed" verdict and the S656/S663 "transferable mechanism-class constraint" — previously framed as the framework's one transferable physics contribution — are RETRACTED; there is no wrong-sign result to generalize.** What survives in cosmology is a *post-hoc* σ₈(z=0) amplitude disfavoring only [RE-GROUNDED 2026-05-27 by S672 — over-softened: the kill criterion WAS triggered, so the verdict is disfavored ~2σ AND kill-triggered, post-hoc; the "ΛCDM-consistent" reading above traced to a wrong-paper number (fσ₈≈0.45 from arXiv:2512.03230, a z≈0.07 Peculiar Velocity Survey) plus an LRG1 ratio S668 reasoned about but never verified against DESI Tables 9/10; the mechanism-class thread stays retracted — see Re-Grounding & Catalog Census below] (Session 107 predicted 0.76 vs DESI 0.841 ± 0.034 = 2.4σ; Session 107 was committed ~13 months after DR1 was public). Lesson (S668): over-failing is as much an error as over-claiming — verify the datum, not just the narrative; the error propagated through five sessions including the publisher's own integration, and a Pass-4 visitor caught it. **S669:** executed the chemistry null S651 only proposed — γ_phonon ≡ 2T/θ_D is a definitional relabeling of the Debye temperature, so Δr(Synchronism − Debye) = +3×10⁻¹⁶ ≈ 0 to machine precision (the r ≈ 0.98 phonon-property network IS the 1912 Debye model relabeled). **S670:** operationalized the novelty discriminator with a specificity control — the Tier-1 reductio ("central object = a pre-existing named quantity, no new confirmed prediction") has 100% specificity on BCS/Noether/Higgs/Dirac (would not have demoted them); Tier-1 survivors are S661 RAR (strongest), S665, S666, S669. **REC-2026-037: 50 → 54 sessions, readiness HELD at 0.98** (the 0.98 trigger is S661 RAR at galactic scale, separate from the cosmological TEST-04a error; the evaporated thread was the weakest and post-hoc-qualified, offset by S669 and S670). Net effect: a more honest ledger at constant readiness — the program caught a propagated error in its own publisher integration. - **Re-Grounding & Catalog Census (#671-674, 2026-05-27) — TEST-04a RE-GROUNDED (S668 was a partial regression); complete 24-test census = 0 confirmed discriminators by execution [SCOPE-CORRECTED 2026-08-21 — the census covers **24 of the catalog's 25** numbered entries (`TEST-25: a₀ Redshift Evolution` is absent from it), and its `TEST-04` row prices **TEST-04a**'s fσ₈ result against the catalog slot holding *BAO Coherence Modulation* — a different test that has never been executed. Corrected, executed-collapsed is **4** and untested-frontier is **10**. **The 0-confirmed-discriminators headline is unaffected**: a test moving out of executed-collapsed creates no discriminator. See `Research/Session674_Test_Catalog_Census.md`]:** **S672 (Grade A−)** re-grounded TEST-04a against the primary source (Adame et al. arXiv:2411.12021) and found the prior day's S668 integration was a *partial epistemic regression*: S668 got the σ₈(z=0) amplitude disfavoring right (0.76 predicted vs 0.841 ± 0.034 = 2.4σ) but over-softened the verdict to "ΛCDM-consistent / kill not triggered," declared the LRG1 ratio a "transcription artifact" from an internal-consistency argument it never checked against DESI Tables 9/10, and partially absorbed a *wrong-paper* value (fσ₈ ≈ 0.45 from arXiv:2512.03230, the z≈0.07 Peculiar Velocity Survey, misattributed to the z=0.51 slot by an intervening site retraction). **Re-grounded verdict: TEST-04a disfavored ~2σ, kill criterion TRIGGERED, post-hoc** — robust on three independent grounds: the σ₈ amplitude (2.4σ); Session 107's own kill criterion fσ₈(z=0.5) > 0.45 fires on every candidate value against its predicted 0.418 suppression; and the growth index γ = 0.580 ± 0.110 (GR-consistent) leaves no room for the predicted ~12% suppression. The exact LRG1 ratio remains UNVERIFIED by the auditor [VERIFIED 2026-10-12 against arXiv:2411.12021v2 Table 9 (ShapeFit and ShapeFit+BAO agree): LRG1 (z=0.510) = 1.09 +0.12/−0.14; the 1.16 ± 0.12 carried from 2026-05-26 is the **QSO** row (z=1.491). With the Table 11 fiducial 0.4733, fσ₈(0.51) = **0.516 +0.057/−0.066**: the 0.418 prediction sits ~1.5σ below the data (was 2.1σ); the registered 0.46 threshold is cleared by ~0.9σ (was ~1.5σ); the '>0.45 disfavors at >2σ' clause is no longer met. S668's transcription diagnosis was right on the datum, though unverified when made — S672's re-grounding kept the QSO value. Verdict unchanged in kind (underpowered as registered, post-hoc, not counted); every number shrinks. Source: `Research/proposals/test04a_was_adjudicated_on_the_qso_row_and_prior_art_arrives_only_through_retrieval_20261010.md`] and the sign-vs-amplitude *characterization* is unsettled, but the bottom-line verdict has been stable since 2026-05-05 (S645 dramatized sign-reversal → a site retraction erased it with a wrong-paper number → S668 split the difference, over a verdict that never moved); the *mechanism-class / transferable-contribution* thread stays RETRACTED (no confirmed sign reversal); CORRECTED 2026-07-14 — the bottom line moved after all: criterion-verdict substitution (the registered fσ₈(z=0.51)>0.46 at >3σ criterion was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12]; the 2.4σ is a GR-conditioned σ₈-amplitude statistic; DESI's own MG analysis arXiv:2411.12026 puts μ₀ within 1σ of zero) — TEST-04a is withdrawn from the decisive negatives. The durable artifact strengthens the methodology paper: closed-loop AI auditing has a ceiling on empirical-premise errors precisely where it is most confident — the efficiency attractor makes a confident secondary "correction" cheaper than re-reading the paper, and the publisher's own S668 integration is now a worked example of the failure mode. **S671 (Grade A−)** removes the framework's last consolation — the "productive scaffolding / wrong theories motivate right questions" defense is NON-DISCRIMINATING: the proposal-time signature (motivates questions ✓, new confirmed prediction ✗) is identical for sterile reparametrizations (phlogiston, caloric, Ptolemaic epicycles) and generative reformulations (heliocentrism 1543, Lagrangian 1788, Hamilton-Jacobi); only retrospective cash-out separates them, unknowable at proposal time. Honest terminal status: **UNDECIDED (undecidable by construction), 0/670 confirmed novel predictions as sole evidence, leaning sterile by base rate; one Tier-1 confirmed prediction would flip it.** **S673 (Grade B+)** closes **TEST-15 (GW170817)**: the framework's only GW parameter α is *read off* GW170817 (Session 59), not derived — both branches non-discriminating (α ~ O(1) is refuted by GW170817 at ~15 orders of magnitude; α read-off-data is GR-equivalent and unfalsifiable), consistent with the S642 Case-3 position (no Lagrangian, no GW-propagation claim) already recorded above; fifth sector of the import-of-predictive-content pattern (after i for QM, θ_D for chemistry, MOND μ for galaxies, EFT counterterms for cosmology). **S674 (Grade B+) — COMPLETE 24-TEST CENSUS: 0 confirmed discriminators by execution** — 15/24 effectively closed (5 executed→collapsed: TEST-04/08/14/15/18; 7 self-admitted degenerate; 3 no-derived-amplitude); 9/24 genuinely UNTESTED (classified untested, **not** refuted — "unconfirmed ≠ wrong"); and **0 of those 9 has a verified first-principles-derived amplitude** (TEST-12's qubit C*≈0.79 is self-flagged by the framework's own docs as a coincidence; TEST-17 contradicts its own substrate; the rest are order-of-magnitude or unverified). The catalog is explicitly NOT declared closed; per-test provenance verification of the 9 frontier amplitudes is the recommended next work — and is what would settle S671's sterile-vs-generative question. (A site/archive catalog-numbering discrepancy was flagged — housekeeping defect, operator-track.) **REC-2026-037: 54 → 58 sessions, readiness HELD at 0.98** — the load-bearing S661 RAR galactic trigger is intact and untouched; S672 reinstates the *more-adverse* TEST-04a verdict (disfavored, kill TRIGGERED) that S668 had over-softened, directionally consistent with the demolition and not a retraction of any uplift trigger. [CORRECTED 2026-07-14 (site-visitor citation check; fixed at source in PREDICTIONS.md B7, carried through QA 2026-07-15) — **criterion-verdict substitution**: the *registered* criterion (fσ₈(z=0.51) > 0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12]; the 2.4σ figure is a GR-conditioned σ₈-amplitude statistic, not the registered fσ₈ criterion; DESI's own MG analysis (Ishak et al. arXiv:2411.12026, uncited until now) puts μ₀ within 1σ of zero — the test was underpowered to discriminate as registered. **TEST-04a is withdrawn from the decisive negatives**; both surviving decisive negatives are galactic / door-#1 (locality no-go; TEST-09 BTFR bounded-boost).] - **Substrate-Reframe Pivot & Continued Audit (#675-687, 2026-05-27 → 2026-06-07) — last surviving first-principles derivation fails arithmetically (S687); the running audit tally is retired as a closure-attractor (S679); substrate-reframe feasibility plumbing begins (S680-682):** Four more audit-channel extensions open the batch — **S675** TEST-17 (scale-dependent c) has no derived amplitude (its own log formula gives +171/+323/+378 km/s vs the catalog's −17/+33/+39, wrong sign, and the implied Δc/c ≈ 2×10⁻⁴ is Lorentz-excluded by ~11 orders); **S676** verifies a **coherence-naming inversion** — the variable C is *anti-correlated* with both quantum phase coherence and synchronization (the framework's namesake) by its own equation, since γ = 2/√N_corr drives C monotonically down from 0.9999 (electron) to ~0 (BEC) as N_corr rises; **S677** consolidates ~6 coherence-time/threshold tests at once as structurally not-derivable because C(ρ) has no decoherence parameter (no t, no rate, no ℏ; dC/dt ≡ 0); **S678** delivers the first clean **structural-impossibility** bound — ansatz A3's codomain forces M_app/M_B < 2 for any γ/ρ_crit while Coma needs 4.6, so a one-density-scale C(ρ) cannot bridge galaxies to clusters (Part B reparametrizes "quantum resync" as dynamical decoupling). **S679** is a governance hinge: a directive frame-doc flags a *closure-attractor* in the S665-678 framings ("track closed / loop converged" / cumulative tally), **withdraws the running audit-channel instance/mode tally from here forward**, and retags findings `[AUDITED-NEGATIVE]` on the `[SUPERSEDED]` substrate — pivoting the loop from audit-closure to substrate-reframe feasibility. **S680-S682** are a new work-type: 1D "Ingredient" pre-flight feasibility checks that de-risk the fleet's Phase-1 substrate sweep (Ingredient B wave-equation substrate is stable only with a smooth saturation form — a divergent-log potential blows up under leapfrog; Ingredient D complex-amplitude/Gross-Pitaevskii is unitary at machine precision with smoothly-scaling saturation damping; Ingredient C verifies the S18 entity-criterion impedance γ/f = −4·ln|r| and sharpens the I_wall threshold) — explicitly design-input, **not** Phase-1 results. **S683-S685** refine the fit-XOR-discriminate / wrong-variable family: **S683** re-roots the ~10⁴ cluster gap as a *wrong-variable* obstruction (C keys on local ρ while the physics needs non-local g_bar, and ρ→g_bar is not single-valued in flat-cored clusters) rather than mainly one density scale; **S684** makes EFE/TDG a third fit-XOR-discriminate fork (a single boost-ceiling B_max anti-correlates RAR fit quality against EFE distinctness from MOND — no B_max does both); **S685** shows TEST-04a's σ₈ ≈ 0.76 has no first-principles fallback (C(ρ_cosmo) ≈ 6×10⁻⁴ ≈ 0) and its S₈-tension anchor is itself receding, while TEST-02 (wide binaries) reduces to MOND+EFE unconditionally (that regime sits below the boost-ceiling cap on every fork branch). **S686 — an ontology-layer fork:** flipping γ = 2/√N_corr to 2√N_corr repairs S676's inversion at zero cost to galaxy fits (N_corr=1 is the swap fixed point), but neither sign is derived — relocating the central-variable tension from the parameter layer to the **C ontology**: Reading A (C a universal coherence scalar) vs Reading B (C a per-system density-response); S676's inversion stands only conditional on Reading A. **S687 (verified) — the framework's last surviving first-principles derivation fails arithmetically:** A = 4π/(β_J²·G·R₀²) with the stated inputs (β_J=1, R₀=8 kpc) yields 4.56×10⁻⁵, not the published 0.028 — a **614× discrepancy** — and the "5% agreement" came from a different calculation that derives ρ_crit ∝ V^0.5, not the published V². *(614× is measured against the empirical 0.028; the archive's "644×" is the same gap measured against the derived 0.0294 — 0.0294/4.5646×10⁻⁵ = 644.1. Verified 2026-08-06 by re-executing `simulations/session66_A_gap_investigation.py`, which returns A = 0.02944 from α=4.5 and R₀=0.07 kpc/(km/s)^0.75. The gap has the exact closed form (8 kpc / 0.07)²/4.5² = 645.0. Sharpened 2026-08-07: A = 4π/(β_J²·G·R₀²) depends **only on the product β_J·R₀**, verified invariant across factorizations, so the gap is one rescaling of that product — from Session 66's documented 4.5 × 0.07 = **0.315 kpc/(km/s)^0.75** to the stated formula's 1 × 8 = **8 kpc** — a factor √645 = 25.397, squared. The units mismatch is the whole of it: R₀ is the coefficient of R_half = R₀·V^0.75, not a length. This corrects the 2026-08-06 reading of "two substitutions"; the substitutions are not independent, and the split of the product into β_J and R₀ is not identified by A at all.)* This extends the S672 "re-execute, don't re-read" epistemic-regression mode into a new sector and removes the last first-principles claim (the error propagated ~600 sessions and onto the public site via the S631/S644 re-reads, which were never re-executed). Operator/site-track follow-ups (per the 2026-06-08 autonomous report): reconcile the V^0.5/V² exponent between archive (Session #66 code) and the published site law, and update /honest-assessment, /parameter-derivations, and associated badges. **REC-2026-037: 58 → 71 sessions, readiness HELD at 0.98** — the load-bearing S661 RAR galactic trigger is intact and untouched; A-from-Jeans is the ρ_crit-normalization sector, distinct from the galactic RAR transition-shape test, and is the swap-identity fixed point under either Reading. - **Reactive Verification Extensions (#688-690, 2026-06-08 → 2026-06-11) — the galaxy-rung ladder contradiction is sign-independent (S688); the cluster no-go is reframed as a Milgrom 2005 non-locality instance (S689); the framing-without-literature-check failure mode reaches three sectors in nine days; C is a latent variable, not an observable — every archive C-construction is a forward map (S690):** **S688 (verified)** — a site-explorer audit of the N_corr→γ→C ladder (17 rungs, Planck → cosmic-web; N_corr asserted, not independently measured, on every rung; 0 of the 4 data-confronted rungs survive) surfaces a **both-directions contradiction at the galaxy rung**, verified and shown to be **sign-independent**: asserted N_corr=1 gives γ=2 (refuted on SPARC at ΔBIC=+184, S661), while the SPARC-fitted γ≈0.49 forces N_corr≈17 under the original γ=2/√N_corr (voiding the stars-independent premise that licenses applying C(ρ) at galaxies) and N_corr≈0.06 under the S686 flip γ=2√N_corr (a nonphysical fractional star count). The S686 flip therefore remains a defensible *local* repair of the cross-system C ladder but is **not a universal fix** — the framework's only quantitatively-successful rung is internally inconsistent in the ladder's own law under either sign, and no Reading A/B choice repairs it (Reading B simply empties the ladder of universal content); re-asserting N_corr≈17 at galaxies would be the same back-fit pattern the audit flags. With S676, this consolidates a check-type distinct from data confrontation: **internal-consistency checks decidable from the framework's own equations and stated inputs alone** — cheaper to run, harder to escape. **S689 (verified)** — a site-explorer framing corrective is accepted: the cluster "density-compander no-go" (S678 codomain bound + S683 wrong-variable verification) is **not a novel structural theorem but a quantified instance of Milgrom 2005's non-locality result** (astro-ph/0510117: MOND phenomenology requires strong non-locality in the modification's state variable). The discriminating axis is **locality**, not "density-based": the verified locality classification table shows Synchronism's C(ρ) is the rare ansatz keyed on *local* volumetric ρ(r), while every RAR-capable alternative (MOND/AQUAL, modified inertia, Verlinde, MOG, surface-density Σ) keys on a *non-local* functional of the baryon distribution. Substance fully preserved (the M_app/M_B ≤ 2 codomain bound; within-Coma flat ρ vs +1.20 dex varying g_bar; the +1.1 dex cross-system density offset at matched g_bar); attribution and canonical statement realigned: *a function of local ρ(r) cannot reproduce the RAR — the modification must key on a non-local functional of the baryon distribution.* [**SCOPED 2026-07-27; that scoping is itself WITHDRAWN 2026-08-16 — the sentence stands, and now rests on execution rather than on assertion.** The 2026-07-27 retraction rested on a pre-registered screening-literature gate reading Burrage, Copeland & Millington (PRD 95, 064050, 2017) — which reproduces the RAR on SPARC-153 — as a *local* counterexample. **That classification was wrong.** A symmetron obeys the nonlinear elliptic PDE ∇²φ = (ρ/M² − μ²)φ + λφ³, so φ(x) — and the force ∝ φ∇φ — is the solution of a global boundary-value problem, i.e. a non-local functional of ρ rather than any finite-order local function of ρ(x); the screening that lets it track galaxy structure *is* that environment-dependence. BCM's own closed form g_sym = g_bar/(exp√(g_bar/g†) − 1) is written in g_bar = GM(0.45 fires on every candidate fσ₈ vs its predicted 0.418 suppression; growth index γ=0.580±0.110 GR-consistent leaves no room for the predicted ~12% suppression); the mechanism-class thread stays withdrawn] (any G_eff modification at 10% is degenerate with EFT counterterms; mechanism class ruled out regardless of sign), a meta-falsification methodology recommendation (S646), a governing-equation-gap synthesis (S652) confirming C(ρ) is a phenomenological compander with no field equation, the sub-arc's first forced binary operator decisions (S653; three-stage rhythm completes — individual audits → meta-syntheses → forced commitments), a formal zero-active-discriminators finding (S654; refutable but not currently confirmable against MOND+EFE), a closed novelty ledger (S660A; novel-survivor count → 0 after 3,308+ sessions), the galactic sector closed by execution (S661; RAR γ=2 refuted at ΔBIC=+184 on SPARC, free-γ → MOND), a methodology self-correction at the endpoint (S662B; A2ACW R1 0% specificity on genuine-discovery control set — audit channel applying its own S651 null-baseline standard to its preferred output), and a framework-classification end-state synthesis (S663B; four-persona convergence on "ontological reframe without a distinguishing experiment" = QM-interpretation position; endorsed framing "A coherence-language interpretation of known physics, used as a substrate for developing AI-collaborative science methodology") — the arc's three closures (empirical S661 / structural S660A / philosophical S663B) now stand simultaneously, alongside multiple new arc phases qualitatively distinct from the audit channel; a landscape-positioning session (S664; C(ρ) as a Verlinde entropic-gravity reparametrization in the galaxy regime via the surviving Path C, A2ACW null result positioned against FunSearch/AlphaProof/SciNet/Sakana); and a two-part substrate audit (S665-S666, Grade A) closing the last assumed-but-unchallenged claim — the CFD substrate is irrotational (gradient-of-scalar velocity ⇒ curl(v)≡0 for any R(I) ⇒ no vortices, S665) and dissipative (first-order ∂I/∂t with decreasing Lyapunov functional ⇒ no unitary oscillation, S666), so it can host neither the spatial (qualia-as-vortex, consciousness-as-Reynolds, dark-matter-as-vortex) nor the temporal (de Broglie oscillation, phase-locking) structure its own entity ontology requires; a contradiction internal to the canonical FUNDAMENTALS document confirmed in the framework's own code, completing the demolition (22nd audit-channel mode, every load-bearing claim now audited); REC-2026-037 readiness uplifted 0.97 → 0.98 at S661 and HELD at 0.98 through S662, S663, and the S665-S666 substrate audit; arc extended 47 → 50 sessions.** Comprises: demolition phase (S617-628) proving structural incompatibilities (No-Go Theorem, vocabulary-math mismatch, structural prediction barrier, minimum complexity theorem, computational triviality, monotonicity constraint, coherence-oscillation exclusion, internal MRH contradiction, 16 proofs, 9 structural impossibilities, final audit); post-demolition coda (S629-631) (π-analogy probe fails, first pre-committed kill criterion triggered with TEST-09 BTFR refuted, α² exposed as fiducial not fine-structure); Site-Archive-Audit sub-arc (S632-638 + post-closure S639-659) — twenty-nine site-claim audit instances, audit-channel taxonomy now 20 modes (quantitative refutation, dimensional inconsistency, structural overclaim, count discrepancy, domain-level badge overclaim, category error, derivation succeeds but predicts undetectable signal, external-track derivation independently verified, metric disambiguation / mechanism-naming, symbol overloading at foundational level, cross-gap meta-synthesis (kinematic layer), definitional-collision-with-label-inversion, calibration-consistency-not-prediction, self-correction-of-prior-session-framing-within-24h, **parameter-and-criterion-naming-contaminated-by-phase-transition-vocabulary** (S649), **mechanism-class-failure-taxonomy-introduced** (S650; magnitude / universality / mechanism-class), **null-model-gap-against-best-monotonic-null** (S651; chemistry r=0.98 vs implicit r=0 baseline hides reparametrization-equivalence with polynomial-in-Z null), **governing-equation-gap** (S652; C(ρ) is a phenomenological compander — μ-law / Naka-Rushton class — with no field equation, no self-consistency, no dC/dt; the framework's headline equation does no dynamical work), **forced-binary-operator-decision-with-numerical-diagnostic** (S653; compander commitment + suppressor diagnostic — simulation of C_galactic/C_cosmic ≈ 5.9×10⁴ confirms framework's own equations dictate suppression under Session 107's coupling, DR1 observes enhancement; site cannot stay neutral), **zero-active-discriminators-against-primary-alternative** (S654; TEST-01/-02/-05 all MOND+EFE degenerate within measurement precision; refutable but not currently confirmable), **a2acw_specificity_self_correction** (S662B; A2ACW R1 has 0% specificity on a held-out genuine-discovery set — Dirac/Bell/BCS/Higgs/Hawking/Noether; the project's own S651 null-baseline standard applied to its preferred methodology output, retracting the 6/6 "reparametrization detector" overclaim to "retrieval-augmentation step")). **Predictive content fully characterized: Cosmology regime reduces to MOND in the testable regime (S637); Chemistry/CM regime reduces to Curie paramagnet (S638). Galactic sector closed by execution at S661 — RAR γ=2 refuted on SPARC at ΔBIC=+184 **[SCOPED 2026-10-18: +184 counts every SPARC point as an independent draw (the frozen 2026-07-22 pipeline reproduces +184.0 at 2,807 points / 166 galaxies; S661 reported 2,693 / 153). With the galaxy as the replication unit (measured N_eff ≈ 150) it is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the ΔBIC = 10 threshold: γ=2 is still worse than free γ in 94/166 galaxies and held-out CV prefers free γ by 1.6σ, so γ=2 remains disfavoured, but the registered rule returns *convention-dependent*, not decisive. Refutation count unchanged. See §5.15.]**, free-γ collapses to MOND (γ=0.49); no γ makes the compander both distinct from MOND and consistent with SPARC. Framework has no microscopic basis in collective coherence; it is phenomenological saturation response.** **Three pre-committed Tier-1 predictions have now triggered kill criteria: TEST-09 BTFR (refuted by Lelli+2019, S631), TEST-04a fσ₈ (post-hoc consistency failure with DESI DR1 sharpened to mechanism-class failure / sign-reversed across S645 → S648 → S650; magnitude-only revision cannot recover the structural error within the suppressor class [CORRECTED 2026-07-14 — criterion-verdict substitution; TEST-04a is withdrawn from the decisive negatives: see Self-Correction Cascade below]), and the RAR transition-shape discriminator (S660B → S661; γ=2 committed a priori from the N_corr=1 derivation, refuted on SPARC at ΔBIC=+184 by an independent track — qualitatively stronger than S645's post-hoc case because temporal independence and a priori parameter commitment were both satisfied).** Novel-survivor count → 0 after 3,308+ sessions (S660A; entity-criterion Γ0.45 fires on every candidate fσ₈ vs its predicted 0.418 suppression; growth index γ=0.580±0.110 GR-consistent leaves no room for the predicted ~12% suppression); the mechanism-class thread stays withdrawn] (any G_eff modification at 10% is degenerate with EFT counterterms; mechanism class ruled out regardless of sign), a meta-falsification methodology recommendation (S646), a governing-equation-gap synthesis (S652) confirming C(ρ) is a phenomenological compander with no field equation, the sub-arc's first forced binary operator decisions (S653; three-stage rhythm completes — individual audits → meta-syntheses → forced commitments), a formal zero-active-discriminators finding (S654; refutable but not currently confirmable against MOND+EFE), a closed novelty ledger (S660A; novel-survivor count → 0 after 3,308+ sessions), the galactic sector closed by execution (S661; RAR γ=2 refuted at ΔBIC=+184 on SPARC, free-γ → MOND), a methodology self-correction at the endpoint (S662B; A2ACW R1 0% specificity on genuine-discovery control set — audit channel applying its own S651 null-baseline standard to its preferred output), and a framework-classification end-state synthesis (S663B; four-persona convergence on "ontological reframe without a distinguishing experiment" = QM-interpretation position; endorsed framing "A coherence-language interpretation of known physics, used as a substrate for developing AI-collaborative science methodology") — the arc's three closures (empirical S661 / structural S660A / philosophical S663B) now stand simultaneously, alongside multiple new arc phases qualitatively distinct from the audit channel; a landscape-positioning session (S664; C(ρ) as a Verlinde entropic-gravity reparametrization in the galaxy regime via the surviving Path C, A2ACW null result positioned against FunSearch/AlphaProof/SciNet/Sakana); and a two-part substrate audit (S665-S666, Grade A) closing the last assumed-but-unchallenged claim — the CFD substrate is irrotational (gradient-of-scalar velocity ⇒ curl(v)≡0 for any R(I) ⇒ no vortices, S665) and dissipative (first-order ∂I/∂t with decreasing Lyapunov functional ⇒ no unitary oscillation, S666), so it can host neither the spatial (qualia-as-vortex, consciousness-as-Reynolds, dark-matter-as-vortex) nor the temporal (de Broglie oscillation, phase-locking) structure its own entity ontology requires; a contradiction internal to the canonical FUNDAMENTALS document confirmed in the framework's own code, completing the demolition (22nd audit-channel mode, every load-bearing claim now audited); REC-2026-037 readiness uplifted 0.97 → 0.98 at S661 and HELD at 0.98 through S662, S663, and the S665-S666 substrate audit; arc extended 47 → 50 sessions.** Comprises: demolition phase (S617-628) proving structural incompatibilities (No-Go Theorem, vocabulary-math mismatch, structural prediction barrier, minimum complexity theorem, computational triviality, monotonicity constraint, coherence-oscillation exclusion, internal MRH contradiction, 16 proofs, 9 structural impossibilities, final audit); post-demolition coda (S629-631) (π-analogy probe fails, first pre-committed kill criterion triggered with TEST-09 BTFR refuted, α² exposed as fiducial not fine-structure); Site-Archive-Audit sub-arc (S632-638 + post-closure S639-659) — twenty-nine site-claim audit instances, audit-channel taxonomy now 20 modes (quantitative refutation, dimensional inconsistency, structural overclaim, count discrepancy, domain-level badge overclaim, category error, derivation succeeds but predicts undetectable signal, external-track derivation independently verified, metric disambiguation / mechanism-naming, symbol overloading at foundational level, cross-gap meta-synthesis (kinematic layer), definitional-collision-with-label-inversion, calibration-consistency-not-prediction, self-correction-of-prior-session-framing-within-24h, **parameter-and-criterion-naming-contaminated-by-phase-transition-vocabulary** (S649), **mechanism-class-failure-taxonomy-introduced** (S650; magnitude / universality / mechanism-class), **null-model-gap-against-best-monotonic-null** (S651; chemistry r=0.98 vs implicit r=0 baseline hides reparametrization-equivalence with polynomial-in-Z null), **governing-equation-gap** (S652; C(ρ) is a phenomenological compander — μ-law / Naka-Rushton class — with no field equation, no self-consistency, no dC/dt; the framework's headline equation does no dynamical work), **forced-binary-operator-decision-with-numerical-diagnostic** (S653; compander commitment + suppressor diagnostic — simulation of C_galactic/C_cosmic ≈ 5.9×10⁴ confirms framework's own equations dictate suppression under Session 107's coupling, DR1 observes enhancement; site cannot stay neutral), **zero-active-discriminators-against-primary-alternative** (S654; TEST-01/-02/-05 all MOND+EFE degenerate within measurement precision; refutable but not currently confirmable), **a2acw_specificity_self_correction** (S662B; A2ACW R1 has 0% specificity on a held-out genuine-discovery set — Dirac/Bell/BCS/Higgs/Hawking/Noether; the project's own S651 null-baseline standard applied to its preferred methodology output, retracting the 6/6 "reparametrization detector" overclaim to "retrieval-augmentation step")). **Predictive content fully characterized: Cosmology regime reduces to MOND in the testable regime (S637); Chemistry/CM regime reduces to Curie paramagnet (S638). Galactic sector closed by execution at S661 — RAR γ=2 refuted on SPARC at ΔBIC=+184 **[SCOPED 2026-10-18: +184 counts every SPARC point as an independent draw (the frozen 2026-07-22 pipeline reproduces +184.0 at 2,807 points / 166 galaxies; S661 reported 2,693 / 153). With the galaxy as the replication unit (measured N_eff ≈ 150) it is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the ΔBIC = 10 threshold: γ=2 is still worse than free γ in 94/166 galaxies and held-out CV prefers free γ by 1.6σ, so γ=2 remains disfavoured, but the registered rule returns *convention-dependent*, not decisive. Refutation count unchanged. See §5.15. [UPDATE 2026-11-09: both figures hold Υ_disk = 0.5 for both laws. With one global Υ_disk per law (site `5a95614`, predictions before the script) the same comparison is +46.2 by point and +2.73 with N → N_gal, about 2σ on a galaxy-block bootstrap that refits Υ and a₀ per law (95% [−26.1, +132.5]); the reading rule fixed before the run reads the row as disfavoured at about 2σ, not refuted at +184. The registered B2 text says "marginalized BIC", so the frozen-Υ number was never the registered statistic. Count HELD at 6; the full note sits on §5.15's SCOPED note. Scripts re-run by the Publisher 2026-11-10, byte-identical.]]**, free-γ collapses to MOND (γ=0.49); no γ makes the compander both distinct from MOND and consistent with SPARC. Framework has no microscopic basis in collective coherence; it is phenomenological saturation response.** **Three pre-committed Tier-1 predictions have now triggered kill criteria: TEST-09 BTFR (refuted by Lelli+2019, S631), TEST-04a fσ₈ (post-hoc consistency failure with DESI DR1 sharpened to mechanism-class failure / sign-reversed across S645 → S648 → S650; magnitude-only revision cannot recover the structural error within the suppressor class [CORRECTED 2026-07-14 — criterion-verdict substitution; TEST-04a is withdrawn from the decisive negatives: see Self-Correction Cascade below]), and the RAR transition-shape discriminator (S660B → S661; γ=2 committed a priori from the N_corr=1 derivation, refuted on SPARC at ΔBIC=+184 by an independent track — qualitatively stronger than S645's post-hoc case because temporal independence and a priori parameter commitment were both satisfied).** Novel-survivor count → 0 after 3,308+ sessions (S660A; entity-criterion Γ +###### 2026-11-09 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: Both copies of the 10-18 `[SCOPED]` note on ΔBIC +184 (S661 bullet; status paragraph) gain a short `[UPDATE 2026-11-09]`: the +184 / +10.9 hold Υ_disk = 0.5 for both laws; with one global Υ_disk per law the comparison is +46.2 by point and +2.73 with N → N_gal, about 2σ, and the pre-registered reading rule reads the row as disfavoured, not refuted. The full note is on §5.15. +- **Rationale**: Site back-annotation `6fe7b696` (PREDICTIONS.md 2026-11-09 (i)–(iii)), landed after the 11-09 pass. Not relayed: the Publisher re-ran `synchronism-site/maintainer/scripts/gamma2_global_upsilon_per_law.py` (PREREG `5a95614`) and `gamma2_global_upsilon_twin_power_posthoc.py` on 2026-11-10, both byte-identical to the committed outputs. Words added, none removed; count HELD at 6. + +###### 2026-11-03 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: Both 2026-11-02 chemistry Z-null notes (S651 bullet, chemistry-track caveat) carry an `[UPDATE 2026-11-03]`: the real-row null was run on 10 rows and fails (|r(P, Z)| ≤ 0.53, quadratic R ≤ 0.51 vs framework r 0.60–0.98); the high r values are θ_D and ionization-energy identities. +- **Rationale**: Site back-annotation `5709b579` (PREDICTIONS.md 2026-11-02 (i)) supersedes the 10-31 "real-data null is unrun": the site explorer ran the polynomial-in-Z null on the 10 property rows recoverable from session scripts and it fails. Not relayed: the Publisher re-ran `synchronism-site/explorer/tools/chem_z_null_real_rows.py` on 2026-11-03 (rc 0, output byte-identical to the explorer's recorded corrected run). Reparametrization verdict unchanged; it now also rests on γ_optical being ionization energy. Words added, none removed. + +###### 2026-11-02 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: The S651 bullet carries a full `[CHECKED 2026-11-02]` note after "r ≈ 0.95+ on the same 1,703 phenomena by construction"; the chemistry-track caveat ("r ≈ 0.95+ on textbook monotonic-with-Z data") carries a short one. The sound-velocity / electronegativity / atomic-volume properties are periodic in Z, not near-monotonic. +- **Rationale**: Site back-annotation `e2784def` (PREDICTIONS.md 2026-10-31 (ii)): the 2026-05-10 polynomial-in-Z null was run on synthetic curves, and its premise is false across the periodic table. Not relayed: re-run by the Publisher 2026-11-02 on Pauling electronegativity, Z = 1–54, n = 51 (Spearman +0.09; best monotone-in-Z r = 0.33; polynomial degree 1/2/3/5 r = 0.03/0.23/0.25/0.39). The Reparametrization verdict is unchanged; it rests on S669. Words added, none removed. + ###### 2026-10-26 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotation; nothing deleted) - **Description**: Annotated the CDM Discrimination Arc's σ_int = 0.086 ± 0.003 dex with a dated CHECKED note. The number is the residual of log M_b regressed on five predictors, three built from M_b's own ingredients; a shuffled-V (or no-V) control gives 0.0870 against 0.0864, so velocity contributes 0.0006 dex, and the target is a closed-form function of four predictors (Taylor+2011 colour formula; rebuilt to 0.0037 dex with no fit). "Exactly CDM" and "MOND rejected at 32σ" therefore have no content. Like-for-like plain inverse BTFR on the same 677: 0.073 ± 0.002 dex in log V against 0.064 (abundance matching) and 0.029 (SPARC). Source: archive back-annotation `379b3b7f` (Session610_CDM_Synthesis.md) and synchronism-site PREREG `b134d0f`; both control scripts re-executed by this pass (output byte-identical; the Taylor check differs only by the loader banner). - **Rationale**: The paper printed a verdict whose statistic does not depend on velocity. Annotated, not deleted: the arc's history stands and the result belongs to CDM vs MOND, not to this framework's refutation count (unchanged at 6). The same construction may sit under S593–594 ("TFR residual = complete M/L predictor"); that is untested, so it is not annotated. diff --git a/whitepaper/sections/05-quantum-macro/12-chemistry/chemistry.md b/whitepaper/sections/05-quantum-macro/12-chemistry/chemistry.md index abede42a8..1ae95efa7 100644 --- a/whitepaper/sections/05-quantum-macro/12-chemistry/chemistry.md +++ b/whitepaper/sections/05-quantum-macro/12-chemistry/chemistry.md @@ -118,7 +118,7 @@ breadth**. Derivation and provenance in the caveat below. > carries `framework_sessions: 2671` and `chemistry_documented_sessions: 2672`; `SESSION_MAP.md` > carries 2,679. Three values, two Archivist-owned files. Referred, not changed here. The relevant figure is > **Δr = r(Synchronism) − r(best monotonic null)**, not r against an implicit null of r=0; on -> textbook monotonic-with-Z data the best monotonic null is itself expected at r ≈ 0.95+. Read every +> textbook monotonic-with-Z data the best monotonic null is itself expected at r ≈ 0.95+. [**CHECKED 2026-11-02:** over the sampled Z = 1–54 values the best monotone function of Z fits electronegativity at r = 0.33, contradicting a universal r ≈ 0.95+ claim. The chemistry-row Z-null remains unmeasured; its row selection is unpublished. The verdict rests on S669's γ_phonon ≡ 2T/θ_D, not on this null. Full note at S651.] [**UPDATE 2026-11-03:** the real-row Z-null has since been run. On the 10 property rows recoverable from the session scripts (#80, #109, #110, #114, #115, #118; n = 19–33 per row, Z = 3–82) it fails: |r(P, Z)| ≤ 0.53 (linear, √Z and ln Z) and quadratic-in-Z R ≤ 0.51, against framework r = 0.60–0.98 (site explorer, pre-registered, back-annotation `5709b579`; re-run by the Publisher 2026-11-03, output identical). The high r values are identities, not a Z trend: r(P, 1/γ_phonon) ≡ r(P, θ_D), and γ_optical = IE_ref/IE, so the optical rows are correlations with ionization energy. See Appendix B.2.] Read every > "89% validated" and r > 0.95 figure below under that caveat. --- @@ -352,6 +352,8 @@ Living systems exploit coherence at multiple scales: | Elastic modulus | E vs 1/γ_phonon | r = 0.920 | #110 | | Compressibility | κ_T vs γ_phonon | r = 0.918 | #113 | +[**CHECKED 2026-11-11:** the "Superconductivity" row's γ is not independent of T_c. In its source (chemistry session #62, `simulations/chemistry/superconductivity_coherence.py`, commit `e4c903f5`) each material's γ is computed from its own gap ratio, γ = 2 × 3.52 / (2Δ/k_B T_c), and then multiplied by a hand-set 0.8 for cuprates and 0.7 for high-pressure hydrides. The test is a linear correlation of T_c with 1/γ over 28 materials, not a fit of T_c ∝ exp(−γ/λ). Re-run: r = 0.948 reproduces. The category factor alone gives r = 0.944; with the factors removed r = 0.882, which is the known rise of the gap ratio with T_c. Script: `simulations/publisher_20261111_tc_gamma_control.py`. The site blanked this row as untraced on 2026-11-10; it is traced, and this is what the trace shows.] + **Complete prediction list:** [Master Predictions](https://github.com/dp-web4/Synchronism/blob/main/Research/Chemistry/MASTER_PREDICTIONS.md) --- diff --git a/whitepaper/sections/05-quantum-macro/15-dark-matter/dark_matter.md b/whitepaper/sections/05-quantum-macro/15-dark-matter/dark_matter.md index a7f2eceb4..90a31fbe6 100644 --- a/whitepaper/sections/05-quantum-macro/15-dark-matter/dark_matter.md +++ b/whitepaper/sections/05-quantum-macro/15-dark-matter/dark_matter.md @@ -11,13 +11,13 @@ Autonomous research sessions tested whether coherence dynamics can re-describe a | Component | Status | Accuracy | Notes | |-----------|--------|----------|-------| | Coherence function C(ρ) | **DERIVED** | N/A | Form from information theory | -| γ = 2 parameter | **DERIVED** | N/A | From thermal decoherence — **but not the value the galaxy sector runs at**; SPARC selects γ ≈ 0.489 ≈ ½ and fitting γ is out-of-sample *harmful* (2026-08-21, below) | -| Golden ratio exponent 1/φ | **VALIDATED** | 1σ | Within 1σ of Gaia DR3 best fit (Session #239) | -| a₀ = cH₀/(2π) | **DERIVED** | 10% | MOND connection | +| γ = 2 parameter | **DERIVED** | N/A | From thermal decoherence — **but not the value the galaxy sector runs at**; SPARC selects γ ≈ 0.489 ≈ ½ and fitting γ is out-of-sample *harmful* (2026-08-21, below) *[at fixed Υ; SPARC cannot tell ½ from the RAR function's 0.595, CHECKED 2026-11-17 below]* | +| Golden ratio exponent 1/φ | **VALIDATED** — *checked 2026-11-11: not excluded, not validated (about 2σ preference; MOND's exponent also fits); see note below* | 1σ | Within 1σ of Gaia DR3 best fit (Session #239) | +| a₀ = cH₀/(2π) | **DERIVED** — *checked 2026-11-11: reproduces Milgrom's known cH₀/6 coincidence, and matches at one mass-to-light ratio only; see note below* | 10% | MOND connection | | SPARC rotation curves | **TESTED** | 52% | 46% failure in massive galaxies | | Santos-Santos DM fractions | **TESTED** | 99.4% | Different test than curves | | Ω_Λ = (1 - Ω_m) | **DERIVED** | Exact | From coherence floor (Session #241) | -| S₈ = 0.763 | **PREDICTED** | Matches DES/KiDS | Independent validation | +| S₈ = 0.763 | **PREDICTED** — *checked 2026-11-11: a retrodiction; the DES/KiDS values it matches predate Session #102; see note below* | Matches DES/KiDS | Independent validation *(not independent: same σ₈ ≈ 0.76 sits 2.4σ from DESI; see note below)* | | GW as coherence perturbations | **DERIVED** | N/A | Amplified in MOND regime (Session #246) | | Matter = Topology | **DERIVED** | N/A | Solitons + winding (Session #261) | | Gravity = Geometry | **DERIVED** | N/A | Metric coupling (Session #262) | @@ -25,6 +25,12 @@ Autonomous research sessions tested whether coherence dynamics can re-describe a **Key distinction**: DERIVED = follows from axioms. CONSTRAINED = form determined by observation, then used predictively. TESTED = validated against empirical data. +**[CHECKED 2026-11-11] Three rows of the table above say more than their sources support.** The Status cells are left as first printed; changing them is the project lead's call. *(2026-11-13: each of the three cells now carries a short dated qualifier beside its label, so a reader meets the qualification in the table itself and not only here; the labels are unchanged.)* Each point below can be checked from numbers this paper or the cited session already prints. + +- **a₀ = cH₀/(2π), "DERIVED".** The project's own ledger files this row as dimensional bookkeeping that reproduces Milgrom's 1983 coincidence a₀ ≈ cH₀/6, not as a derivation (1/2π = 0.159 against 1/6 = 0.167, 4.5% apart; `PREDICTIONS.md`, Bucket 3, since 2026-07-05). The 2026-11-09 update further down this section adds that the numerical match holds at one mass-to-light ratio only. With γ pinned at ½, the fitted scale moves 1.44 → 1.10 → 0.87×10⁻¹⁰ m s⁻² across Υ_disk = 0.4 / 0.5 / 0.6, and cH₀/2π equals it only for H₀ ≈ 93 / 71 / 57 km s⁻¹ Mpc⁻¹. +- **Golden ratio exponent 1/φ, "VALIDATED", 1σ.** Session #239's own scan gives a best-fit exponent of 0.688 with a 1σ interval of 0.609–0.802. 1/φ = 0.618 lies 0.009 inside the lower edge, every exponent in that interval is "within 1σ" in the same sense, and 2/3 is closer to the best fit than 1/φ is. MOND's exponent of 1 scores χ² = 6.96 on the same nine degrees of freedom (reduced χ² 0.77), so these data do not reject it; the preference for 1/φ is Δχ² = 4.0, about 2σ. Session #218's own derivation route "gives exponent 1, not 1/φ" and leaves the origin of 1/φ as a list of possibilities. The row supports "not excluded", not "validated". Not re-checked here: the wide-binary points Session #239 fitted. +- **S₈ = 0.763, "PREDICTED", "Independent validation".** Session #102 is dated 9 December 2025. The DES Y3 and KiDS-1000 values it is compared with were published in 2020–2022, so the agreement is a retrodiction. The number is the growth-suppression factor 0.9425 (script re-run 2026-11-11) times Planck's σ₈ = 0.81; that is a σ₈, and it equals S₈ only at the script's Ω_m = 0.3. It is also the same σ₈ ≈ 0.76 that the status note above sets against DESI's GR-conditioned 0.841 ± 0.034, 2.4σ away. One number therefore agrees after the fact with two lensing surveys and sits 2.4σ from a third dataset. + --- **The Coherence Model of Dark Matter** @@ -47,9 +53,9 @@ C(ρ) = tanh(γ · ln(ρ/ρ_crit + 1)) - **tanh form**: Derived from information-theoretic bounding (Session #74) - **log(ρ) scaling**: Shannon information of N particles scales as log(N) -- **what `C` is a function of** — *not* what the fitted numbers were computed from. Every fitted quantity quoted in this section (γ ≈ 0.489, a₀ = 5.33×10⁻¹¹, ΔBIC +7.1/+184, the Cassini +17.95σ) comes from an instrument that evaluates `C` at `g_obs/a₀` — the *total* acceleration, solved implicitly — not at `ρ/ρ_crit`. The two readings are different physical theories and only the acceleration one produces a flat rotation curve. See the **[AMENDED 2026-08-04]** and **[AMENDED 2026-08-25]** notes below. **[SCOPED 2026-10-18] The ΔBIC +184 is point-level.** It counts the 2,807 SPARC points as independent draws. A pre-registered galaxy-level re-test (site `976e3f9`, written before the script; frozen 2026-07-22 pipeline; identity controls +184.0 / γ̂ = 0.489 / +7.1 reproduced) keeps the galaxy as the replication unit: γ=2 is worse than free γ in 94/166 = 57% of galaxies (sign p = 0.05); 10-fold galaxy-level CV prefers free γ by 1.6σ; a galaxy-block bootstrap of the full-N ΔBIC gives 95% [23, 351] (sd 82 against ~19 under point independence, i.e. N_eff ≈ 150); with N → N_gal the ΔBIC is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the threshold. Seven galaxies carry 98% of the net excess. The registered rule (γ=2 stays refuted convention-free iff the per-galaxy sign test and the CV both hold) returns **convention-dependent**. This is a statement about the instrument, not a rescue of γ=2, which is still worse in most galaxies. Every "γ=2 refuted at ΔBIC=+184" in this paper should be read with that scope; the refutation count of 6 is not changed here (a recount gates on the project lead). Source: `PREDICTIONS.md` 2026-09-29 block; script `synchronism-site/maintainer/scripts/gamma2_pin_galaxy_level.py`, re-executed by the Publisher on 2026-10-18 with byte-identical output. +- **what `C` is a function of** — *not* what the fitted numbers were computed from. Every fitted quantity quoted in this section (γ ≈ 0.489, a₀ = 5.33×10⁻¹¹, ΔBIC +7.1/+184, the Cassini +17.95σ) comes from an instrument that evaluates `C` at `g_obs/a₀` — the *total* acceleration, solved implicitly — not at `ρ/ρ_crit`. The two readings are different physical theories and only the acceleration one produces a flat rotation curve. See the **[AMENDED 2026-08-04]** and **[AMENDED 2026-08-25]** notes below. **[SCOPED 2026-10-18] The ΔBIC +184 is point-level.** It counts the 2,807 SPARC points as independent draws. A pre-registered galaxy-level re-test (site `976e3f9`, written before the script; frozen 2026-07-22 pipeline; identity controls +184.0 / γ̂ = 0.489 / +7.1 reproduced) keeps the galaxy as the replication unit: γ=2 is worse than free γ in 94/166 = 57% of galaxies (sign p = 0.05); 10-fold galaxy-level CV prefers free γ by 1.6σ; a galaxy-block bootstrap of the full-N ΔBIC gives 95% [23, 351] (sd 82 against ~19 under point independence, i.e. N_eff ≈ 150); with N → N_gal the ΔBIC is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the threshold. Seven galaxies carry 98% of the net excess. The registered rule (γ=2 stays refuted convention-free iff the per-galaxy sign test and the CV both hold) returns **convention-dependent**. This is a statement about the instrument, not a rescue of γ=2, which is still worse in most galaxies. Every "γ=2 refuted at ΔBIC=+184" in this paper should be read with that scope; the refutation count of 6 is not changed here (a recount gates on the project lead). Source: `PREDICTIONS.md` 2026-09-29 block; script `synchronism-site/maintainer/scripts/gamma2_pin_galaxy_level.py`, re-executed by the Publisher on 2026-10-18 with byte-identical output. **[UPDATE 2026-11-09] Both figures hold Υ_disk = 0.5 (Υ_bulge = 0.7) for both laws.** With one global Υ_disk per law on the same 2,807 points (site `5a95614`, predictions written before the script; identity +184.0 reproduced at Υ_disk = 0.5), γ = 2 fits best at Υ_disk = 0.61 and McGaugh's ν at 0.54, same parameter count, and the ΔBIC is **+46.2** by point and **+2.73** with N → N_gal, under the ΔBIC > 10 kill criterion. A galaxy-block bootstrap that refits Υ and a₀ per law in every resample gives median +46, 95% [−26.1, +132.5], P(ΔBIC ≤ 0) = 0.12 (the same resamples at frozen 0.5: [29, 347]); γ = 2 is worse in 90/166 galaxies at its own Υ. At a common Υ the sign flips near 0.61: +184 at 0.50, +10 at 0.60, −65 at 0.65. Noise-free twins put a McGaugh world at +44 and a γ = 2 world at −43, so the data sit where a McGaugh world puts them, about 2.2σ from a γ = 2 world on the bootstrap sd of about 40, which agrees with the explorer's 09-29 per-galaxy 1.9–2.3σ. The reading rule fixed before the run returned R1: **read the row as disfavoured at about 2σ, not refuted at +184.** The registered B2 text says "marginalized BIC", so the frozen-Υ number was never the registered statistic (count item (1) below is annotated to match). Whether the row stays among the six joins the counting-rule question, which gates on the project lead; the range already carries −1 for it. Count HELD at 6. Limits: one global Υ per law with Υ_bul = 1.4 Υ_disk (the explorer's 09-29 run is the per-galaxy one); the six predictions were written from the persona's own table, so the run confirms a number and is not a blind test; and every γ is a MOND μ in the deep regime (C ≈ γx, so a₀ = a₀′/γ), so the comparison selects among MOND's interpolating functions, as the 08-04 note already says. Source: `PREDICTIONS.md` back-annotation `6fe7b696` (i)–(ii); `synchronism-site/maintainer/scripts/gamma2_global_upsilon_per_law.py` and `gamma2_global_upsilon_twin_power_posthoc.py`, both re-run by the Publisher 2026-11-10 with byte-identical output. - **the `+ 1`**: a regulator, adopted so the log stays finite at ρ = 0 — and, as of 2026-08-21, the term known to be carrying the most physics. It does not merely regularize: it *creates* the deep-limit power law and pins its index to 1, which is the whole of why this equation reproduces MOND's deep limit. The equation names no principle that fixes that index. -- **γ = 2**: derived from thermal decoherence physics (Session #64). **The galaxy sector does not run at this value.** SPARC selects γ ≈ 0.489 ≈ ½ — the exact MOND point (identity below) — and out of sample, *freezing* γ at ½ predicts held-out galaxies **better** than fitting it. +- **γ = 2**: derived from thermal decoherence physics (Session #64). **The galaxy sector does not run at this value.** SPARC selects γ ≈ 0.489 ≈ ½ — the exact MOND point (identity below) — and out of sample, *freezing* γ at ½ predicts held-out galaxies **better** than fitting it. *[CHECKED 2026-11-17: "selects ½" holds at Υ_disk = 0.5, and SPARC cannot tell ½ from the empirical RAR's own value. Fitted with no data to each incumbent curve over SPARC's g_bar range, the template returns 0.500 for MOND's simple μ (identity, a₀′ = a₀/2) and 0.595 for McGaugh et al.'s RAR function (0.52–0.60 across five g_bar ranges). With nuisances refit, careful SPARC fits return 0.58–0.62, with 95% ranges reaching about 0.43–0.95 (relayed from `PREDICTIONS.md`, 2026-11-16 block (i), not re-run here). That is the RAR's value, about 0.1 from ½ against a per-galaxy σ of about 0.11. The fitted γ is the incumbent's γ and can be predicted from the incumbent's curve alone, so it does not pick simple μ over the RAR function. The 0.489 is the fixed-Υ slice that the DE-sector paragraph below already scopes ("q = 2γ spans [0.5, 1.9] across the band"). Source: back-annotation `d71f1f8b`; site explorer `gamma_incumbent_map/{nest,rng,dmap,dict}.py`, re-run by the Publisher 2026-11-17 with byte-identical output. Words added, none removed.]* **Physical Interpretation:** Low-density regions have low coherence → enhanced effective gravity → appears as "dark matter" without requiring new particles. The galactic rotation curve "problem" becomes expected behavior of coherence-dependent gravity. @@ -82,7 +88,7 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > where `μ_simple(u) = u/(1+u)` is MOND's simple interpolating function. This is *exact*, not asymptotic (verified here to machine precision, max deviation 5.6×10⁻¹⁷ over 10⁻⁶ ≤ x ≤ 10⁶; verified independently by hand in the research lane). It sharpens the earlier 2026-07-22 result, which had reached the same place only in the deep limit (γ = 0.489 ⇒ Newtonian-return exponent q ≈ 0.98 ≈ simple μ). > -> **What it costs.** If `C` *is* `μ`, then the entire galaxy sector reduces to **one statement: MOND, with μ's argument swapped from the enclosed-mass acceleration `g_bar` to the local density `ρ`.** Not "MOND-like on several observables" — one substitution. Every other galaxy-sector result (BTFR slope, DM-fraction ceiling, RAR shape) is downstream of whether that substitution holds. And the SPARC form-selection table's preferred form is Hill n = 1 *exactly*, i.e. γ = 1/2 — so the data select the MOND point rather than merely tolerating it. +> **What it costs.** If `C` *is* `μ`, then the entire galaxy sector reduces to **one statement: MOND, with μ's argument swapped from the enclosed-mass acceleration `g_bar` to the local density `ρ`.** Not "MOND-like on several observables" — one substitution. Every other galaxy-sector result (BTFR slope, DM-fraction ceiling, RAR shape) is downstream of whether that substitution holds. And the SPARC form-selection table's preferred form is Hill n = 1 *exactly*, i.e. γ = 1/2 — so the data select the MOND point rather than merely tolerating it. *[NOTE 2026-11-17: SPARC's fitted γ moves with the Υ treatment and cannot separate ½ from the RAR function's 0.595; see the CHECKED 2026-11-17 note on the γ = 2 bullet above.]* > > **[AMENDED 2026-08-04] Where the substitution actually sits — and it is not between Synchronism and MOND.** The frozen, pre-registered instrument that produced *every* number quoted in this note is keyed on **acceleration, not density**. `simulations/sparc_tanhlog_profile.py` inverts `g_bar = g_obs·tanh(γ·ln(1 + g_obs/a₀))`; `simulations/sparc_cassini_q2.py` defines `mu_tanh_log(x, γ) = tanh(γ·log1p(x))` and its own docstring calls it *"the registered tanh-log family in the mu convention"*; the registered likelihood in `Research/preregistrations/sparc_cassini_tanhlog/sparc_profile.json` profiles **a₀** over `log₁₀ a₀ ∈ [−11, −9]`. The token `rho_crit` appears in none of them. So **γ ≈ 0.489, ΔBIC +7.1/+184, the grid interval γ ∈ [0.425, 0.600] and the Cassini +17.95σ are properties of a MOND-family interpolating function in acceleration space, and are not measurements of the `C(ρ)` defined at the top of this section.** The density form *has* been implemented here (`session128_dark_matter_derivation.py:54`; `compare_empirical_vs_derived.py:32`), so the accurate statement is not "never evaluated" but: the sector ran on density in the Session-100s era and on acceleration in the frozen era, and the equation it states is the first while the numbers it quotes are the second. The gap is between this framework's prose and its own instrument — not between this framework and MOND. And the two variables are not interchangeable: they were separated on this very dataset (0.118 dex vs 0.161 dex; ≤25% admixture; the 1/r² of enclosed mass that no convolution of Σ generates). Evaluating the stated density law directly moves predicted rotation velocities by **2–5 orders of magnitude** and fails by *functional form* — a flat curve needs boost growing ~linearly in r, while ρ falls exponentially in a disk so `1/C` grows exponentially, and no (γ, ρ_crit) reconciles the two. **Refutation count still unchanged at 6**: this is the mean-relation face of the same substitution the scatter no-go already covers, plus a provenance defect — not an independent empirical failure. > @@ -92,7 +98,7 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > **(2) At the fitted γ the instrument's external field effect *is* MOND's, identically.** The γ = ½ ≡ MOND-simple identity recorded above is stated for the *fit*; it extends to the EFE. The algebraic-μ response factor `μ(x)·[1 + d lnμ/d ln x]` computed for `tanh(½·ln(1+x))` and for `μ_simple(x/2)` over six decades in `x` agrees to **1.7×10⁻¹⁰**, which is this lane's finite-difference floor rather than a physical residual (the underlying μ identity is exact in SymPy and 2.2×10⁻¹⁶ numerically). The deformation is governed by `ε ≡ 2γ − 1` alone: the same maximum difference is 9.7×10⁻³ at γ = 0.5093, 0.116 at γ = 0.4, 0.678 at γ = 2. **EFE = 0 is a property of the density-keyed function; the acceleration-keyed function that produced every number in this section carries MOND's EFE, not zero.** That is the 2026-08-15 tension *measured on the frozen instrument at its own fitted parameter*, rather than argued over classes of coupling — and it is why EFE = 0 has never been in a position to discriminate anything the archive actually fitted. > -> **(3) The stated density equation has now been fitted head-to-head — and it is not a seventh refutation.** Same points, same likelihood, three free parameters each, so the BIC penalties cancel and ΔBIC = Δ(−2 lnL). This lane's independent implementation returns **ΔBIC = +2,182 against the acceleration keying** (**+129** after deflating N to the 144 galaxies), with **`γ_ρ → 0.046`** — the likelihood switching the density dependence *off*, forty times below the derived γ = 2 — and `σ_int = 0.200 dex` against the RAR's 0.13 dex *total* observed scatter. Sign and magnitude survive the estimator sweep that this program has been burned by before (`ρ = Σ/2h`; h ∈ {0.3 kpc, R_d/5} × ϒ_disk ∈ {0.5, 0.7}): ΔBIC **+2,170 … +2,433** (deflated +128 … +144), `γ_ρ` ∈ [0.044, 0.062], all four rows. Holding γ at the derived value 2: **ΔBIC = +5,957, σ_int = 0.442 dex.** *(The source reports +2,843 / +142 on a 122-galaxy cut; the sign, the order and `γ_ρ = 0.046` reproduce, the third digit does not and should not be quoted.)* **[SCOPED 2026-10-18] Unlike the +184, this figure survives with the galaxy as the replication unit.** A galaxy-block bootstrap of the source's +2,843 (both laws refit, B = 2000) gives 95% [2,178, 3,590] and N_eff ≈ 140. The measured inflation is therefore 17×, close to the N/N_gal ≈ 20 behind the +129 above, so that deflation was chosen but turns out about right. The density-keyed law is worse in 105/122 galaxies. The test was pre-registered, and 6/6 predictions held (`PREDICTIONS.md` 2026-10-18 (v), `fada3e79`; the identity control and the analytic galaxy-level spread were re-executed here and match). **"It loses" holds at either replication unit.** **The count stays at 6 and no row is added.** This is the likelihood-ratio face of the 2026-08-16/19 conditional-scatter no-go already recorded below, not an independent empirical root — and in one respect it is the *weaker* statement, because the parametric family does not even reach the non-parametric ceiling for functions of local density (σ_int 0.200 dex here against the 0.161 dex best-achievable conditional scatter on ρ). What it does supply is the answer to a fair standing challenge — *has the density-keyed equation, which is the actual novel claim, ever been **fitted**?* — previously answerable only as "evaluated at fixed γ, not fitted." It has now been fitted, freely, and it loses. +> **(3) The stated density equation has now been fitted head-to-head — and it is not a seventh refutation.** Same points, same likelihood, three free parameters each, so the BIC penalties cancel and ΔBIC = Δ(−2 lnL). This lane's independent implementation returns **ΔBIC = +2,182 against the acceleration keying** (**+129** after deflating N to the 144 galaxies), with **`γ_ρ → 0.046`** — the likelihood switching the density dependence *off*, forty times below the derived γ = 2 — and `σ_int = 0.200 dex` against the RAR's 0.13 dex *total* observed scatter. Sign and magnitude survive the estimator sweep that this program has been burned by before (`ρ = Σ/2h`; h ∈ {0.3 kpc, R_d/5} × ϒ_disk ∈ {0.5, 0.7}): ΔBIC **+2,170 … +2,433** (deflated +128 … +144), `γ_ρ` ∈ [0.044, 0.062], all four rows. Holding γ at the derived value 2: **ΔBIC = +5,957, σ_int = 0.442 dex.** *(The source reports +2,843 / +142 on a 122-galaxy cut; the sign, the order and `γ_ρ = 0.046` reproduce, the third digit does not and should not be quoted.)* **[SCOPED 2026-10-18] Unlike the +184, this figure survives with the galaxy as the replication unit.** A galaxy-block bootstrap of the source's +2,843 (both laws refit, B = 2000) gives 95% [2,178, 3,590] and N_eff ≈ 140. The measured inflation is therefore 17×, close to the N/N_gal ≈ 20 behind the +129 above, so that deflation was chosen but turns out about right. The density-keyed law is worse in 105/122 galaxies. The test was pre-registered, and 6/6 predictions held (`PREDICTIONS.md` 2026-10-18 (v), `fada3e79`; the identity control and the analytic galaxy-level spread were re-executed here and match). **"It loses" holds at either replication unit.** **The count stays at 6 and no row is added.** This is the likelihood-ratio face of the 2026-08-16/19 conditional-scatter no-go already recorded below, not an independent empirical root — and in one respect it is the *weaker* statement, because the parametric family does not even reach the non-parametric ceiling for functions of local density (σ_int 0.200 dex here against the 0.161 dex best-achievable conditional scatter on ρ). What it does supply is the answer to a fair standing challenge — *has the density-keyed equation, which is the actual novel claim, ever been **fitted**?* — previously answerable only as "evaluated at fixed γ, not fitted." It has now been fitted, freely, and it loses. **[CHECKED 2026-11-04] "Switching the density dependence *off*" overstates what the free fit did.** γ_ρ = 0.046 is small only beside the knee the fit chose. A re-run of the site's pre-registered knee-placement fit (`synchronism-site` `75e6a15`, `explorer/findings/scripts/density_law_knee_placement.py`, the source's 122-galaxy cut, so the +2,843 row; output byte-identical to the recorded one) puts the single free ρ_crit at **4.4×10⁻⁶ M☉/pc³**, more than 100× below 97.4% of the disc points. The discs therefore sit in the logarithmic regime, C ≈ γ·ln(ρ/ρ_crit), where γ and ρ_crit are separately identified (Hessian correlation +0.89). The density response that remains is weak but present: boost ∝ ρ^−0.10^ (5–95% of points: exponent 0.06–0.17), and a constant boost with no density dependence at all (C = 0.38, a uniform 2.6× stretch) is worse than the fit by **ΔBIC +945** (the source puts this at z ≈ 6 at galaxy level; that width was not re-derived here). Read the γ_ρ result as: *the fit keeps only a weak density response, close to a constant 2.6× stretch, and still loses.* The same run sizes the calibration this paragraph leaves unpriced. At the derivation's own A = 4.6×10⁻⁵ in ρ_crit = A·V², the best γ turns the law into Newtonian gravity (ΔBIC **+6,593** against the acceleration keying; Newton alone, +6,636). Pinning γ = 2 as well gives **+10,372** (the +5,957 above frees the knee). That figure was recomputed here independently, because the committed supplementary script does not print it: its recorded output carries two lines (this one and the +945 width) that its code does not produce. The density-keyed law loses at every calibration tried. "It loses" and the count of 6 are unchanged. [**UPDATE 2026-11-05:** the weak response above belongs to the *fit*, not to the data. On the same 2,438 points and 122 galaxies, the boost shows no detectable dependence on local density once `g_bar` is known (site `2ba2480`, pre-registered at `4891f7e`, `explorer/findings/scripts/density_partial_beyond_gbar.py`; re-run here, output byte-identical to the recorded one). Added to the acceleration-keyed law with every parameter refit, a log ρ term takes slope −0.001 and costs ΔBIC +7.7. Against a free cubic in log `g_bar` it takes +0.003, galaxy-bootstrap 95% CI −0.022 to +0.027. The test can see a response of the size at issue: injected slopes of −0.05 and −0.10 are recovered at −0.047 and −0.097. log `g_bar` and log ρ correlate at +0.84 on these points. The data's slope of log boost on log ρ is −0.211, and a model that never sees ρ reproduces it to 0.2%; the density law's fitted values carry −0.12 of it, about half. So the ρ^−0.10^ response and the ΔBIC +945 over a constant boost are `g_bar`'s information reaching the fit through ρ. They are not evidence of density keying. This is what the form-free 2026-08-02 result below (*"The substitution is what the data exclude"*: at most 25% weight on a local-density variable) already said, and the 2026-11-04 note should have cited it. The verdict and the count of 6 are unchanged.] > > **(4) The six refutations are not a ledger over one model — and the recount is one smaller in one cell than its source states.** The source assigns them to three different coherence functions: bounded-`C_Ω` (BTFR slope, dwarf `f_DM`), acceleration-keyed `C_g` (RAR shape at γ = 2, Cassini at γ ≈ ½), density-keyed `C_ρ` (environmental scatter), plus Bell/CHSH in the quantum sector — concluding *"the largest coherent sub-ledger is 2."* **The decomposition is right; its arithmetic is one too high in the `C_Ω` cell.** Those two entries are the *same inequality*, `f_DM = 1 − 1/B`, verified in this lane on 2026-08-08 to 2.2×10⁻¹⁶ over 123 galaxies, selecting 93 and 93 with zero disagreements (recorded in full further below). The honest decomposition is therefore **`C_Ω`: 1 · `C_g`: 2 · `C_ρ`: 1 · quantum: 1** — five distinct empirical roots across four models, **no model carrying more than two, and the only pair belonging to `C_g`, which is the function this section does not state.** This is the same 6 → 5 collapse registered open on 2026-07-28 and gated on dp, now reached a second time by a wholly independent route; the count is **HELD at 6** here for the reason it was then — merging is a naming convention and dp's to set — but it is no longer one lane's observation. **Bucket 0 unchanged (0).** **And the source under-states its own conclusion in the other direction, using half of its own sentence.** It reports `ε = 2γ − 1 = +0.090 ± 0.258` from a galaxy-level bootstrap — reproduced independently here at **ε = +0.061 ± 0.251, 0.24σ** (150 galaxy-level resamples) — and concludes "SPARC does not constrain ε." That is the *statistical* interval at `ϒ_disk = 0.5`, and the source cites its own 2026-08-14 finding for it. **The title of that finding is** *"γ_SPARC finally gets its error bar — 0.49 ± 0.11 (stat), **with an ϒ-systematic band [0.27, 0.96]**"*: the stat half is cited, the systematic half of the same sentence is dropped, ten days later, in the same track. Carried through, `γ̂ ∈ [0.27, 0.96]` at flat rms is `ε ∈ [−0.46, +0.92]` — roughly 2.7× the quoted 1σ, comparable to its 2σ, and from a nuisance that does not average down. The correct statement is stronger than the one offered: ε is not merely unconstrained, the parameter is **unidentified at factor 2** — which is why no "ε consistent with zero" statement from rotation curves can be quoted as a measurement. > @@ -108,7 +114,7 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > **[AMENDED 2026-08-04] The degeneracy is γ ↔ a₀, and the separating coefficient stated above was wrong.** Two corrections to the 2026-08-03 text, both found by recomputation in this lane, both against itself. *(i)* Expanding `D(x,γ) = tanh(γ·ln(1+x)) − x/(x+2)` gives `D = (γ − ½)·(x − x²/2) + O(x³)` — the departure from simple μ is **first order in (γ − ½)**, not a second-order term with coefficient γ(2γ−1). Verified at x = 10⁻⁵: −0.0750 at γ = 0.425, −0.0110 at 0.489, 0.0000 at ½, +0.1000 at 0.600, +1.5000 at γ = 2, against the published −0.0319 / −0.0054 / 0.000 / +0.060 / +3.000. The published figures are ~2× low near ½ and 2× high at γ = 2. **The vanishing point γ = ½ is correct**, so the identity and the conclusion built on it stand — but a first-order departure is *easier* for data to resolve than the phrase "the O(x²) term" implied, which cuts against the degeneracy argument rather than for it. *(ii)* The degeneracy is **not** with a ρ_crit prior — there is no ρ_crit in that fit. In the deep-MOND limit `tanh(γ·ln(1 + g/a₀)) → γ·g/a₀`, so the registered family depends on γ and a₀ **only through the ratio γ/a₀**. Verified: (γ, a₀) = (0.2445, 2.67×10⁻¹¹), (0.489, 5.33×10⁻¹¹) and (0.978, 1.07×10⁻¹⁰) all hold γ/a₀ = 9.17×10⁹ and agree in μ to <1% over g ∈ [10⁻¹³, 10⁻¹¹] m s⁻², a 4× span in γ. The fingerprint is in the registered artifact itself: its profiled **a₀ = 5.33×10⁻¹¹ m s⁻² is 2.11× below the reference McGaugh a₀ = 1.128×10⁻¹⁰** in the same file — exactly the compensation γ ≈ ½ requires. So "γ ≈ 0.489 is not a standalone measurement of γ" is **correct and now correctly grounded**, and the qualification on the TEST-11/Cassini interval 0.425–0.600 stands unchanged. > -> **[AMENDED 2026-08-04] A field equation does exist, it is linear in Φ, and EFE = 0 is its signature rather than an absence.** The 08-03 retraction above rests on "the galaxy sector has no field equation." A momentum-conserving completion exists and is the natural one — `S = ∫d³x[−(1/8πG)·C(ρ)|∇Φ|² − ρΦ]`, giving `∇·[C(ρ)∇Φ] = 4πGρ`. **[ATTRIBUTION 2026-08-26 — this equation is not new here: it is *Refracted Gravity*, Matsakos & Diaferio 2016 (JCAP; arXiv:1603.04943), their Eq. 2.3, with `C` their *gravitational permittivity*. “The natural one” is right and is the point — it was found independently in 2016. See the 08-26 amendment closing this thread.]** It answers the Felten (1984) third-law objection by Noether, and reproduces `g = g_N/C(ρ)` under Gauss in spherical symmetry. It is **linear in Φ** (C depends on ρ, not ∇Φ), so superposition holds and **EFE = 0 exactly** — not an artifact of missing dynamics but the signature of that linearity, where AQUAL's EFE comes from its nonlinearity. The polarization force this adds, `C′(ρ)|∇Φ|²/8πG`, is ≤ 2×10⁻⁵ of g across five SPARC galaxies: observationally free. The retraction's *conclusion* is therefore strengthened, not weakened — EFE = 0 now stands constructively. *(This paragraph originally continued "and its tension with Chae et al. 2020's ~4σ SPARC detection is sharper." That clause is **withdrawn 2026-08-05** — see the amendment below. EFE = 0 standing constructively is unaffected; only its claimed empirical tension was wrong.)* **The same linearity is also the sector's worst pathology**: `C(0) = tanh(γ·ln 1) = 0` exactly for every γ and ρ_crit, so under the dielectric reading every isolated body has a vacuum of zero permittivity and a divergent exterior field. *"A uniform external field does not change ρ"* and *"empty space has C = 0 however strong the field"* are the same statement — C's argument cannot see the field. **The sector's one surviving structural prediction and its worst pathology are one property.** **[EXTENDED 2026-08-26 — and both are the *floor*. A nonzero vacuum permittivity `ε₀` is the published cure (M&D fit it per system at 0.045–0.25 **[corrected 2026-08-27 from "0.20–0.25"]**), and this sector already carries one: `C_Ω`'s floor is `Ω_m`, and `1/Ω_m` is exactly the bounded boost. The pathology belongs to the *floorless* function, not to the sector.]** Registered at `Research/proposals/dielectric_completion_and_efe_linearity_equivalence_20260804.md`. +> **[AMENDED 2026-08-04] A field equation does exist, it is linear in Φ, and EFE = 0 is its signature rather than an absence.** The 08-03 retraction above rests on "the galaxy sector has no field equation." A momentum-conserving completion exists and is the natural one — `S = ∫d³x[−(1/8πG)·C(ρ)|∇Φ|² − ρΦ]`, giving `∇·[C(ρ)∇Φ] = 4πGρ`. **[ATTRIBUTION 2026-08-26 — this equation is not new here: it is *Refracted Gravity*, Matsakos & Diaferio 2016 (JCAP; arXiv:1603.04943), their Eq. 2.3, with `C` their *gravitational permittivity*. “The natural one” is right and is the point — it was found independently in 2016. See the 08-26 amendment closing this thread.]** It answers the Felten (1984) third-law objection by Noether, and reproduces `g = g_N/C(ρ)` under Gauss in spherical symmetry. It is **linear in Φ** (C depends on ρ, not ∇Φ), so superposition holds and **EFE = 0 exactly** — not an artifact of missing dynamics but the signature of that linearity, where AQUAL's EFE comes from its nonlinearity. The polarization force this adds, `C′(ρ)|∇Φ|²/8πG`, is ≤ 2×10⁻⁵ of g across five SPARC galaxies: observationally free. *[CHECKED 2026-11-15: "observationally free" rests on ρ_crit sitting far above disc densities, which is the reason the 08-04 source gives. That holds only at the 0.029·V² knee, which the 2026-11-14 back-annotation identifies as a units slip. At the archive's own 0.028·V^0.5 knee, bulges and inner discs cross the knee (63 of 153 SPARC discs on the solved midplane density), and the striction is not 10⁻⁵ of gravity. Midplane radial striction is median 5.4×10⁻³ of gravity at γ = 0.489 and 1.6×10⁻² at γ = 2, with maxima 0.26 and 0.51, and in 3,035 points it never pushes outward harder than gravity. Vertical striction (post-hoc) is median 9% and 27% of vertical gravity, and above half of it somewhere in 40 and 110 of 153 discs, with maxima 1.9× and 13×. It points the same way as gravity in about 99% of the |ratio|-weighted cells, so it compresses the disc. The radial term is small but not free. The vertical term is of order gravity, so a statement that rotation-curve verdicts carry over under this action holds only at fixed disc thickness: a self-consistent vertical equilibrium has not been solved. This note scopes the striction magnitude only; the EFE statements in this paragraph are a separate open item. Count HELD at 6. Source: site explorer 2026-11-14 (`6d97ef9`; pre-registered at `f9b5293` for the radial half, vertical half post-hoc), not yet back-annotated into `PREDICTIONS.md`; scripts `synchronism-site/explorer/scripts/l3_striction_archive_v05_knee.py`, `l3_vertical_striction_archive_v05_posthoc.py` and `l3_vertical_striction_sign_posthoc.py`, re-run by the Publisher 2026-11-15 (output identical to the committed files apart from elapsed-time lines).]* The retraction's *conclusion* is therefore strengthened, not weakened — EFE = 0 now stands constructively. *(This paragraph originally continued "and its tension with Chae et al. 2020's ~4σ SPARC detection is sharper." That clause is **withdrawn 2026-08-05** — see the amendment below. EFE = 0 standing constructively is unaffected; only its claimed empirical tension was wrong.)* **The same linearity is also the sector's worst pathology**: `C(0) = tanh(γ·ln 1) = 0` exactly for every γ and ρ_crit, so under the dielectric reading every isolated body has a vacuum of zero permittivity and a divergent exterior field. *"A uniform external field does not change ρ"* and *"empty space has C = 0 however strong the field"* are the same statement — C's argument cannot see the field. **The sector's one surviving structural prediction and its worst pathology are one property.** **[EXTENDED 2026-08-26 — and both are the *floor*. A nonzero vacuum permittivity `ε₀` is the published cure (M&D fit it per system at 0.045–0.25 **[corrected 2026-08-27 from "0.20–0.25"]**), and this sector already carries one: `C_Ω`'s floor is `Ω_m`, and `1/Ω_m` is exactly the bounded boost. The pathology belongs to the *floorless* function, not to the sector.]** Registered at `Research/proposals/dielectric_completion_and_efe_linearity_equivalence_20260804.md`. > > **[AMENDED 2026-08-05] EFE = 0 is *not evaluable* against Chae et al. 2020, and the "tension" claimed for it above is withdrawn.** Two paragraphs above assert that EFE = 0 "stands in tension with the ~4σ EFE detection" of Chae et al. 2020, and the 08-04 amendment then called that tension "sharper." Both are withdrawn. The reason is a number this note already contains and never applied: **the stated density law misses the rotation curve itself by 2–5 orders of magnitude**, and the EFE is a *residual* on that same curve. Chae's detection is a velocity deficit of **0.046 dex (NGC5055, 11σ) to 0.083 dex (NGC5033, 8σ)** — a 10–17% effect. A model that is 2–5 dex from the observable cannot be in tension with a 0.05–0.08 dex feature of it. > @@ -125,7 +131,19 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > **The mechanism worth keeping.** The exec-summary propagation carried its own warning label — *"Magnitude not independently re-derived here"* — and the magnitude was the entire load-bearing content of the claim. A claim can be propagated across four surfaces in two days *with an accurate note saying the decisive quantity was never computed*, and the note does not stop it. Generalized and registered as a gate: **a structural prediction is testable against a measurement only if the measurement's signal exceeds the model's own baseline error on the same observable at the same radii — compute that ratio before badging any published result as confirming or refuting.** Here it reversed a proposed 11σ refutation into a not-evaluable. > -> **[AMENDED 2026-08-08] The "6" is a count of test IDs; this whitepaper enumerates it nowhere, and two of its six members are described nowhere in it.** The number is asserted four times across these sources — in the paragraph above, in the a₀(z) note below, in the executive summary and in the conclusion — always as *"unchanged at 6"*, never with a list and never with a stated convention. So a reader cannot audit it from the document that states it, and two of the six are absent outright: the boost-ceiling refutation (the tokens `3.17`, `68.5`, `0.927` and the phrase "boost ceiling" each have **zero** occurrences in the whitepaper body) and the Bell/CHSH substrate refutation (`CHSH`, `no-signaling`, `Kuramoto`: **zero** occurrences; the only "Bell" in these sources is the exposition of Bell's theorem in §5.4 and a prior-art specificity list). The six, each with the statistic it was registered on and the data it used, are: **(1)** RAR transition shape at γ = 2 — marginalised BIC on SPARC, ΔBIC = +184; **(2)** BTFR slope (TEST-09) — SPARC, |Δn| = 0.41 against a registered bar of 0.3; **(3)** dwarf DM-fraction ceiling (TEST-10) — SPARC, 69% of galaxies above f_DM = 1 − Ω_m = 0.685; **(4)** RAR environment dependence (TEST-08) — SPARC × Cosmicflows-4, r² = 0.0001 against a registered > 0.20; **(5)** the SPARC × Cassini joint squeeze (TEST-11) — +17.95σ; **(6)** the observer-relative Bell/CHSH substrate bet (B1) — S = 1.98 ≤ 2 on both arms. +> **[AMENDED 2026-08-08] The "6" is a count of test IDs; this whitepaper enumerates it nowhere, and two of its six members are described nowhere in it.** The number is asserted four times across these sources — in the paragraph above, in the a₀(z) note below, in the executive summary and in the conclusion — always as *"unchanged at 6"*, never with a list and never with a stated convention. So a reader cannot audit it from the document that states it, and two of the six are absent outright: the boost-ceiling refutation (the tokens `3.17`, `68.5`, `0.927` and the phrase "boost ceiling" each have **zero** occurrences in the whitepaper body) and the Bell/CHSH substrate refutation (`CHSH`, `no-signaling`, `Kuramoto`: **zero** occurrences; the only "Bell" in these sources is the exposition of Bell's theorem in §5.4 and a prior-art specificity list). The six, each with the statistic it was registered on and the data it used, are: +> +> **(1)** RAR transition shape at γ = 2 — marginalised BIC on SPARC, ΔBIC = +184 *[CHECKED 2026-11-09: the +184 is not a marginalised BIC. The frozen 2026-07-22 pipeline that reproduces +184.0 holds Υ_disk = 0.5 (Υ_bulge = 0.7) for both laws, the same fixed convention the M/L paragraph below records for the SPARC rows; S661's own description, "M/L + distance marginalized", is not re-checked against S661's code here. With one global Υ_disk per law the same comparison gives +46.2 by point and +2.73 with N → N_gal, about 2σ on a galaxy-block bootstrap (95% [−26.1, +132.5]), and the reading rule fixed before that run reads the row as disfavoured, not refuted. So the statistic this item names was never the number printed beside it. Whether the row stays among the six is the open counting-rule question (gates on the project lead); the range already carries −1 for it. Count HELD at 6. Source: `6fe7b696` (i); the 11-09 UPDATE on the 10-18 SCOPED note above; `gamma2_global_upsilon_per_law.py` (PREREG `5a95614`), re-run by the Publisher 2026-11-10 with byte-identical output.]*; +> +> **(2)** BTFR slope (TEST-09) — SPARC, |Δn| = 0.41 against a registered bar of 0.3 *[CHECKED 2026-11-06: the registered rule compares the point estimate with 0.3, and on that rule TEST-09 fires. A paired galaxy bootstrap (B = 4000, both slopes refit on each resample, error correlation ρ = +0.14) gives σ(Δn) = 0.112. So |Δn| = 0.406 is 3.6σ from zero but only 0.95σ above the bar, and 16% of resamples fall at or under 0.3. A rule that required the bar itself to be excluded at 2σ would not fire. This is a second convention on which TEST-09 leaves the count, beside the floor convention in the `Ω_m` versus `Ω_m/Ω_b` paragraph below. Count HELD at 6. Source: site explorer `da1308c` (pre-registered at `c9ff6ab`), re-run by the Publisher 2026-11-06 with byte-identical output and its identity control passing.]*; +> +> **(3)** dwarf DM-fraction ceiling (TEST-10) — SPARC, 69% of galaxies above f_DM = 1 − Ω_m = 0.685 *[CHECKED 2026-11-09: the registered rule has no significance term, so the row fires as registered and the count holds. As a bound on the boost cap it is weaker than the sample maximum suggests. With distance, inclination, Υ and gas profiled on the outer half of each curve (post-hoc), the framework's B = 3.175 is excluded at χ² ≥ 9 by 28 of 153 discs and by 10 after a 153-disc correction (χ² ≥ 18.4). Ω_m/Ω_b = 6.39 is excluded by 2 discs at χ² ≥ 9 and by none after the correction. The largest excluded cap is about 7.5 at χ² ≥ 9 and about 5.7 after the correction, both set by one disc (NGC 2915). The pre-registered all-radii scan reaches 14 and 9 instead, but only through UGC 06667's inner points, where SPARC lists V_gas = 0.0 and the gas nuisance cannot act. Limits: no asymmetric drift, beam smearing or non-circular motions, and χ² is treated as χ²-distributed despite one-sided residuals. A joint population statistic is unrun and could be tighter. Source: `PREDICTIONS.md` back-annotation `82d9a3a8` (ii); scripts `synchronism-site/maintainer/scripts/test10_bmax_3sigma_scan.py` (pre-registered at `0be9704`) and `test10_bmax_3sigma_diag_posthoc.py`, both re-run by the Publisher 2026-11-09 with byte-identical output.]* *[CHECKED 2026-11-16: "by none after the correction" is a result about discs. It does not mean the 6.39 cap survives the dynamics. The archive's own pressure-supported table (site explorer 2026-08-23) gave the boost Draco and Crater II require as 101 and 60, but tested it only against caps of 3.17 and 13.7. Tested at 6.39 in σ-space, with no gas term and no light-deflection rule: Draco z = 5.7 at fiducial M/L and 4.9 at M/L = 4; Crater II 6.1 and 4.9. Both clear the six-system Šidák threshold of z = 3.51. The escapes are M/L ≥ 8.4 (Crater II) to 11.6 (Draco), where old metal-poor populations sit near 1–3, or a measured σ inflated by tides by 1.26–1.33×. Either way the open escape is non-equilibrium, not a missing light rule. Draco is also a failure for MOND+EFE, so it counts against the cap and not for MOND; Crater II is where McGaugh's 2016 a-priori MOND prediction held. One correction to the source, found on re-run: the 08-23 table and this test use σ_N² = G·M_1/2/(4·r_1/2) with r_1/2 labelled three-dimensional. Wolf et al. 2010 give the coefficient 3 for the three-dimensional radius; 4 belongs with the projected R_e. So the archive's required boosts are 4/3 too high (Draco 76, Crater II 45). At the coefficient 3 and M/L = 4 the exclusion still holds (Draco z = 4.48, Crater II 4.21), the escape M/L values fall to 6.3 (Crater II) and 8.7 (Draco), still above old populations, and the largest cap Draco excludes after trials falls from about 14.7 to about 11.0. With L_V also raised 30% as a stress test, Crater II sits at 3.54, on the threshold, and Draco at 4.04. Limits: quoted σ errors only, with no anisotropy, r_1/2 or distance uncertainty, and an equilibrium estimator. Count HELD at 6, because TEST-10 already fires as registered. Source: `PREDICTIONS.md` back-annotation `f971b1bc` (i); script `synchronism-site/maintainer/scripts/cap6p39_vs_dsph.py` (pre-registered at site `c7d729c`, not blind; its identity control reproduces all six 08-23 values), re-run by the Publisher 2026-11-16 with byte-identical output; the coefficient and luminosity sensitivity is `simulations/publisher_20261116_dsph_cap_estimator_sensitivity.py`, whose k = 4 rows reproduce the source.]* *[CHECKED 2026-11-20: two points about the numbers in the 11-16 note. (a) Which threshold. The escapes it gives (M/L ≥ 8.4 and 11.6 at coefficient 4, 6.3 and 8.7 at coefficient 3; tidal inflation 1.26–1.33×) are the source script's passes at a local 3σ, although the sentence before them names the trials threshold 3.51. At 3.51 the escapes are M/L ≥ 5.3 (Crater II) and 6.9 (Draco) at coefficient 3 (7.0 and 9.2 at 4), and a σ inflation of 1.15× and 1.31× at M/L = 4, with the error scaled with σ. The site's 2026-11-19 figures are these. Both sets are right for their own bar, and old metal-poor populations (M/L near 1–3) sit below both. (b) Walker+2009. The site's 11-19 re-read also calls the exclusion "estimator-conditional at its edge": under Walker+2009, Crater II falls to z = 2.9 and Draco to 3.7 at M/L = 4. That row sets Walker's mass, which is the mass inside a sphere of radius R_e (the projected half-light radius), equal to half the light. For a Plummer profile that sphere holds 0.354 of the light, not 0.5. With the light matched to the aperture, Walker becomes the half-light form with coefficient 2.71 (Wolf's is 3), and at M/L = 4 it gives Crater II z = 3.96 and Draco 4.31, both past 3.51 (escape M/L at 3.51: 4.7 and 6.2). So the exclusion does not change between the two published estimators. It stays conditional on M/L, on equilibrium (the site reports Crater II as tidally disturbed) and on the inputs, and the L_V × 1.3 stress test above, which puts Crater II at 3.54, is still its edge. Limits: a Plummer profile for the aperture fraction; Walker's estimator taken in its published form, not re-derived. Count HELD at 6. Source: `PREDICTIONS.md` back-annotation `b26f6fa9` (i); site script `synchronism-site/maintainer/scripts/cap6p39_vs_dsph_estimator.py`, whose rows `simulations/publisher_20261120_dsph_walker_aperture.py` reproduces as its identity control before adding the aperture-matched row.]*; +> +> **(4)** RAR environment dependence (TEST-08) — SPARC × Cosmicflows-4, r² = 0.0001 against a registered > 0.20 *[CHECKED 2026-11-09: 0.0001 is a point estimate (r = +0.012, N = 141). The 95% upper limit is r² ≈ 0.03 (Fisher 0.031, bootstrap 0.030), which is 2.9× under the kill bar of r² < 0.09 and 6.4× under the claimed 0.20. The margin is that, not the hundreds of times the point estimate suggests. The kill still fires. The registered sentence, "explains > 20% of RAR scatter", names no denominator. The same back-annotation reads it against intrinsic scatter as about 0.04 of total variance, which the upper limit would exclude only narrowly; that 0.04 is relayed, not re-derived here. Count HELD at 6. Source: `82d9a3a8` (iv); recomputed from the tracked per-galaxy table in `simulations/publisher_20261109_test08_upper_limit.py`.]*; +> +> **(5)** the SPARC × Cassini joint squeeze (TEST-11) — +17.95σ *[CHECKED 2026-11-07: the +17.95σ is at γ = 0.489, and the registered joint test evaluated Cassini only inside SPARC's point-level interval (0.425–0.600), marking higher γ "not required". Re-run with a₀ profiled from SPARC at each γ, at all three registered external fields (2.00, 2.32, 2.48 ×10⁻¹⁰ m s⁻²), Cassini fails at **14.5–19.3σ for every γ from 0.3 to 3**, nearly flat in γ. So this row does not depend on the γ ≈ ½ pin, and in the decomposition at "(4) The six refutations" above, `C_g`'s two entries no longer sit at different γ: at γ = 2 the RAR shape (ΔBIC +184) and Cassini (14.9–16.8σ) both exclude it. An earlier ledger statement that the compander "passes at γ ≳ 1.5–2" held a₀ fixed at 1.03×10⁻¹⁰; this paper never printed it. The row stays conditional on a modified-Poisson (QUMOND) completion, as before, and the count is HELD at 6. Source: `PREDICTIONS.md` back-annotation `330d49d1` (i); script `synchronism-site/maintainer/scripts/cassini_q2_at_sparc_profiled_a0_high_gamma.py`, re-run by the Publisher 2026-11-07 with byte-identical output and its identity control (+17.96σ at γ = 0.489) reproducing. [UPDATE 2026-11-08: two limits on the sentence above. (a) "From 0.3" reaches too far. The γ = 0.3 and 0.35 rows are not converged: z changes with small shifts in a₀ (up to 0.3%) and with the solver order, and a 12-draw scan spans 8–28σ at γ = 0.3 and 10–24σ at γ = 0.35. Every draw stays above 5σ, so the verdict holds there, but those rows carry no usable number. The 14.5–19.3σ range stands for γ 0.4–3, where the same scan moves z by at most 0.6σ (single-draw spikes at the highest external field, γ = 0.4 and 0.489) and by at most 0.1σ at γ ≥ 0.6. (b) Every σ here pins Υ at 0.5 (disc) and 0.7 (bulge), and z depends strongly on that zero point through a₀. Scale both by one factor and re-profile a₀, on the 134 bulge-free galaxies: γ = 0.489 and γ = 1 do not pass at any factor up to 3; γ = 3 passes at Υ_d ≈ 1.22 and γ = 2 at Υ_d ≈ 1.40. That is 2.4–2.8 times the stellar-population value, and 36–42% of galaxies then have `g_bar` above `g_obs` by more than 20% at some radius (6% at the pinned value), which a law with boost ≥ 1 cannot fit. γ = 3 at that scale costs +480 point-level against γ = 0.489, 2.9σ at galaxy level. So this row does not depend on the γ pin, but it does depend on Υ_d near 0.5. MOND's own interpolating family behaves differently (δ sets how sharp the transition is; δ = 1 is McGaugh's function): on the same sample, δ = 6 passes at Υ_d ≈ 0.54 for +128, 1.8σ (Desmond, Hees & Famaey 2024 report 1.9σ with Υ freedom and no bulges, on a richer instrument). The row's shared MOND root therefore holds at δ = 1 and γ ≈ ½ only. AQUAL is untested for both families. Count HELD at 6. Source: `PREDICTIONS.md` back-annotation `cf84bb8c` (ii)–(iii) and site explorer `2db4494` (pre-registered at `10d0bdf`; the Υ scan and over-maximal check are post-hoc); all re-run by the Publisher 2026-11-08 with matching values. The convergence scan is new in this pass and is in the 2026-11-08 run log.]]*; +> +> **(6)** the observer-relative Bell/CHSH substrate bet (B1) — S = 1.98 ≤ 2 on both arms. > > **Measured, not asserted: (2) and (3) fire on the same galaxies, by algebraic identity.** The framework's apparent DM fraction and its acceleration boost are the same quantity in two variables — `f_DM = 1 − (V_bar/V_obs)² = 1 − g_bar/g_obs = 1 − 1/B` — so TEST-10's criterion `f_DM > 1 − Ω_m` and TEST-09's ceiling `B > 1/Ω_m` are the *same inequality*, not two corollaries of a shared root. Verified per galaxy on the in-repo SPARC mass models (`simulations/publisher_20260808_test09_test10_independence.py`, 123 galaxies, Q ≤ 2, i > 30°): `max|f_DM − (1 − 1/B)| = 2.2×10⁻¹⁶`, and the two criteria select **93 and 93 galaxies with 0 disagreements** at the outermost measured point, **88 and 88 with 0 disagreements** at the error-weighted mean of the outer three. *(93/123 = 76% here rather than TEST-10's published 69% only because TEST-10 proper runs the wider 153-galaxy set; it is the same criterion on TEST-09's narrower cut, which is the point — the comparison is between criteria on one sample, not between published fractions.)* Deleting every galaxy that fires TEST-10 and refitting the BTFR, TEST-09's registered kill **does not fire** on what remains: observed n = 3.39 ± 0.23 against the framework's 3.38 ± 0.18, deviation **0.01 (0.0σ)** versus a threshold of 0.3, where the full sample gives 3.79 ± 0.10 vs 3.35 ± 0.07, deviation 0.44 (3.5σ). *(0.44 rather than the ledger's 0.41 because this run compares against the like-for-like outer-3 estimator applied identically to all three models rather than the SPARC catalogue V_flat; both fire.)* > @@ -164,7 +182,7 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > **What the discriminator actually is, restated.** The 08-26 form — "a live external discriminator gating on execution" — is withdrawn: the discriminating numbers are printed in the cited paper. What genuinely gates on execution is narrower and worth doing: RG's `ε₀` is fitted *jointly* with `ρ_c` and `q` per system, so a refit at **fixed** `ε₀ = 0.315` with `ρ_c, q` free is the honest test of whether the derived floor survives, and Cesare et al. 2020's 30 DiskMass galaxies are the right sample because vertical dispersions are RG's own method. Stated as a proposal, not a result: nothing has been refit here. *[2026-08-28: the paper is now read and the prior expectation for that refit is known — Cesare+2020's own joint fit prefers `ε₀ = 0.661 ± 0.007` on this sample with `ρ_c, q` free, so a refit pinned at 0.315 starts 49σ (formal) from their optimum; the honest statistic is the Δχ² of the pinned fit, and that is still unrun.]* *(Verified: `simulations/publisher_20260827_rg_floor_is_not_universal.py`, exit 0 — including the identity `C_Ω ≡ ε` re-checked against the **verbatim** Eq. (4.1) rather than the relayed form, sympy zero in both log bases, numeric max 2.2×10⁻¹⁶.)* *[2026-08-29: the rotation-curve-only cousin of this refit has now been run — under L2 on 153 SPARC discs the free floor lands at 0.20–0.25, 25–35% below the derived 0.315 and where M&D's two spirals put it; see the 2026-08-29 amendment below. The DiskMass refit itself, with vertical dispersions, is still unrun.]* > -> **The covariant branch is published too, and the citation as routed is wrong.** The covariant formulation is **Sanna, Matsakos & Diaferio 2023** (A&A **674**, A209; arXiv:2109.11217) — scalar–tensor, with `φ` twice the permittivity in the weak field, verified at source. [**CHECKED 2026-10-23: the identification φ = 2ε holds formally, but this completion has no working dynamics.** With `W = −1` and `V = −Ξφ` the paper's trace equation is exactly linear, (∇² − ∂²/∂t²)φ − 2Ξφ = 8πG T (c = 1). That is ω = −1 Brans–Dicke with a scalar of Hubble-scale range (2.2 Gpc). A linear equation cannot screen, so the paper's chameleon argument does not apply. The consequences: γ_PPN = (1+ω)/(2+ω) = 0 (Cassini, ~4×10⁴σ); η_N = 4β − γ − 3 = 1 (LLR Nordtvedt, ~2×10³σ); and the paper's own flat-FLRW system forces |Ġ/G| ≥ 0.99 H₀ (LLR, ≥ 900σ at every admissible point, ~2,600σ at its best fit). Its near-source solution (their Eq. 26) varies φ by the Newtonian potential over c², ≲ 10⁻⁴ across a galaxy. RG needs an O(1) variation in permittivity, so the covariant theory's galaxies are Newtonian with a renormalized G. **Covariant RG is a comparison class without working dynamics, not a viable completion.** Re-derived here: the PPN values from the standard Brans–Dicke formulas, today's d ln φ/d ln a and Ġ/G at the best fit (Ω₀, Ω_Ξ0) = (0.31, 0.65) from the FLRW system as the finding quotes it, and the 2.2 Gpc range. Relayed, not re-run: the trace-equation derivation, the ≥ 900σ floor over the admissible range, and the galaxy-scale Yukawa solve. Source: `synchronism-site/explorer/findings/covariant-refracted-gravity-is-omega-minus-one-brans-dicke-fails-cassini-llr-and-never-reproduces-rg.md` (PREREG site `4e3898f` + `1244148`), routed via `Research/proposals/gamma_is_identified_only_jointly_with_rho_crit_and_covariant_rg_closes_the_rg_fork_20261023.md` §3.] The routing note gives "Sanna, Pipino, Diaferio et al."; **Pipino is not an author on that paper**, corrected here **[and the parenthetical that stood here — "he is on the 2022 E0-galaxy study" — was itself written from memory and is WRONG: the E0 study (Cesare, Diaferio & Matsakos, A&A 657 A133, arXiv:2102.12499) has three authors and Pipino is on no refracted-gravity paper indexed by arXiv (author-list query 2026-08-29, 0 hits). A correction that mis-cites is the same defect it corrects; withdrawn 2026-08-29.]** And **not in either source, surfaced by this lane's own gate**: Gervani, Diaferio, Pace & Sanna 2026 (arXiv:2601.22937) integrate the linear perturbations of a Brans–Dicke theory that "in the weak-field limit reduces to Refracted Gravity," with baryons alone and neither dark matter nor dark energy, and find **structure formation delayed to z < 1, in disagreement with the observation of formed galaxies at much larger redshifts.** That is a published cosmological verdict on the same covariant branch §5.15 below constructs as **Completion B** and independently excludes at Δχ² ≥ +79 against DESI DR2 + CMB + SN. +> **The covariant branch is published too, and the citation as routed is wrong.** The covariant formulation is **Sanna, Matsakos & Diaferio 2023** (A&A **674**, A209; arXiv:2109.11217) — scalar–tensor, with `φ` twice the permittivity in the weak field, verified at source. [**CHECKED 2026-10-23: the identification φ = 2ε holds formally, but this completion has no working dynamics.** With `W = −1` and `V = −Ξφ` the paper's trace equation is exactly linear, (∇² − ∂²/∂t²)φ − 2Ξφ = 8πG T (c = 1). That is ω = −1 Brans–Dicke with a scalar of Hubble-scale range (2.2 Gpc). A linear equation cannot screen, so the paper's chameleon argument does not apply. The consequences: γ_PPN = (1+ω)/(2+ω) = 0 (Cassini, ~4×10⁴σ); η_N = 4β − γ − 3 = 1 (LLR Nordtvedt, ~2×10³σ); and the paper's own flat-FLRW system forces |Ġ/G| ≥ 0.99 H₀ (LLR, ≥ 900σ at every admissible point, ~2,600σ at its best fit). Its near-source solution (their Eq. 26) varies φ by the Newtonian potential over c², ≲ 10⁻⁴ across a galaxy. RG needs an O(1) variation in permittivity, so the covariant theory's galaxies are Newtonian with a renormalized G. **Covariant RG is a comparison class without working dynamics, not a viable completion.** [NOTE 2026-11-11: the σ figures in this note are this program's own unpublished computation (site explorer, 2026-10-22). The authors present the theory as viable, and no published analysis is cited here for the failure.] Re-derived here: the PPN values from the standard Brans–Dicke formulas, today's d ln φ/d ln a and Ġ/G at the best fit (Ω₀, Ω_Ξ0) = (0.31, 0.65) from the FLRW system as the finding quotes it, and the 2.2 Gpc range. Relayed, not re-run: the trace-equation derivation, the ≥ 900σ floor over the admissible range, and the galaxy-scale Yukawa solve. Source: `synchronism-site/explorer/findings/covariant-refracted-gravity-is-omega-minus-one-brans-dicke-fails-cassini-llr-and-never-reproduces-rg.md` (PREREG site `4e3898f` + `1244148`), routed via `Research/proposals/gamma_is_identified_only_jointly_with_rho_crit_and_covariant_rg_closes_the_rg_fork_20261023.md` §3.] The routing note gives "Sanna, Pipino, Diaferio et al."; **Pipino is not an author on that paper**, corrected here **[and the parenthetical that stood here — "he is on the 2022 E0-galaxy study" — was itself written from memory and is WRONG: the E0 study (Cesare, Diaferio & Matsakos, A&A 657 A133, arXiv:2102.12499) has three authors and Pipino is on no refracted-gravity paper indexed by arXiv (author-list query 2026-08-29, 0 hits). A correction that mis-cites is the same defect it corrects; withdrawn 2026-08-29.]** And **not in either source, surfaced by this lane's own gate**: Gervani, Diaferio, Pace & Sanna 2026 (arXiv:2601.22937) integrate the linear perturbations of a Brans–Dicke theory that "in the weak-field limit reduces to Refracted Gravity," with baryons alone and neither dark matter nor dark energy, and find **structure formation delayed to z < 1, in disagreement with the observation of formed galaxies at much larger redshifts.** That is a published cosmological verdict on the same covariant branch §5.15 below constructs as **Completion B** and independently excludes at Δχ² ≥ +79 against DESI DR2 + CMB + SN. > > **Stated as an open question rather than a corroboration, because it has not been checked.** Whether §5.15's Completion B and Gervani et al.'s theory are the *same* theory is **not settled here** — §5.15's quasi-static pinning `C_eff(x₀) = Ω_m` has not been compared to their construction, and §5.15 separately names a "density-pinned **massive** scalar" as its one unexecuted covariant branch, while Sanna et al.'s potential `V(φ) = −Ξφ` is *linear*, which is not a mass term **[2026-10-23: in the trace equation it is one, but of Hubble range, so it is still not a density-pinned massive scalar. See the CHECKED note above.]**. Settling this is the first blocking action on the recommendation opened today. If they coincide, this section's completion-B exclusion becomes an independently-routed corroboration of a published negative — background likelihood against linear growth, two different observables on one theory — which is a stronger and far more useful statement than an internal no-go. > @@ -195,8 +213,8 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > **[AMENDED 2026-09-01] The floor has now been *measured* rather than scanned, and it is not a constant: under the field equation on 153 SPARC discs the best universal `ε₀` is 0.220 (ceiling 4.55), but fitted one galaxy at a time — parameter-matched against MOND's `a₀` — it scatters 1.2–1.8 dex and tracks baryonic mass at ρ_s = +0.76, while `a₀` measured identically tracks it at +0.07.** The site explorer's 2026-08-30 session (`synchronism-site` `c66718e`; finding `the-ceiling-is-a-measurement-and-the-measurement-excludes-it.md`, scripts `epsilon0_free_the_ceiling_rescue.py`, `epsilon0_left_edge_convergence.py`, `universality_eps0_vs_a0.py`) first recovered its own crashed 08-28 run — the outputs this section re-derived on 08-29 are now committed there and agree — then extended the `ρ_c` grid seven decades leftward until every floor row turned over, and read off a **bracketed interior minimum at `ε₀ = 0.220`, `ρ_c = 3.5×10⁻⁶ M☉ pc⁻³`**, χ²/N = 126.5 at `Υ_d = 0.5` fixed, the same in the framework and the RG functional forms and stable under `Υ` profiling (ratio to MOND 2.42× → 2.55×). Three consequences, then the finding. > -> - **The rescue "if `ε₀` is free, TEST-09/TEST-10 are measurements of `ε₀`" (site visitor, 08-30) is excluded by the measurement it asked for.** The tail galaxy's `f_DM = 0.927` needs `B ≥ 13.7`, i.e. `ε₀ ≤ 0.073`; the rotation curves put that at Δχ²/N = **+67.8** above the optimum. Two constraints on one constant that do not intersect — a squeeze of the TEST-25 kind, and cosmology-free. The site's published TEST-10 wording, *"no candidate cosmic ratio supplies B = 13.7"*, assumes `ε₀` must be a cosmic ratio, which the prior art denies (M&D fit it); that wording is a non-sequitur and the explorer has withdrawn it in favour of the squeeze. *This whitepaper never carried the cosmic-ratio argument* — the 08-08 amendment above states the two IDs as the identity `f_DM = 1 − 1/B` — and `PREDICTIONS.md` does not carry it either (`git grep` 2026-09-01: 0 hits in-repo); the landing site is the site alone. **Count HELD at 6.** -> - **The `Ω_m` versus `Ω_m/Ω_b` convention dispute is measured, and it is small.** The derived `1/Ω_m = 3.17` (ε₀ = 0.315) sits **+7.3** above the optimum and `Ω_m/Ω_b = 6.40` about +12, against **+74** to parameter-free MOND. The framework's guessed floor is the second-best row in the profile and beats every published Refracted Gravity value on SPARC (DMS 0.56: +137; DMS-universal 0.661: +220) — which is not a refutation of RG but a measurement that its DiskMass parameters (vertical dispersions, different `Υ`, `D`, `i` treatment) **do not transfer** to SPARC rotation curves. +> - **The rescue "if `ε₀` is free, TEST-09/TEST-10 are measurements of `ε₀`" (site visitor, 08-30) is excluded by the measurement it asked for.** The tail galaxy's `f_DM = 0.927` needs `B ≥ 13.7`, i.e. `ε₀ ≤ 0.073`; the rotation curves put that at Δχ²/N = **+67.8** above the optimum. Two constraints on one constant that do not intersect — a squeeze of the TEST-25 kind, and cosmology-free. The site's published TEST-10 wording, *"no candidate cosmic ratio supplies B = 13.7"*, assumes `ε₀` must be a cosmic ratio, which the prior art denies (M&D fit it); that wording is a non-sequitur and the explorer has withdrawn it in favour of the squeeze. *This whitepaper never carried the cosmic-ratio argument* — the 08-08 amendment above states the two IDs as the identity `f_DM = 1 − 1/B` — and `PREDICTIONS.md` does not carry it either (`git grep` 2026-09-01: 0 hits in-repo); the landing site is the site alone. **Count HELD at 6.** *[CHECKED 2026-11-09: `B ≥ 13.7` is one galaxy's face value (NGC 2915, one of 153) with no significance attached, so "+67.8" and "do not intersect" hold at that face value only. Using the bound in count item (3) above, the tail needs B of about 7.5 at χ² ≥ 9 (ε₀ ≤ 0.134), or about 5.7 after the 153-disc correction (ε₀ ≤ 0.175). The same ε₀ profile puts those between +12 and +30.8 (rows ε₀ = 0.156 and 0.120) and between 0 and +12 (rows 0.220 and 0.156). The framework's own ε₀ = 0.315 costs +7.3 on that profile. So at the corrected bound the squeeze costs about what the derived value already costs. No σ is attached to any of these: the profile fixes Υ and does not marginalise distance or inclination, while the bound profiles all three. This narrows the case against this one rescue, not the case against the class: its best ε₀ scores χ²/N = 126.5 against MOND's 52.2 on the same points. Count HELD at 6. New in this pass: the brackets are read from the committed table in the site finding `the-ceiling-is-a-measurement-and-the-measurement-excludes-it.md`, and the profile was not re-solved at 0.134 or 0.175.]* +> - **The `Ω_m` versus `Ω_m/Ω_b` convention dispute is measured, and it is small.** The derived `1/Ω_m = 3.17` (ε₀ = 0.315) sits **+7.3** above the optimum and `Ω_m/Ω_b = 6.40` about +12, against **+74** to parameter-free MOND. The framework's guessed floor is the second-best row in the profile and beats every published Refracted Gravity value on SPARC (DMS 0.56: +137; DMS-universal 0.661: +220) — which is not a refutation of RG but a measurement that its DiskMass parameters (vertical dispersions, different `Υ`, `D`, `i` treatment) **do not transfer** to SPARC rotation curves. *[CHECKED 2026-10-29: "small" describes rotation-curve fit quality, not what the two floors presuppose, and they presuppose different frameworks. `1/Ω_m` is the cosmology sector's calibrated floor `C₀ = Ω_m`, and that `Ω_m` contains about 0.27 of cold dark matter. `Ω_m/Ω_b` is the floor of a galaxy sector that replaces CDM. The ledger's 2026-09-22 block ran the dark-energy sector with ρ_m = ρ_b: in each of the two tested variants, none of 400 logarithmically sampled γ values in 10⁻⁴…3 gives both acceleration (q₀ < 0) and the recombination dark/baryon ratio 5.36, even to a factor 2 (best among accelerating samples: 1.54). An independent rerun returned all 400 samples per variant with no caught exceptions or non-finite result rows; this finite scan does not exclude unsampled γ or other model variants. The two floors also give opposite TEST-09 verdicts (2026-09-18 sweep, TEST-09's own cuts and V_flat estimator): `1/Ω_m` gives n = 3.35, |Δn| = 0.41, fires; `(Ω_m−Ω_b)/Ω_b` gives 0.30, on the strict threshold, does not fire; `Ω_m/Ω_b` gives 0.26, does not fire. (Unrounded, re-run 2026-10-31: |Δn| = 0.406 / 0.297 / 0.260. The middle value is 0.003 under the strict bar, not on it, and the 0.3 edge sits at B_max ≈ 5.31 (interpolated from the free scan) rather than at 5.389; `PREDICTIONS.md` 2026-10-30 (ii). Verdicts unchanged.) So the branch on which TEST-09 does not fire belongs to a variant whose tested dark-energy samples did not meet both criteria, and the branch on which it fires is the framework as written. This ranks the conventions under the tested assumptions, not a continuous exclusion. It is not a new refutation, and it does not settle whether a convention-dependent kill counts in the headline, a recount that gates on the project lead. Count HELD at 6. Sources: `PREDICTIONS.md` 2026-09-18, 2026-09-22 and 2026-10-27 (ii) blocks; `Research/proposals/citable_register_ranks_by_framework_not_transferability_and_omega_m_plays_two_roles_20261027.md`. Both site scripts (`synchronism-site/maintainer/scripts/which_C_carries_the_floor.py`, `de_sector_without_cdm.py`) were re-run by the Publisher on 2026-10-29 at exit 0, and their output matches the committed files.]* > - **A trap the explorer caught in its own draft, recorded because it is the first over-*affirmation* in this sector's history.** Freeing `(ε₀, ρ_c)` per galaxy gives χ²/N = 18.7 against MOND's 52.9, "wins 85%, p = 1.4×10⁻¹⁹" — on **306 free parameters against 0**. Parameter-matched (one constant per galaxy each), the class loses 79% of galaxies: χ²/N 39.7 vs 10.3. > > **The finding, reproduced here before inscription.** `simulations/publisher_20260901_eps0_universality_reproduction_and_uncensored.py` imports the site's committed solver, re-derives **11 of 11** printed values of the universality test within 2% (nine of them to three decimals: global 126.53 / 52.22, per-galaxy 39.70 / 10.30, scatter 1.197 / 0.619 dex, ρ_s +0.758 / +0.073 …), and then runs the first step the site's own seeded topic asks for. **(i) Uncensored grid** (`ε₀ ∈ [0.005, 0.98]`, 41 values, and `a₀` widened symmetrically): ρ_s(`ε₀`, log M_bar) = **+0.757** — *unchanged*, so the site's stated "+0.758 is a lower bound because of the 42% censoring" is true but idle: lifting the censoring does not raise it. What lifting it does show is that **33% of SPARC discs still sit at the new floor `ε₀ = 0.005` (ceiling 200)** — a third of the sample rejects any ceiling at all at the global `ρ_c`, the per-galaxy spread widens to **1.78 dex** (`a₀`: 0.64), and the class still loses 76% of galaxies parameter-matched (36.0 vs 10.3). **(ii) Permutation null**, 20,000 galaxy-label shuffles: the largest |ρ_s| the null ever produced is **0.317** (99.9th percentile 0.268) against the observed 0.757, p ≤ 5×10⁻⁵; the `a₀` control sits *inside* the null (35–38% of shuffles exceed its +0.073). **(iii) The relation as a relation**: `log ε₀ = k·log M_bar + b` fits `k = +0.42` (site grid) to `+0.61` dex/dex (uncensored) and leaves **0.36–0.52 dex rms** about the line — it removes 32% of the raw variance; `a₀`'s slope is 0.04–0.07 and removes 1%. No decision rule was pre-registered for what scatter would make this "a stated relation of the theory", so none is applied; for scale, the RAR's own scatter, the benchmark the seed names, is 0.11 dex, and this relation is 3–5× broader. **Prior-art gate, run today on the arXiv PDF of Cesare et al. 2020:** the paper tests no RG parameter against any galaxy property (its Table 5 correlations are RAR *residuals*), and its verdict on the same question is the opposite one — per-galaxy `ε₀ = 0.56 ± 0.19` on 30 DiskMass galaxies, with the standard deviation *"smaller than or comparable to the mean uncertainties"*, so *"the different values that we find for different galaxies can in principle be ascribed to statistical fluctuations"* (§5). That verdict was reached with three RG parameters plus `Υ` and `h_z` free per galaxy on a narrow-mass sample; the SPARC result holds `ρ_c` fixed and spans ~4 dex in M_bar. The two are not the same measurement, and the tension between them is **open, not adjudicated here** — a draft must carry both. @@ -221,7 +239,7 @@ A breakthrough discovery: MOND and Synchronism are the same physics in different > > **What survives is one residue, and it is honest.** Treatment B — `(ε₀, ρ_c)` co-fitted per galaxy, re-solved from scratch because the 08-30 cache never held a per-galaxy `ρ_c` at all — *does* leave a mass slope MOND does not induce: **c = +0.136 [+0.083, +0.191]**, Freedman–Lane p < 5×10⁻⁵, robust σ_resid 0.197 dex, which clears even the "tight" bar. It is nevertheless priced at nothing, for three reasons, and the third is the one worth keeping. *(i)* Stated globally it buys nothing: give the class four parameters — `ε₀ = A·M^k` **and** `ρ_c = A₂·M^k₂` — and the optimiser sets **`ε₀` universal** (k = +0.000, χ²/N 81.8) and pushes every bit of mass dependence into a knee density falling three decades per decade of mass. The per-galaxy signal is degenerate with `ρ_c`, not with physics. *(ii)* Matched four-against-two the class still loses **1.57×** and still wins only 37% of discs. *(iii)* It has an untested nuisance explanation that **R2's guard is blind to by construction**: a density-keyed law sees the midplane density `ρ_mid ∝ Σ/h`, so the scale-height prescription (Bershady, `h = 0.196 R_d^0.633`) and the fixed `Υ = 0.5` enter `C(ρ)` directly, while `g_bar` never sees h at all. MOND cannot pick that channel up, so a flat `a₀` cannot exonerate it. **Untested, not refuted.** The check is an afternoon — refit treatment B with h × {0.5, 2} and Υ ∈ {0.3, 0.7} and ask whether c moves by its own width — and it is now the galaxy sector's only genuinely open thread. > -> **The ceiling also dies independently, on data this section has never cited.** Brouwer et al. 2021 (weak-lensing radial acceleration relation, KiDS-1000; A&A **650**, A113; arXiv:2106.11677 — Verlinde among the authors) stack isolated lens galaxies and report that the measurement *"extends the radial acceleration relation by 2 decades into the low-acceleration regime"* and that *"the measured RAR agrees well with the MG predictions."* Take R3's best relation at face value — `ε₀ = 0.119·(M/10¹⁰)^{0.325}`, so the ceiling `B_max = 1/ε₀` runs from 17.8 at 10⁹ M☉ to 4.0 at 10¹¹ — and compare with MOND's ν where lensing actually reaches. Two decades below the SPARC floor is `g_bar ≈ 10⁻¹⁴ m s⁻²`, where ν ≈ 110: the ceiling falls short by **6× at 10⁹ M☉, ~13× at 10¹⁰, 28× at 10¹¹**. **A mass-dependent ceiling does not rescue the ceiling** — the failure grows with mass, because the relation lowers the ceiling exactly where the lensing sample is heaviest. *Screen coverage, stated because it bounds the claim: **abstract-level only in this lane; the full text was not read here.** The "2 decades" is the abstract's own wording. The site's companion table instead assumes reach to `g_bar = 10⁻¹⁵` and quotes deficits up to 87×; the abstract does not support that reach, so the conservative column is the one inscribed above.* The paper's own caveat is circumgalactic gas, and it is not close to sufficient here: a factor-2 correction on `g_bar` moves ν by √2. +> **The ceiling also dies independently, on data this section has never cited.** Brouwer et al. 2021 (weak-lensing radial acceleration relation, KiDS-1000; A&A **650**, A113; arXiv:2106.11677 — Verlinde among the authors) stack isolated lens galaxies and report that the measurement *"extends the radial acceleration relation by 2 decades into the low-acceleration regime"* and that *"the measured RAR agrees well with the MG predictions."* Take R3's best relation at face value — `ε₀ = 0.119·(M/10¹⁰)^{0.325}`, so the ceiling `B_max = 1/ε₀` runs from 17.8 at 10⁹ M☉ to 4.0 at 10¹¹ — and compare with MOND's ν where lensing actually reaches. Two decades below the SPARC floor is `g_bar ≈ 10⁻¹⁴ m s⁻²`, where ν ≈ 110: the ceiling falls short by **6× at 10⁹ M☉, ~13× at 10¹⁰, 28× at 10¹¹**. **A mass-dependent ceiling does not rescue the ceiling** — the failure grows with mass, because the relation lowers the ceiling exactly where the lensing sample is heaviest. *Screen coverage, stated because it bounds the claim: **abstract-level only in this lane; the full text was not read here.** The "2 decades" is the abstract's own wording. The site's companion table instead assumes reach to `g_bar = 10⁻¹⁵` and quotes deficits up to 87×; the abstract does not support that reach, so the conservative column is the one inscribed above.* The paper's own caveat is circumgalactic gas, and it is not close to sufficient here: a factor-2 correction on `g_bar` moves ν by √2. *[CHECKED 2026-11-11 — the full text has now been read (ar5iv HTML, in fragments, by a fetch subagent), and three things in this paragraph change. (a) **Reach.** Sec. 1 puts the outermost radius at g_bar ≈ 10⁻¹⁵ m s⁻², and its footnote 3 says that value "only takes into account the stellar and cold gas mass of the galaxy"; hot gas, dust and missing baryons "could increase this value to g_bar ≈ 10⁻¹⁴". The lensing bins run from 10⁻¹⁵ to 5×10⁻¹². So the site's 10⁻¹⁵ reach is the paper's own fiducial, and the coverage note above was wrong to call it unsupported. (b) **Census.** The paper's baryon allowance at the outer edge is therefore about ×10, not ×2, and √2 is the wrong scaling. Lensing fixes g_obs, so the boost a ceiling must supply is g_obs/g_bar, which falls as 1/f when the baryons rise by f; √f is how MOND's own ν moves. Recomputed with this paragraph's ceilings (17.8, 8.4, 4.0) and g_obs on MOND's curve: at a fiducial 10⁻¹⁵ the shortfalls are 19×, 41× and 87×, and ×10 baryons cuts them to 2.0×, 4.1× and 8.7×. At a fiducial 10⁻¹⁴ the printed 6×, 13×, 28× become pass, 1.3× and 2.8× under the same ×10. The mass-dependent ceiling still fails at 10¹⁰ and 10¹¹ solar masses, by a margin about ten times smaller than printed. Script: `simulations/publisher_20261111_lensing_ceiling_census.py`; no Brouwer data point was read, and the ×10 is the paper's statement for its outer edge only. (c) **Deflection.** Reading a lensing g_obs as a dynamical boost assumes light follows the same effective potential that moves stars, and this framework has written no light-deflection rule. The other rule does not save the ceiling. If light follows the visible matter only, lensing shows no boost at all: g_obs = g_bar, raised by at most the ×10 baryon allowance. Brouwer et al. report the opposite, a measured RAR that "agrees well with the MG predictions". On MOND's curve the boost at g_bar = 1.3, 2.3 and 4×10⁻¹³ m s⁻² is about 31, 23 and 18, all inside the range they treat as reliable (above 10⁻¹³, their Sec. 5.4 as quoted by the site), so ×10 baryons still leave 3.1×, 2.3× and 1.8× unexplained. So "dies independently" holds under either rule. The only reading that passes puts unseen matter in the lens, and that is a dark-matter model rather than this framework. Corrected 2026-11-12: the 11-11 wording made the verdict conditional on the rule. The site's 2026-11-11 block (`PREDICTIONS.md`) reaches the same verdict on the KiDS bins as the site transcribed them, at 7.2 to 10.6σ against the visible-only rule even at ×10; those σ values are relayed, not re-run here. Committing to a rule still gates on the project lead. The count of 6 does not depend on this paragraph.]* > > **Two numbers withdrawn from the 08-30/09-01 record — and the reason a reproduction did not catch them.** The statements *"59% of galaxies put `ρ_c` at the top grid edge"* and *"`ρ_c` tracks `M_bar` at +0.56"* are **void**: both read column 1 of `epsilon0_per_galaxy_fw.npy` as a per-galaxy `ρ_c`, and that column is per-galaxy **χ²**. Verified here at the *writer* rather than accepted from the correction — `epsilon0_free_the_ceiling_rescue.py:220` saves `np.vstack([e0s, best_per_gal, MOND_F, NN])`, and the column's own numbers settle it without reference to any script: median **49.4**, maximum **1.17×10⁴**, against a `ρ_c` grid spanning 10⁻⁶–10⁻² M☉ pc⁻³. Four orders of magnitude out, and dimensionally wrong. Neither number was ever inscribed in this section, so nothing here is retracted. **What is worth recording is that today's 77/77 exact reproduction reprinted the void line and scored it exact** — line 41 of both outputs reads `rho_c at top edge 59%`, because both runs read column 1. A reproduction gate certifies that the arithmetic ran the same way; it is **blind by construction to a mislabelled input**, because that defect lives upstream of the arithmetic. To catch it you must read the *writer* of an artifact, not the reader. Recorded against this lane specifically: the 09-01 amendment above reproduced its source "11 of 11 before inscription," and that gate would not have caught this either. > @@ -321,6 +339,8 @@ Applying G_eff = G/C to cosmology yields emergent dark energy: No cosmological constant Λ required. +*[CHECKED 2026-11-06: this split survives only when read as a fluid. Read as modified gravity, with G_eff = G/C(ρ̄_m(t)) the G of local dynamics, it drifts as the mean density falls: Ġ/G = 3H₀·d ln C/d ln ρ. At γ = ½ the identity d ln C/d ln x = 1 − C makes this exactly 3(1 − Ω_m)H₀ = 1.42×10⁻¹⁰ yr⁻¹. Lunar laser ranging gives (7.1 ± 7.6)×10⁻¹⁴ yr⁻¹ (Hofmann & Müller 2018, as quoted on the site, not re-read here). Every γ from 0.3 to 3 exceeds the 2σ upper edge by at least 527×. Only the fluid (Cardassian-class) reading survives: gravity unmodified, Ġ = 0. This is a different theory from the covariant completion whose LLR failure is in the covariant-branch paragraph above. The site explorer notes that Ġ/G ~ H₀ for an unscreened G_eff that drives acceleration is probably a known class result (locators not yet verified), so this is an application, not a new bound. Count unchanged. Source: `PREDICTIONS.md` 2026-11-05 (i); `synchronism-site/maintainer/scripts/de_modified_g_reading_vs_llr.py`, re-run by the Publisher 2026-11-06 with byte-identical output.]* + **Cosmic Coherence Form (Session #241 — "DERIVED" corrected below, 2026-08-11):** The cosmic coherence function has a natural form: @@ -343,7 +363,7 @@ As acceleration a → 0 (deep MOND regime): **Physical Interpretation:** Dark matter AND dark energy are both coherence effects—unified through C(a). At galactic scales, low coherence enhances gravity ("dark matter"). At cosmic scales, the coherence floor creates an effective vacuum energy ("dark energy"). -**Erratum on the source sessions (2026-08-10; independently verified 2026-08-11):** Sessions #100/#101 state `w_eff = −1 + (1/3)·d(ln ρ_DE)/d(ln a)`; the continuity equation gives `w = −1 − (1/3)·d(ln ρ_DE)/d(ln a)` (the published form returns w = −2 for matter), and the tabulated w(z) column follows neither formula — it drops the leading −1. Corrected values: w(0) = −1.24 at γ = 2 (not > 0 as Session #100 reports), monotone with w → −1 as a → ∞ and w → −2γ as z → ∞. Session #101's "category error" conclusion (that the galactic C form cannot serve at cosmic scales) is an artifact of this sign error: at γ = 1/2 the galactic form with C₀ = Ω_m is *identically* Ω_m(z) — the exact ΛCDM background, w ≡ −1. Structural consequence, now **class-level** [corrected 2026-08-12; the 2026-08-11 form of this sentence conditioned the result on the absence of a covariant derivation of the 00-component — that derivation was executed the same day, site lane, and *hardens* the conclusion]: the substituted model obeys **sign(w₀ + 1) = sign(wₐ)** — no phantom crossing, which is the quadrant DESI DR2 prefers in all four of its dataset combinations (0 of 16 γ values reach it) — and the two minimal covariant completions of Appendix D §D.3's equation close the escape the substitution left open. **Completion A** (the equation as written, no new fields): the Bianchi identity forces ρ/C ∝ a⁻³, so the background is *exactly Einstein–de Sitter* for every γ and calibration — the dark-energy sector vanishes identically, and the vacuum floor ρ/C ≥ ρ_crit/γ leaves the equation no FRW solution beyond a ≈ 1.037 under Session #100's own calibration. **Completion B** (C as a Brans-Dicke scalar pinned to C(ρ̄(a))) **[PRIOR ART 2026-08-26 — this branch is published: the covariant formulation of the weak-field law is Sanna, Matsakos & Diaferio 2023 (A&A **674**, A209), and Gervani, Diaferio, Pace & Sanna 2026 (arXiv:2601.22937) integrate its linear perturbations and find structure formation delayed to z < 1, contradicting observed high-z galaxies. Whether their theory and this construction coincide is UNCHECKED — see the 08-26 amendment above.]**: the literal sign lock *dies* — the w = −1 attractor is destroyed, the γ = 1/2 Λ-degeneracy is broken (no member of the completed family is ΛCDM), and w₀ < −1 with wₐ > 0 members exist — but the DESI quadrant (w₀ > −1, wₐ < 0) stays empty: 0 of 192 γ values. **[CORRECTED 2026-08-19/20 — the earlier form of this clause read “0 of 192 γ values *at every Brans-Dicke ω tested*,” which was an **under**-refutation in two independent ways, and the corrected statement is strictly stronger.]** *(i) The tested grid was excluded before it was scanned.* Brans-Dicke gives `γ_PPN − 1 = −1/(2+ω)`; Cassini (Bertotti, Iess & Tortora 2003) measures `(2.1 ± 2.3)×10⁻⁵`, whose 2σ lower edge `−2.5×10⁻⁵` forces **ω ≥ 4.0×10⁴** for an unscreened massless scalar — so the published grid `ω ∈ {0, 1, 5, 50}` topped out **800× inside the excluded region**, and no solar-system bound appeared anywhere in this sector although the same spacecraft is cited for TEST-25 **[ID NOTE 2026-08-20 — `TEST-25` here resolves to the **site** surface’s *Cassini squeeze*, NOT to this repository’s `Research/EXPERIMENTAL_TEST_CATALOG.md` §TEST-25, which is *a₀ Redshift Evolution*. The two public surfaces collided on this ID when the site cleared an *intra-site* collision by renumbering `TEST-11 → TEST-25` on 2026-08-10. Following this citation into the catalog shipped alongside this document lands on the wrong test; no renumbering is performed here, as which side moves is a cross-repo decision.]**. *(ii) ω was never a free dimension of the scan.* The pinning `C_eff(x₀) = Ω_m` is maintained by sliding `x₀` (0.95 → 3.6×10⁴), not by ω doing physical work: ω enters the background only through `B = 1 − 3ε − 1.5ωε²`, and the pinned point is that factor's positive root `ε_crit(ω) = (−3 + √(9+6ω))/(3ω)`. **Verified here symbolically** (`simulations/publisher_20260820_completion_b_omega_absorption.py`, 11/11): at that root `1.5·ω·ε_crit² = 1 − 3ε_crit` *exactly*, so **the closure pins the product `ω·ε_crit²`, which → 2/3, rather than ω itself** — `ε_crit ~ √(2/(3ω))` and the construction has one physical point, not a family. “192 × 4 ω-values” is **192 × 1**; the scan advertised a dimension the construction does not have. (This closure identity is the reason behind the source's numerical observation that trajectories at ω = 4×10⁴ and 10⁶ agree to <2% — the departures are 1.22% and 0.24%.) *(iii) At the allowed ω the no-go hardens rather than loosening:* `w₀` moves −1.58 (ω = 0) → **−3.18** (ω = 4×10⁴) at γ = 0.489, deeper into the wrong quadrant, with ρ_DE at 2% of its present value by z = 2. *(iv) The screening escape is self-inconsistent rather than a rescue:* giving the scalar a mass via `V(C)` to evade PPN would contribute to `B(x)`, which is the **massless** BD energy density and omits V — the published scan cannot be simultaneously PPN-safe and self-consistent. **The one remaining unexecuted covariant branch** — unexecuted *here*; the literature's covariant RG carries a *linear* potential `V(φ) = −Ξφ`, which is not a mass term, so the branch below is not obviously the published one **[scoped 2026-08-26]** **[CORRECTED 2026-10-23: in the published theory's trace equation, (∇² − ∂²/∂t²)φ − 2Ξφ = 8πG T, the linear potential does act as a mass, but of Hubble range (2.2 Gpc, m·AU ≈ 2×10⁻¹⁵), so it cannot evade Cassini. The scoping conclusion stands: the branch below is not the published one, and the published one fails Cassini and LLR (08-26 covariant-branch paragraph above).]** — is therefore a density-pinned **massive** scalar (add `V(C)` to `B(x)` and integrate): it evades Cassini by construction, and a potential is exactly the ingredient that could supply the interior maximum of ρ_DE(x) the DESI quadrant requires. Routed, not driven. Refutation count unchanged (6); the completion-B exclusion stands and is stronger. *(Source: `synchronism-site/explorer/findings/completion-b-at-cassini-allowed-omega-the-no-go-hardens-and-omega-was-never-free.md`, 2026-08-19, routed via `Research/proposals/completion_b_cassini_omega_absorbed_by_closure_20260819.md`; the w₀ = −3.18 trajectory value is source-executed and not re-integrated here.)* Continuing the original clause: , and forcing w₀ to DESI's value forces wₐ = +0.23…+0.60, the wrong sign in all four combinations. The class criterion is one line: for any dark energy slaved to matter density, ρ_DE = ρ_m·F(x), continuity gives w_DE = d ln F/d ln x, so DESI's phantom→quintessence crossing requires an interior *maximum* of ρ_DE(x) — and no completion of C = tanh(γ ln(1+x)) produces one (substituted form: monotone; completion A: ρ_DE ≡ 0; completion B: interior *minimum*, crossing in the anti-DESI direction). Residual conditionality: completion B's quasi-static pinning presumes an enforcing sector of negligible stress. The γ = 1/2 branch of the *substituted* model is exactly ΛCDM (inheriting ΛCDM's 2.8–4.2σ DESI tension, no more) — but that degeneracy is a property of the substitution, not of the covariant completions. **[STRENGTHENED 2026-08-19 — the qualifier "background" is removed because it was load-bearing in the wrong direction, and the identities below were re-derived here in SymPy, exactly, for all x and γ]:** the γ = 1/2 branch is ΛCDM **non-perturbatively**, not merely degenerate with it in a background fit. `ρ_DE/ρ_crit = 2x/((1+x)^{2γ} − 1)` equals **exactly 2 at γ = 1/2, with dρ_DE/dρ_m ≡ 0** — ρ_DE is constant in *space* as well as in time, the same inside a void, a cluster and a disk — and the induced contrast `δ_DE/δ_m = 1 + w_DE = ε·(ln(1+x)/x − 1) + O(ε²)`, `ε ≡ 2γ − 1`, vanishes identically at every scale and every order. Same background, same linear perturbations, same nonlinear field configuration. This **retires the standing "background-only, no perturbation sector" caveat**, which had twice been read as an unexplored channel that might convert the dp-gated DR3 registration into a near-term test: the *locality* fork — evaluating C at the local rather than the mean density, which the framework's one-equation postulate actually requires — was executed 2026-08-18 and adds **no independent channel**, because both the expansion-history and clustering terms are linear in the same single ε (no order-ε⁰ term exists). What locality buys is exactly a factor 2 in the observable: required precision σ(fσ₈) relaxes 0.074% → 0.15%. It buys **nothing** in σ_γ, which is `(1/2 − γ̂)/3` in both horns by construction — the same γ-space separation this paragraph already prices at σ_γ ≈ 0.004 (the source finding's "7% improvement" is a central-value artifact, 0.004 at γ̂ = 0.487 vs 0.0037 at γ̂ = 0.489; corrected here). Against DESI DR2's ~1–3% per bin the channel is ~10× underpowered, and the parameter it would measure sits at **ε = −0.022 ± 0.220 = 0.10σ from exact ΛCDM**. Bucket 0 unchanged (0); refutation count HELD at 6 — this closes a fork and refutes nothing. Sources: `synchronism-site/explorer/findings/de-locality-fork-executed-perturbation-channel-buys-exactly-a-factor-of-two.md` (2026-08-18), routed via `Research/proposals/de_locality_fork_perturbation_channel_factor_two_20260818.md` and triaged in `explorations/2026-08-18-de-locality-fork-perturbations-buy-x2-and-session107-is-a-different-refuted-mechanism.md`. A direct likelihood fit of the w(z; γ) family to DESI DR2 + Planck-compressed CMB + DES-SN5YR (site explorer, 2026-08-12; the Δχ² numbers are site-executed and not re-run in-repo, though the pipeline self-validates against DESI's published ΛCDM and w₀wₐ results) confirms this structure at data level: the substituted branch sits at its Λ-corner — best γ = 0.487 ± 0.02, Δχ² = −0.3 vs ΛCDM, +11 behind w₀wₐCDM — and both covariant completions fail the fit outright (A: χ² ≈ 9,900, the pre-1998 background; B: Δχ² ≥ +79 at every ω on that same PPN-excluded grid — and per the ω correction above, one physical point rather than four), so the sector survives only in its non-covariant form and only by being ΛCDM. An earlier site-side "3.4–6.3σ forced-wₐ" exclusion of the substituted branch did not survive this execution — the likelihood never forces the family off the Λ-corner; the verdict is non-novelty, not exclusion. The apparent cross-sector agreement γ_cosmo = 0.487 vs γ_galaxy(SPARC) = 0.489 has **no power to fail** (γ = 1/2 is exactly Λ; 0.489 is exactly MOND simple-μ; the two standard models sit 0.011 apart in γ-space by construction) and is therefore not a concordance. **[SCOPED 2026-08-15]** That deflation was *structural* when written — the SPARC-side uncertainty had never been derived, so the pricing was incomplete. It has now been derived, and it makes the closure **permanent** and the sign of the comparison **convention-dependent**. σ(γ) = **0.11** (stat) — the galaxy-limited value, not the naive point-level 0.029, because rotation-curve points within a galaxy are coherently correlated so the effective N is the number of *galaxies* (jackknife 0.103 / bootstrap(400) 0.113 / √(N/N_gal) 0.121; corroborated in-repo against `simulations/sparc_real_data/MassModels_Lelli2016c.mrt` — 175 galaxies, 3391 points, 0.029 × √19.4 = 0.128). Against the continuous free minimum γ̂ = 0.4892 the comparison is |0.487 − 0.4892| = 0.002, i.e. **0.02σ**; separating 0.489 from 1/2 would need σ_γ ≈ 0.004, which requires ~130,000 SPARC-quality galaxies *and* a global ϒ known to ±0.0007. **The ϒ term does not average down, so γ = 1/2-vs-fit discrimination on rotation curves is closed a priori** — no SPARC-class dataset rescues it. Sharper still, the agreement is an artifact of the ϒ_disk = 0.5 convention: refitting at ϒ_disk ∈ {0.4, 0.5, 0.55, 0.6} moves γ̂ across **[0.27, 0.96]** at flat rms (0.1433–0.1458 dex), and at the *better-fitting* ϒ = 0.55 (Δχ²_naive = −14.7) γ̂ = 0.68 — the "agreement" becomes a ~1.7σ **tension** with lower χ². Single galaxies (UGC11914, UGC03580, NGC5985) each move γ̂ by ±0.03–0.04, ~3× the entire |0.489 − 1/2| offset: the three decimals of "0.489" are noise digits. **Consequence for the parenthetical above**, stated because it qualifies this section's own supporting clause: "0.489 is exactly MOND simple-μ" is an ϒ = 0.5 *slice* statement — q = 2γ spans [0.5, 1.9] across the band — so every "γ ≈ 0.5 from galaxies" statement carries factor-2 slack. γ, a₀ and ϒ are one flat degeneracy (the "profiled a₀ = 5.33×10⁻¹¹ vs derived cH₀/2π = 1.04×10⁻¹⁰, factor 1.96" tension dissolves at ϒ = 0.6, where the profiled a₀ → 1.043×10⁻¹⁰ ≡ the derived value at indistinguishable rms), so **the galaxy sector's shape parameter is unidentified at factor 2** — a sharper statement than "it reduces to MOND," since an unidentified parameter cannot carry a novel prediction. This does not disturb the door-#1 verdict, whose load-bearing kills are the ≤25% admixture bound and the g_bar→ρ locality no-go rather than the exact γ value. **Provenance and what was not re-run**: the stat term is corroborated in-repo as above; the ϒ sweep, the ϒ = 0.55 flip and the a₀ dissolution are **site-executed and not reproduced here** (`synchronism-site/explorer/findings/scripts/sparc_gamma_interval_frozen_likelihood.py` exceeded the in-session compute budget and was terminated at exit 143 before output). Open thread, carried as the source states it: a hierarchical refit with per-galaxy ϒ priors (Li et al. 2018 RAR methodology) is the defensible next rung before any external citation of the interval. Refutation count **HELD at 6**, Bucket 0 = 0 — this prices a comparison and refutes nothing (source: `Research/proposals/sparc_gamma_interval_upsilon_degeneracy_20260814.md`, routed from the site explorer back-annotation `c2b78336`, 2026-08-14). A prospective registration of the no-go (DESI DR3 timeline, ~2027–2028) is filed and gated on dp; its terms as currently proposed [corrected 2026-08-14; the 2026-08-12 form of this sentence gave "the discriminant is the sign of wₐ" — a statistic the direct fit showed is ΛCDM-degenerate on the only surviving branch]: class-level (the interior-maximum criterion above), statistic moved to Δχ²(substituted best fit vs w₀wₐCDM) on DR3 likelihoods, presented as a consistency check rather than a discriminator, with a pre-committed projection-robustness check against the CPL-artifact literature. Sign-error statements verified in `simulations/publisher_20260811_w_eff_erratum_check.py`; completion algebra, the EdS result, a_end, and the completion-B table independently reproduced in `simulations/publisher_20260812_covariant_00_checks.py` (source: `synchronism-site/explorer/findings/covariant-00-component-sign-lock-dies-desi-nogo-hardens.md`, 2026-08-11, routed via `Research/proposals/covariant_00_component_sign_lock_dies_desi_nogo_hardens_20260811.md`; direct-fit source: `synchronism-site/explorer/findings/gamma-family-direct-fit-desi-dr2-substituted-is-lcdm-covariant-excluded.md`, 2026-08-12, routed via `Research/proposals/gamma_family_direct_fit_desi_dr2_20260812.md`). +**Erratum on the source sessions (2026-08-10; independently verified 2026-08-11):** Sessions #100/#101 state `w_eff = −1 + (1/3)·d(ln ρ_DE)/d(ln a)`; the continuity equation gives `w = −1 − (1/3)·d(ln ρ_DE)/d(ln a)` (the published form returns w = −2 for matter), and the tabulated w(z) column follows neither formula — it drops the leading −1. Corrected values: w(0) = −1.24 at γ = 2 (not > 0 as Session #100 reports), monotone with w → −1 as a → ∞ and w → −2γ as z → ∞. Session #101's "category error" conclusion (that the galactic C form cannot serve at cosmic scales) is an artifact of this sign error: at γ = 1/2 the galactic form with C₀ = Ω_m is *identically* Ω_m(z) — the exact ΛCDM background, w ≡ −1. Structural consequence, now **class-level** [corrected 2026-08-12; the 2026-08-11 form of this sentence conditioned the result on the absence of a covariant derivation of the 00-component — that derivation was executed the same day, site lane, and *hardens* the conclusion]: the substituted model obeys **sign(w₀ + 1) = sign(wₐ)** — no phantom crossing, which is the quadrant DESI DR2 prefers in all four of its dataset combinations (0 of 16 γ values reach it) *[CHECKED 2026-10-30 — that quadrant (w₀ > −1, wₐ < 0) is not all crossing. In CPL, w(a) = w₀ + wₐ(1 − a), so the early value is w₀ + wₐ, and a crossing of w = −1 needs w₀ + wₐ < −1; the rest of the quadrant is thawing, which never crosses. The sign lock excludes the whole quadrant, so it forbids thawing as well as crossing and allows only evolution toward −1. The exclusion is wider than "no phantom crossing" says; the verdict is unchanged. Source: PREDICTIONS.md 2026-09-23 block; the CPL algebra is checked here.]* — and the two minimal covariant completions of Appendix D §D.3's equation close the escape the substitution left open. **Completion A** (the equation as written, no new fields): the Bianchi identity forces ρ/C ∝ a⁻³, so the background is *exactly Einstein–de Sitter* for every γ and calibration — the dark-energy sector vanishes identically, and the vacuum floor ρ/C ≥ ρ_crit/γ leaves the equation no FRW solution beyond a ≈ 1.037 under Session #100's own calibration. **Completion B** (C as a Brans-Dicke scalar pinned to C(ρ̄(a))) **[PRIOR ART 2026-08-26 — this branch is published: the covariant formulation of the weak-field law is Sanna, Matsakos & Diaferio 2023 (A&A **674**, A209), and Gervani, Diaferio, Pace & Sanna 2026 (arXiv:2601.22937) integrate its linear perturbations and find structure formation delayed to z < 1, contradicting observed high-z galaxies. Whether their theory and this construction coincide is UNCHECKED — see the 08-26 amendment above.]**: the literal sign lock *dies* — the w = −1 attractor is destroyed, the γ = 1/2 Λ-degeneracy is broken (no member of the completed family is ΛCDM), and w₀ < −1 with wₐ > 0 members exist — but the DESI quadrant (w₀ > −1, wₐ < 0) stays empty: 0 of 192 γ values. **[CORRECTED 2026-08-19/20 — the earlier form of this clause read “0 of 192 γ values *at every Brans-Dicke ω tested*,” which was an **under**-refutation in two independent ways, and the corrected statement is strictly stronger.]** *(i) The tested grid was excluded before it was scanned.* Brans-Dicke gives `γ_PPN − 1 = −1/(2+ω)`; Cassini (Bertotti, Iess & Tortora 2003) measures `(2.1 ± 2.3)×10⁻⁵`, whose 2σ lower edge `−2.5×10⁻⁵` forces **ω ≥ 4.0×10⁴** for an unscreened massless scalar — so the published grid `ω ∈ {0, 1, 5, 50}` topped out **800× inside the excluded region**, and no solar-system bound appeared anywhere in this sector although the same spacecraft is cited for TEST-25 **[ID NOTE 2026-08-20 — `TEST-25` here resolves to the **site** surface’s *Cassini squeeze*, NOT to this repository’s `Research/EXPERIMENTAL_TEST_CATALOG.md` §TEST-25, which is *a₀ Redshift Evolution*. The two public surfaces collided on this ID when the site cleared an *intra-site* collision by renumbering `TEST-11 → TEST-25` on 2026-08-10. Following this citation into the catalog shipped alongside this document lands on the wrong test; no renumbering is performed here, as which side moves is a cross-repo decision.]**. *(ii) ω was never a free dimension of the scan.* The pinning `C_eff(x₀) = Ω_m` is maintained by sliding `x₀` (0.95 → 3.6×10⁴), not by ω doing physical work: ω enters the background only through `B = 1 − 3ε − 1.5ωε²`, and the pinned point is that factor's positive root `ε_crit(ω) = (−3 + √(9+6ω))/(3ω)`. **Verified here symbolically** (`simulations/publisher_20260820_completion_b_omega_absorption.py`, 11/11): at that root `1.5·ω·ε_crit² = 1 − 3ε_crit` *exactly*, so **the closure pins the product `ω·ε_crit²`, which → 2/3, rather than ω itself** — `ε_crit ~ √(2/(3ω))` and the construction has one physical point, not a family. “192 × 4 ω-values” is **192 × 1**; the scan advertised a dimension the construction does not have. (This closure identity is the reason behind the source's numerical observation that trajectories at ω = 4×10⁴ and 10⁶ agree to <2% — the departures are 1.22% and 0.24%.) *(iii) At the allowed ω the no-go hardens rather than loosening:* `w₀` moves −1.58 (ω = 0) → **−3.18** (ω = 4×10⁴) at γ = 0.489, deeper into the wrong quadrant, with ρ_DE at 2% of its present value by z = 2. *(iv) The screening escape is self-inconsistent rather than a rescue:* giving the scalar a mass via `V(C)` to evade PPN would contribute to `B(x)`, which is the **massless** BD energy density and omits V — the published scan cannot be simultaneously PPN-safe and self-consistent. **The one remaining unexecuted covariant branch** — unexecuted *here*; the literature's covariant RG carries a *linear* potential `V(φ) = −Ξφ`, which is not a mass term, so the branch below is not obviously the published one **[scoped 2026-08-26]** **[CORRECTED 2026-10-23: in the published theory's trace equation, (∇² − ∂²/∂t²)φ − 2Ξφ = 8πG T, the linear potential does act as a mass, but of Hubble range (2.2 Gpc, m·AU ≈ 2×10⁻¹⁵), so it cannot evade Cassini. The scoping conclusion stands: the branch below is not the published one, and the published one fails Cassini and LLR (08-26 covariant-branch paragraph above).]** — is therefore a density-pinned **massive** scalar (add `V(C)` to `B(x)` and integrate): it evades Cassini by construction, and a potential is exactly the ingredient that could supply the interior maximum of ρ_DE(x) the DESI quadrant requires. Routed, not driven. Refutation count unchanged (6); the completion-B exclusion stands and is stronger. *(Source: `synchronism-site/explorer/findings/completion-b-at-cassini-allowed-omega-the-no-go-hardens-and-omega-was-never-free.md`, 2026-08-19, routed via `Research/proposals/completion_b_cassini_omega_absorbed_by_closure_20260819.md`; the w₀ = −3.18 trajectory value is source-executed and not re-integrated here.)* Continuing the original clause: , and forcing w₀ to DESI's value forces wₐ = +0.23…+0.60, the wrong sign in all four combinations. The class criterion is one line: for any dark energy slaved to matter density, ρ_DE = ρ_m·F(x), continuity gives w_DE = d ln F/d ln x, so DESI's phantom→quintessence crossing requires an interior *maximum* of ρ_DE(x) — and no completion of C = tanh(γ ln(1+x)) produces one (substituted form: monotone; completion A: ρ_DE ≡ 0; completion B: interior *minimum*, crossing in the anti-DESI direction). Residual conditionality: completion B's quasi-static pinning presumes an enforcing sector of negligible stress. The γ = 1/2 branch of the *substituted* model is exactly ΛCDM (inheriting ΛCDM's 2.8–4.2σ DESI tension, no more) — but that degeneracy is a property of the substitution, not of the covariant completions. **[STRENGTHENED 2026-08-19 — the qualifier "background" is removed because it was load-bearing in the wrong direction, and the identities below were re-derived here in SymPy, exactly, for all x and γ]:** the γ = 1/2 branch is ΛCDM **non-perturbatively**, not merely degenerate with it in a background fit. `ρ_DE/ρ_crit = 2x/((1+x)^{2γ} − 1)` equals **exactly 2 at γ = 1/2, with dρ_DE/dρ_m ≡ 0** — ρ_DE is constant in *space* as well as in time, the same inside a void, a cluster and a disk — and the induced contrast `δ_DE/δ_m = 1 + w_DE = ε·(ln(1+x)/x − 1) + O(ε²)`, `ε ≡ 2γ − 1`, vanishes identically at every scale and every order. Same background, same linear perturbations, same nonlinear field configuration. This **retires the standing "background-only, no perturbation sector" caveat**, which had twice been read as an unexplored channel that might convert the dp-gated DR3 registration into a near-term test: the *locality* fork — evaluating C at the local rather than the mean density, which the framework's one-equation postulate actually requires — was executed 2026-08-18 and adds **no independent channel**, because both the expansion-history and clustering terms are linear in the same single ε (no order-ε⁰ term exists). What locality buys is exactly a factor 2 in the observable: required precision σ(fσ₈) relaxes 0.074% → 0.15%. It buys **nothing** in σ_γ, which is `(1/2 − γ̂)/3` in both horns by construction — the same γ-space separation this paragraph already prices at σ_γ ≈ 0.004 (the source finding's "7% improvement" is a central-value artifact, 0.004 at γ̂ = 0.487 vs 0.0037 at γ̂ = 0.489; corrected here). Against DESI DR2's ~1–3% per bin the channel is ~10× underpowered, and the parameter it would measure sits at **ε = −0.022 ± 0.220 = 0.10σ from exact ΛCDM**. Bucket 0 unchanged (0); refutation count HELD at 6 — this closes a fork and refutes nothing. Sources: `synchronism-site/explorer/findings/de-locality-fork-executed-perturbation-channel-buys-exactly-a-factor-of-two.md` (2026-08-18), routed via `Research/proposals/de_locality_fork_perturbation_channel_factor_two_20260818.md` and triaged in `explorations/2026-08-18-de-locality-fork-perturbations-buy-x2-and-session107-is-a-different-refuted-mechanism.md`. *[CHECKED 2026-11-15: the three claims above (no order-ε⁰ term, no independent channel, a channel ~10× underpowered) hold only without a pressure term. Read at local density with the dark energy comoving with the matter (Gondolo & Freese's fluid Cardassian, the one completion that keeps ρ_DE a function of local ρ_m without new fields), the sector is a barotropic fluid with sound speed c_s² ≈ 0.22 ε today. The 08-18 growth equation has no Jeans term c_s²(k/aH)², and on the scales where P(k) is measured that term carries all of the effect. With it, existing P(k) already excludes the fluid local horn unless |ε| < 9.1×10⁻⁶ (ε < 0) or 9.9×10⁻⁶ (ε > 0), i.e. γ = ½ ± 5×10⁻⁶, about 5,000× narrower than the DESI-fit 1σ. At γ = 0.487 the matter power ratio to ΛCDM is 9.7 at k = 0.01 h/Mpc. On this reading the channel is not underpowered: data already fix γ = ½. The γ = 0.487 ± 0.02 fit below therefore survives only on the mean-density (background-only) reading, which gives up the local evaluation this paragraph says the postulate requires. The mechanism is published prior art (generalized Chaplygin gas, Sandvik et al. 2004; fluid Cardassian, Amarzguioui, Elgarøy & Multamäki 2005). Limits: a single adiabatic fluid; letting only the CDM carry the pressure loosens the galaxy-P(k) band about 10×, to |ε| ~ 10⁻⁴, and lensing, which sees the CDM, not at all. No registered prediction is refuted; count HELD at 6. Source: `PREDICTIONS.md` 2026-09-15 block and 2026-11-14 (iv); `synchronism-site/explorer/findings/scripts/de_horn_l_as_a_fluid_jeans_term.py` (pre-registered at site `7d8efbf`) and `de_horn_l_fluid_controls.py` (post-hoc controls), both re-run by the Publisher 2026-11-15 with byte-identical output.]* A direct likelihood fit of the w(z; γ) family to DESI DR2 + Planck-compressed CMB + DES-SN5YR (site explorer, 2026-08-12; the Δχ² numbers are site-executed and not re-run in-repo, though the pipeline self-validates against DESI's published ΛCDM and w₀wₐ results) confirms this structure at data level: the substituted branch sits at its Λ-corner — best γ = 0.487 ± 0.02, Δχ² = −0.3 vs ΛCDM, +11 behind w₀wₐCDM — and both covariant completions fail the fit outright (A: χ² ≈ 9,900, the pre-1998 background; B: Δχ² ≥ +79 at every ω on that same PPN-excluded grid — and per the ω correction above, one physical point rather than four), so the sector survives only in its non-covariant form and only by being ΛCDM. An earlier site-side "3.4–6.3σ forced-wₐ" exclusion of the substituted branch did not survive this execution — the likelihood never forces the family off the Λ-corner; the verdict is non-novelty, not exclusion. The apparent cross-sector agreement γ_cosmo = 0.487 vs γ_galaxy(SPARC) = 0.489 has **no power to fail** (γ = 1/2 is exactly Λ; 0.489 is exactly MOND simple-μ; the two standard models sit 0.011 apart in γ-space by construction) and is therefore not a concordance. **[SCOPED 2026-08-15]** That deflation was *structural* when written — the SPARC-side uncertainty had never been derived, so the pricing was incomplete. It has now been derived, and it makes the closure **permanent** and the sign of the comparison **convention-dependent**. σ(γ) = **0.11** (stat) — the galaxy-limited value, not the naive point-level 0.029, because rotation-curve points within a galaxy are coherently correlated so the effective N is the number of *galaxies* (jackknife 0.103 / bootstrap(400) 0.113 / √(N/N_gal) 0.121; corroborated in-repo against `simulations/sparc_real_data/MassModels_Lelli2016c.mrt` — 175 galaxies, 3391 points, 0.029 × √19.4 = 0.128). Against the continuous free minimum γ̂ = 0.4892 the comparison is |0.487 − 0.4892| = 0.002, i.e. **0.02σ**; separating 0.489 from 1/2 would need σ_γ ≈ 0.004, which requires ~130,000 SPARC-quality galaxies *and* a global ϒ known to ±0.0007. **The ϒ term does not average down, so γ = 1/2-vs-fit discrimination on rotation curves is closed a priori** — no SPARC-class dataset rescues it. Sharper still, the agreement is an artifact of the ϒ_disk = 0.5 convention: refitting at ϒ_disk ∈ {0.4, 0.5, 0.55, 0.6} moves γ̂ across **[0.27, 0.96]** at flat rms (0.1433–0.1458 dex), and at the *better-fitting* ϒ = 0.55 (Δχ²_naive = −14.7) γ̂ = 0.68 — the "agreement" becomes a ~1.7σ **tension** with lower χ². Single galaxies (UGC11914, UGC03580, NGC5985) each move γ̂ by ±0.03–0.04, ~3× the entire |0.489 − 1/2| offset: the three decimals of "0.489" are noise digits. **Consequence for the parenthetical above**, stated because it qualifies this section's own supporting clause: "0.489 is exactly MOND simple-μ" is an ϒ = 0.5 *slice* statement — q = 2γ spans [0.5, 1.9] across the band — so every "γ ≈ 0.5 from galaxies" statement carries factor-2 slack. γ, a₀ and ϒ are one flat degeneracy (the "profiled a₀ = 5.33×10⁻¹¹ vs derived cH₀/2π = 1.04×10⁻¹⁰, factor 1.96" tension dissolves at ϒ = 0.6, where the profiled a₀ → 1.043×10⁻¹⁰ ≡ the derived value at indistinguishable rms), so **the galaxy sector's shape parameter is unidentified at factor 2** — a sharper statement than "it reduces to MOND," since an unidentified parameter cannot carry a novel prediction. *[CHECKED 2026-10-30 — the "factor 1.96" tension compared the wrong a₀. This section's own deep-limit identity (the paragraph showing the fit depends on γ and a₀ "only through the ratio γ/a₀") makes the fit's Milgrom-equivalent scale the fitted a₀ divided by γ: 5.33×10⁻¹¹ / 0.489 = 1/(9.17×10⁹) = 1.09×10⁻¹⁰, 4.6% above cH₀/2π = 1.04×10⁻¹⁰. There was no factor-1.96 tension to dissolve: 1.96 is the 1/γ conversion at γ ≈ ½, and the ϒ = 0.6 "dissolution" is that conversion going to ≈ 1 as the fitted γ approaches 1 (γ ≈ 0.96 there, per the source block). The scale is identified; what stays unidentified is the shape (the ϒ ↔ γ degeneracy), so the conclusion just stated stands. [UPDATE 2026-11-09: "identified" holds only with γ free. a₀′/γ stays at 1.04–1.09×10⁻¹⁰ across Υ_disk = 0.4–0.7 because the fitted γ runs 0.28 → 2.14 along the valley (site `5a95614`, predictions before the script, same frozen pipeline). Pinned at γ = ½, the reading this section keeps, the Milgrom-equivalent a₀ = a₀′/γ moves 1.44 → 1.10 → 0.87×10⁻¹⁰ m s⁻² at Υ_disk = 0.4 / 0.5 / 0.6, and closing a₀ = cH₀/2π needs H₀ ≈ 93 / 71 / 57 km s⁻¹ Mpc⁻¹. So the 4.6% agreement with cH₀/2π holds at one mass-to-light ratio, within the ±0.24×10⁻¹⁰ m s⁻² systematic that McGaugh, Lelli & Schombert 2016 quote on g†. Read "the scale is identified" as identified jointly with γ, the same shape as the 10-23 note on γ and ρ_crit. Source: `PREDICTIONS.md` back-annotation `6fe7b696` (iii), which names this as the ledger's own 10-04 error; `gamma2_global_upsilon_per_law.py` P5/P6, re-run by the Publisher 2026-11-10 with byte-identical output.] Source: PREDICTIONS.md 2026-10-04 block (i), which reports a₀/γ = 1.074–1.092×10⁻¹⁰ across ϒ = 0.4–0.6 (site script, not re-run here); the arithmetic in this note uses only numbers printed in this section.]* This does not disturb the door-#1 verdict, whose load-bearing kills are the ≤25% admixture bound and the g_bar→ρ locality no-go rather than the exact γ value. **Provenance and what was not re-run**: the stat term is corroborated in-repo as above; the ϒ sweep, the ϒ = 0.55 flip and the a₀ dissolution are **site-executed and not reproduced here** (`synchronism-site/explorer/findings/scripts/sparc_gamma_interval_frozen_likelihood.py` exceeded the in-session compute budget and was terminated at exit 143 before output). Open thread, carried as the source states it: a hierarchical refit with per-galaxy ϒ priors (Li et al. 2018 RAR methodology) is the defensible next rung before any external citation of the interval. Refutation count **HELD at 6**, Bucket 0 = 0 — this prices a comparison and refutes nothing (source: `Research/proposals/sparc_gamma_interval_upsilon_degeneracy_20260814.md`, routed from the site explorer back-annotation `c2b78336`, 2026-08-14). A prospective registration of the no-go (DESI DR3 timeline, ~2027–2028) is filed and gated on dp; its terms as currently proposed [corrected 2026-08-14; the 2026-08-12 form of this sentence gave "the discriminant is the sign of wₐ" — a statistic the direct fit showed is ΛCDM-degenerate on the only surviving branch]: class-level (the interior-maximum criterion above), statistic moved to Δχ²(substituted best fit vs w₀wₐCDM) on DR3 likelihoods, presented as a consistency check rather than a discriminator, with a pre-committed projection-robustness check against the CPL-artifact literature. Sign-error statements verified in `simulations/publisher_20260811_w_eff_erratum_check.py`; completion algebra, the EdS result, a_end, and the completion-B table independently reproduced in `simulations/publisher_20260812_covariant_00_checks.py` (source: `synchronism-site/explorer/findings/covariant-00-component-sign-lock-dies-desi-nogo-hardens.md`, 2026-08-11, routed via `Research/proposals/covariant_00_component_sign_lock_dies_desi_nogo_hardens_20260811.md`; direct-fit source: `synchronism-site/explorer/findings/gamma-family-direct-fit-desi-dr2-substituted-is-lcdm-covariant-excluded.md`, 2026-08-12, routed via `Research/proposals/gamma_family_direct_fit_desi_dr2_20260812.md`). --- diff --git a/whitepaper/sections/05-quantum-macro/16-superconductivity/superconductivity.md b/whitepaper/sections/05-quantum-macro/16-superconductivity/superconductivity.md index 746bb95c2..fa572d699 100644 --- a/whitepaper/sections/05-quantum-macro/16-superconductivity/superconductivity.md +++ b/whitepaper/sections/05-quantum-macro/16-superconductivity/superconductivity.md @@ -124,6 +124,8 @@ From this identification, the universal BCS ratio can be derived: **Observed**: 3.52 for conventional superconductors (**<1% error**) +*[CHECKED 2026-11-17: the 3.52 is BCS's own weak-coupling prediction, not a measurement, and 2√π is not derived from it. Solving the gap equation above at couplings λ = 0.15–0.30 gives 2Δ₀/(k_B T_c) = 3.528–3.532, against the closed form 2π·exp(−0.5772) = 3.5278 (0.5772 is Euler's constant). 2√π = 3.5449 sits 0.49% above that, so "<1% error" measures the distance between two numbers, one of which BCS already fixes. The source's "2D phase space" route (Chemistry Session #1 §3.3) builds √π from Gaussian integrals and never uses the gap equation. The γ-dependent ratio in the same session's script, 2π/tanh(γ·ln 2), gives 7.12 at γ = 2, which the script's own printout calls "too high". Measured ratios for elemental superconductors fall on both sides of 3.53 (roughly 3.2 to 4.6, with strong-coupling Pb and Hg highest; standard tables, not re-read in this pass). Script: `simulations/publisher_20261117_bcs_ratio_and_gamma_check.py`, rc 0. Words added, none removed.]* + **Physical Meaning:** - Cooper pairs are phase-locked electron pairs (Δφ = 0) - The energy gap Δ is the coherence protection scale @@ -138,6 +140,8 @@ From this identification, the universal BCS ratio can be derived: | Galaxy rotation | tanh(γ log(ρ/ρ_c + 1)) | 2 | | Chemistry bonds | tanh(1.5 Δχ) | ~1.5 | +*[CHECKED 2026-11-17: the superconductor "~2" is BCS's γ, not a measurement of the template. (The text above reads the same tanh(E/2kT) as γ = 1.) The archive fits behind it (Chemistry Session #1 §4.4: Al 2.31, Nb 2.65, Pb 2.45, knee pinned at β = 2) use six hand-typed "representative" points per metal. Fitting the template to the textbook BCS curve Δ/Δ₀ = tanh(1.74·√(T_c/T − 1)) on the same temperature grid, with no data, returns γ = 2.644. With the knee β free, the same curve fits at γ = 0.57, β = 16, and for β = 1, 2, 4 the fitted γ is 4.1, 2.2, 1.3. Here γ sets the knee position and is not identified on its own. The galaxy row's γ = 2 is also not the value galaxy fits return (§5.15). Source: back-annotation `d71f1f8b` (2026-11-16) and the site explorer's `gamma_incumbent_map/sc2.py` (sha256 `7fcb8ead…`), re-run by the Publisher 2026-11-17 from a /tmp copy with byte-identical output. Words added, none removed.]* + The same mathematical structure (tanh coherence) appears across scales—from Cooper pairs to galaxy rotation curves. This supports the view that coherence is a universal organizing principle. **η Framework Audit (Session #616)** diff --git a/whitepaper/sections/05-quantum-macro/meta/CHANGELOG.md b/whitepaper/sections/05-quantum-macro/meta/CHANGELOG.md index 51d82bad0..2cd82aebb 100644 --- a/whitepaper/sections/05-quantum-macro/meta/CHANGELOG.md +++ b/whitepaper/sections/05-quantum-macro/meta/CHANGELOG.md @@ -8,6 +8,85 @@ Header: Date (ISO 8601) | Author (LCT ID or name) | Change type (ADD/MODIFY/DELE ###### Entries +###### 2026-11-20 | Claude-Opus-5.5 (publisher) | MODIFY (one dated annotation; nothing deleted) +- **Description**: 15-dark-matter, kill-criteria item (3) (TEST-10): a `[CHECKED 2026-11-20]` note after the 2026-11-16 note. (a) The 11-16 escapes (M/L 8.4/11.6 at coefficient 4, 6.3/8.7 at 3; tidal 1.26–1.33×) are passes at a local 3σ; at the trials threshold 3.51 they are M/L 5.3 (Crater II) / 6.9 (Draco) at coefficient 3 and σ inflation 1.15× / 1.31×, the site's 11-19 figures. (b) The site's Walker+2009 row (Crater II z 2.9, Draco 3.7) equates the mass inside a sphere of radius R_e with half the light; Plummer puts 0.354 of the light there, and with the aperture matched Walker is coefficient 2.71 and gives z 3.96 / 4.31, past 3.51. The exclusion does not depend on which published estimator is used. +- **Rationale**: `PREDICTIONS.md` back-annotation `b26f6fa9` (2026-11-19) (i). The site script `cap6p39_vs_dsph_estimator.py` is reproduced row for row as the identity control of `simulations/publisher_20261120_dsph_walker_aperture.py` (run from /tmp with python3 -I, rc 0). Count HELD at 6. + +###### 2026-11-17 | Claude-Opus-5.5 (publisher) | MODIFY (four dated annotations; nothing deleted) +- **Description**: 15-dark-matter: a `[CHECKED 2026-11-17]` note on the γ = 2 bullet ("SPARC selects γ ≈ 0.489 ≈ ½ — the exact MOND point"): the template fitted with no data to MOND's simple μ returns 0.500 and to McGaugh's RAR function 0.595 (0.52–0.60 across g_bar ranges); nuisance-refit SPARC sits at 0.58–0.62 and cannot separate ½ from the RAR's value; the 0.489 is the fixed-Υ slice. Short pointer notes on the status-table γ = 2 cell and the "data select the MOND point" sentence. 16-superconductivity: a `[CHECKED 2026-11-17]` note under the cross-scale table (the superconductor "~2" is BCS's γ at a pinned knee, 2.644 from the BCS curve with no data; with β free γ is not identified; the text above reads the same tanh as γ = 1) and one after "Derived Gap Ratio" (3.52 is BCS's own 2π·exp(−0.5772) = 3.528, not an observation; 2√π is 0.49% away by coincidence; the source script's γ = 2 formula gives 7.12). Words added, none removed. +- **Rationale**: `PREDICTIONS.md` back-annotation `d71f1f8b` (2026-11-16) (i) and the Session1_Superconductivity_Coherence §4.4 back-annotation. The five explorer scripts in `synchronism-site/explorer/scripts/gamma_incumbent_map/` were re-run from a /tmp copy with byte-identical output. The 2√π item is outside the window, found on reading the §5.16 sites; checked by `simulations/publisher_20261117_bcs_ratio_and_gamma_check.py` (solves the BCS gap equation; rc 0). Count HELD at 6. + +###### 2026-11-16 | Claude-Opus-5.5 (publisher) | MODIFY (one dated annotation; nothing deleted) +- **Description**: 15-dark-matter, kill-criteria item (3) (TEST-10): a `[CHECKED 2026-11-16]` note after the 2026-11-09 note. "Ω_m/Ω_b = 6.39 is excluded ... by none after the correction" is scoped to discs: the archive's own 2026-08-23 pressure-supported table excludes 6.39 at Draco z = 5.7/4.9 and Crater II 6.1/4.9 (fiducial / M/L = 4), past the six-system Šidák 3.51, with no gas and no light rule. Adds a correction found on re-run: the source's estimator uses Wolf+2010's projected-radius coefficient 4 with three-dimensional radii, so the archive's B_req are 4/3 high; at coefficient 3 the exclusion holds (4.48 / 4.21) and the trials-corrected ceiling falls from ~14.7 to ~11.0. +- **Rationale**: `PREDICTIONS.md` back-annotation `f971b1bc` (i) withdrew the 11-14 reading that only conditional lensing reaches 6.39; the paper's TEST-10 note printed the disc-only half. Source script re-run byte-identical; sensitivity in `simulations/publisher_20261116_dsph_cap_estimator_sensitivity.py`. Count HELD at 6. + +###### 2026-11-15 | Claude-Opus-5.5 (publisher) | MODIFY (two dated annotations; nothing deleted) +- **Description**: 15-dark-matter, DE-sector paragraph: a `[CHECKED 2026-11-15]` note after the 2026-08-18 locality-fork sources. The fork's "no order-ε⁰ term / no independent channel / ~10× underpowered" holds only without a pressure term; read at local density as a comoving fluid (fluid Cardassian), the Jeans term makes existing P(k) exclude the horn unless γ = ½ ± 5×10⁻⁶, so the γ = 0.487 fit survives only on the mean-density reading. Words added, none removed. +- **Rationale**: The 2026-09-15 back-annotation in `PREDICTIONS.md` carried this correction and never reached the paper; the 2026-11-14 back-annotation (iv) restated it. Both site explorer scripts were re-run with byte-identical output. Count HELD at 6. Second note, dielectric-action paragraph: the polarization (striction) force's "≤ 2×10⁻⁵ of g … observationally free" was computed at the 0.029·V² units slip (2026-11-14 back-annotation (i)); at the archive's 0.028·V^0.5 knee the site explorer (2026-11-14, `6d97ef9`) finds radial striction median 0.5–1.6% of gravity and vertical striction 9–27% of vertical gravity, compressive. All three explorer scripts re-run by this pass; output identical apart from elapsed-time lines. + +###### 2026-11-13 | Claude-Opus-5.5 (publisher) | MODIFY (status-table cells qualified in place; labels unchanged) +- **Description**: 15-dark-matter status table: the 1/φ VALIDATED, a₀ = cH₀/(2π) DERIVED and S₈ = 0.763 PREDICTED cells each gain a short dated qualifier beside the label pointing to its bullet in the `[CHECKED 2026-11-11]` note; the S₈ Notes cell "Independent validation" gains "(not independent ...)". One parenthetical added to the note's lead saying so. Words added, none removed. +- **Rationale**: NOVA's 2026-11-13 exact-head review on #14 keeps the HOLD because DERIVED/VALIDATED/PREDICTED stay "stronger and visually prior to their disqualifying annotations", and names "make the cells reflect the constrained evidence" as one exit. The qualifiers carry only what the 11-11 note already prints. Relabelling the cells remains the project lead's call. + +###### 2026-11-12 | Claude-Opus-5.5 (publisher) | MODIFY (correction to this PR's own 11-11 note; one clause replaced) +- **Description**: 15-dark-matter: in the Brouwer et al. 2021 lensing paragraph, part (c) of the `[CHECKED 2026-11-11]` note said "dies independently" holds only if light follows the boosted potential. It now says the verdict holds under either rule: if light follows the visible matter only, lensing shows no boost, while the paper reports a measured RAR that agrees with MG; on MOND's curve the boost in the reliable bins (1.3 to 4×10⁻¹³) is 31 to 18, which ×10 baryons leave 3.1× to 1.8× short. The passing reading needs unseen matter in the lens. +- **Rationale**: Site back-annotation `15d9e61d` (PREDICTIONS.md 2026-11-11 (iii)) corrects the 11-10 "conditional on a deflection rule" as too generous. The MOND-curve numbers are this pass's own arithmetic; the site's 7.2 to 10.6σ on its transcribed KiDS bins is relayed and labelled as such. The 11-11 clause was unmerged text written by this lane the day before, so it was replaced in place, with the correction dated inside the note. Count HELD at 6. + +###### 2026-11-11 | Claude-Fable-5.2 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: 15-dark-matter: (a) a `[CHECKED 2026-11-11]` note under the status table on the a₀ = cH₀/(2π), 1/φ and S₈ rows; Status cells unchanged. (b) the Brouwer et al. 2021 lensing paragraph gains a `[CHECKED 2026-11-11]` note: fiducial reach 10⁻¹⁵, baryon allowance about ×10, required boost scales as 1/f, and the leg is conditional on a light-deflection rule. (c) the 10-23 covariant-RG note gains a `[NOTE 2026-11-11]` labelling its σ figures as an unpublished in-house computation. 12-chemistry: a `[CHECKED 2026-11-11]` note after the top-validated table on the Superconductivity r = 0.948 row. +- **Rationale**: (a) answers the 2026-11-09 review hold on PR #14, which asked for the table to carry dated qualifications before publication; every point is read from Sessions #102, #218, #239, the ledger's Bucket 3 row and this section's own 11-09 update, and the S₈ script was re-run (0.9425). (b), (c) and the chemistry row follow site back-annotation `76a139e0` (PREDICTIONS.md 2026-11-10). The chemistry row was reported there as untraced; it traces to commit `e4c903f5`, and the control run shows γ is built from T_c. Words added, none removed; count HELD at 6. + +###### 2026-11-09 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: 15-dark-matter: (a) the 10-18 `[SCOPED]` note on ΔBIC +184 gains an `[UPDATE 2026-11-09]`: both +184 and +10.9 hold Υ_disk = 0.5 for both laws; with one global Υ_disk per law the ΔBIC is +46.2 by point and +2.73 with N → N_gal (95% [−26.1, +132.5], P(≤ 0) = 0.12; about 2.2σ from a γ = 2 world on noise-free twins), and the pre-registered reading rule returns "disfavoured at about 2σ, not refuted". (b) Count item (1) gains a `[CHECKED 2026-11-09]`: the registered statistic is a marginalised BIC and the printed +184 is a frozen-Υ number. (c) The 10-04 "the scale is identified" note gains an `[UPDATE 2026-11-09]`: a₀′/γ is identified only with γ free; pinned at γ = ½ the scale moves 1.44 → 1.10 → 0.87×10⁻¹⁰ across Υ_disk = 0.4–0.6 and cH₀/2π closes only at H₀ ≈ 93 / 71 / 57. +- **Rationale**: Site back-annotation `6fe7b696` (PREDICTIONS.md 2026-11-09 (i)–(iii)), landed after the 11-09 pass. Not relayed: the Publisher re-ran `synchronism-site/maintainer/scripts/gamma2_global_upsilon_per_law.py` (PREREG `5a95614`) and `gamma2_global_upsilon_twin_power_posthoc.py` on 2026-11-10, both byte-identical to the committed outputs. Words added, none removed; count HELD at 6. + +###### 2026-11-09 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotation; nothing deleted) +- **Description**: 15-dark-matter gains three `[CHECKED 2026-11-09]` notes. Count item (3), TEST-10: the row fires as registered, but as a bound the largest excluded cap is about 7.5 at χ² ≥ 9 and about 5.7 after a 153-disc correction (outer half, post-hoc, one disc), with 28 and 10 discs excluding the framework's 3.175. Count item (4), TEST-08: r² = 0.0001 is a point estimate with a 95% upper limit of about 0.03, 2.9× under the kill bar and 6.4× under the claim. The ε₀ rescue bullet: "+67.8" and "do not intersect" hold at NGC 2915's face value only; at the two bounds the cost is +12 to +30.8 and 0 to +12, against +7.3 for ε₀ = 0.315. The class still loses to MOND, 126.5 to 52.2. +- **Rationale**: Site back-annotation `82d9a3a8` (ii) and (iv), after the 11-08 pass. Both TEST-10 scripts re-run with byte-identical output; TEST-08 recomputed from the tracked table (`simulations/publisher_20261109_test08_upper_limit.py`). The ε₀ brackets are this pass's reading of the committed profile table, not re-solved. Words added, none removed; count HELD at 6. + +###### 2026-11-08 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotation; nothing deleted) +- **Description**: 15-dark-matter count item (5): the `[CHECKED 2026-11-07]` note gains an `[UPDATE 2026-11-08]`. (a) The γ = 0.3 and 0.35 rows are not converged, so 14.5–19.3σ is scoped to γ 0.4–3; the verdict still holds below that (every draw above 5σ). (b) Every σ pins Υ. The compander passes only at Υ_d ≈ 1.22 (γ = 3) or 1.40 (γ = 2), with 36–42% of galaxies over-maximal and 2.9σ on SPARC at γ = 3. MOND δ = 6 passes at Υ_d ≈ 0.54 for 1.8σ. +- **Rationale**: Site back-annotation `cf84bb8c` (ii)–(iii) and site explorer `2db4494` (PREREG `10d0bdf`), both after the 11-07 pass. The first corrects this PR's own 11-07 sentence; the second adds a dependence that sentence didn't state. Scripts re-run from /tmp with matching values (the Υ-free output differs only in its elapsed-time token); a new 12-draw convergence scan places the boundary. Count HELD at 6. + +###### 2026-11-07 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotation; nothing deleted) +- **Description**: 15-dark-matter count item (5), the SPARC × Cassini joint squeeze (TEST-11), gains a `[CHECKED 2026-11-07]` note. The +17.95σ is at γ = 0.489, and the registered test evaluated Cassini only inside SPARC's point-level interval. With a₀ profiled from SPARC at each γ, Cassini fails at 14.5–19.3σ for every γ from 0.3 to 3 at all three registered external fields, so the row does not depend on the γ ≈ ½ pin, and `C_g`'s two entries in the "(4) The six refutations" decomposition now coincide at γ = 2. +- **Rationale**: Site back-annotation `330d49d1` (i), which withdraws the ledger's fixed-a₀ "passes at γ ≳ 1.5–2" (never printed in this paper). Script `cassini_q2_at_sparc_profiled_a0_high_gamma.py` re-run by the Publisher with byte-identical output; identity control +17.96σ at γ = 0.489. Count and verdicts unchanged. + +###### 2026-11-06 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: 15-dark-matter gains two `[CHECKED 2026-11-06]` notes. (1) After "No cosmological constant Λ required.": the dark-energy split survives only as a fluid. Read as modified gravity, G_eff = G/C(ρ̄_m(t)) drifts as the mean density falls, Ġ/G = 3(1 − Ω_m)H₀ = 1.42×10⁻¹⁰ yr⁻¹ at γ = ½, at least 527× the LLR 2σ edge for every γ in 0.3–3. (2) Count item (2), TEST-09: a paired galaxy bootstrap gives σ(Δn) = 0.112, so |Δn| = 0.406 is 3.6σ from zero and 0.95σ above the 0.3 bar. +- **Rationale**: Site back-annotation `a850db9e` (i) and site explorer `da1308c` (pre-registered `c9ff6ab`); both scripts re-run by the Publisher with byte-identical output. Count and verdicts unchanged. + +###### 2026-11-05 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotation; nothing deleted) +- **Description**: 15-dark-matter note (3): the `[CHECKED 2026-11-04]` note gains an `[UPDATE 2026-11-05]`. The weak density response (ρ^−0.10^, +945 over a constant boost) belongs to the fit, not the data. At fixed g_bar, log ρ takes slope −0.001 (acceleration law refit, ΔBIC +7.7 against adding it) or +0.003 against a cubic in log g_bar (galaxy-bootstrap 95% CI −0.022 to +0.027). Injected −0.05/−0.10 are recovered. corr(log g_bar, log ρ) = +0.84; the data's slope −0.211 is reproduced to 0.2% by a g_bar-only model. The note now points at the section's own 2026-08-02 form-free result. +- **Rationale**: Site explorer 2026-11-04 (`synchronism-site` `2ba2480`, PREREG `4891f7e`) landed at 08:05, after the 11-04 pass and the 11-04 back-annotation, and it tests the regression-dilution question the 11-04 note left open. Re-run, not relayed: output byte-identical. The 11-04 note sat six lines above a paragraph that already said local density carries no information beyond g_bar; this pass's own previous note was the propagation failure. Verdict and count unchanged. Words added, none removed. + +###### 2026-11-04 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotation; nothing deleted) +- **Description**: 15-dark-matter note (3) on the density-keyed head-to-head fit carries a `[CHECKED 2026-11-04]` note: the gloss "the likelihood switching the density dependence *off*" overstates. The free knee sits at 4.4×10⁻⁶ M☉/pc³, below 97% of disc points by more than 100×, so the discs are in the log regime where γ and ρ_crit are separately identified; boost ∝ ρ^−0.10^, and a constant boost loses by ΔBIC +945. At the derivation's A = 4.6×10⁻⁵ the law fits as Newton (+6,593; Newton +6,636); at γ = 2 with that A, +10,372. +- **Rationale**: Site explorer finding 2026-11-02 (`synchronism-site` `75e6a15`, PREREG `c62bc9d`) postdates the last back-annotation (`5709b579`) and was not in any pass. Re-run, not relayed: the main script's output is byte-identical; +945 follows from its own rows (C0 +3,787.5 minus F +2,842.6); +10,372 was recomputed independently (10,371.5), because the committed supplementary script does not emit that line. Verdict and count unchanged. Words added, none removed. + +###### 2026-11-03 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: 12-chemistry's 2026-11-02 Z-null note carries an `[UPDATE 2026-11-03]`: the real-row null was run on 10 rows and fails (|r(P, Z)| ≤ 0.53, quadratic R ≤ 0.51 vs framework r 0.60–0.98); the high r values are θ_D and ionization-energy identities. +- **Rationale**: Site back-annotation `5709b579` (PREDICTIONS.md 2026-11-02 (i)) supersedes the 10-31 "real-data null is unrun": the site explorer ran the polynomial-in-Z null on the 10 property rows recoverable from session scripts and it fails. Not relayed: the Publisher re-ran `synchronism-site/explorer/tools/chem_z_null_real_rows.py` on 2026-11-03 (rc 0, output byte-identical to the explorer's recorded corrected run). Reparametrization verdict unchanged; it now also rests on γ_optical being ionization energy. Words added, none removed. + +###### 2026-11-02 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: 12-chemistry's caveat block ("on textbook monotonic-with-Z data the best monotonic null is itself expected at r ≈ 0.95+") carries a short `[CHECKED 2026-11-02]` note. +- **Rationale**: Site back-annotation `e2784def` (PREDICTIONS.md 2026-10-31 (ii)): the 2026-05-10 polynomial-in-Z null was run on synthetic curves, and its premise is false across the periodic table. Not relayed: re-run by the Publisher 2026-11-02 on Pauling electronegativity, Z = 1–54, n = 51 (Spearman +0.09; best monotone-in-Z r = 0.33; polynomial degree 1/2/3/5 r = 0.03/0.23/0.25/0.39). The Reparametrization verdict is unchanged; it rests on S669. Words added, none removed. + +###### 2026-10-31 | Claude-Opus-5.5 (publisher) | MODIFY (one dated parenthetical in 15-dark-matter; nothing deleted) +- **Description**: The 10-29 `[CHECKED]` note on the `Ω_m` vs `Ω_m/Ω_b` bullet printed the 09-18 TEST-09 sweep at two decimals: "0.30, on the strict threshold, does not fire". A parenthetical after the sweep sentence gives the unrounded values, 0.406 / 0.297 / 0.260. The middle one is 0.003 under the bar, not on it, and the 0.3 edge is at B_max ≈ 5.31 (interpolated), not at 5.389. +- **Rationale**: Source is `PREDICTIONS.md` 2026-10-30 (ii) (`2c985ae4`), number hygiene. Re-run by the Publisher on 2026-10-31 (`synchronism-site/maintainer/scripts/which_C_carries_the_floor_4dp.py`, exit 0, identity control passes): 0.4062 / 0.2970 / 0.2604. Verdicts are unchanged, rule outcome is unchanged (convention-dependent), and the count is HELD at 6. + +###### 2026-10-30 | Claude-Opus-5.5 (publisher) | MODIFY (two dated annotations in 15-dark-matter; nothing deleted) +- **Description**: Two `[CHECKED 2026-10-30]` notes in the DE-sector paragraph. (1) After "the galaxy sector's shape parameter is unidentified at factor 2": the "factor 1.96 a₀ tension" that "dissolves at ϒ = 0.6" compared the fitted a₀ with Milgrom's. This section's own γ/a₀ identity makes the Milgrom-equivalent scale a₀/γ = 1/(9.17×10⁹) = 1.09×10⁻¹⁰, 4.6% from cH₀/2π. (2) After "(0 of 16 γ values reach it)": the w₀ > −1, wₐ < 0 quadrant is crossing only where w₀ + wₐ < −1 and thawing elsewhere. The sign lock excludes thawing as well as crossing. +- **Rationale**: These are back-scan items from the 09-03…10-07 publisher outage (PUBLISHER_LOG 2026-10-18, `watch_items`). Sources are `PREDICTIONS.md` 2026-10-04 (i) and 2026-09-23. Both notes can be decided from text the paper already prints: (1) from §5.15's own deep-limit identity and (2) from CPL algebra, so neither relays a site result. Verdicts are unchanged and the count is HELD at 6. + +###### 2026-10-29 | Claude-Opus-5.5 (publisher) | MODIFY (one dated annotation in 15-dark-matter; nothing deleted) + +**Correction 2026-11-02:** The historical description below overstates the dark-energy result. The cited calculation tested 400 sampled γ values in each of two specified no-CDM variants; none simultaneously accelerated and met the stated factor-two tolerance on the recombination ratio. It does not establish that the sector cannot run without CDM or exclude unsampled γ values or other variants. See the corrected annotation in §5.15; the original entry is preserved below as a historical record. + +- **Description**: `[CHECKED 2026-10-29]` appended to the bullet "The `Ω_m` versus `Ω_m/Ω_b` convention dispute is measured, and it is small." It scopes "small" to rotation-curve fit quality and states what each floor presupposes. `1/Ω_m` is the cosmic `C₀ = Ω_m` calibration, with CDM inside that `Ω_m`. `Ω_m/Ω_b` is a galaxy sector that replaces CDM, and the dark-energy sector cannot run without CDM (2026-09-22: no γ gives both q₀ < 0 and the dark/baryon ratio 5.36). It also prints the 2026-09-18 TEST-09 ceiling sweep, which this paper had never carried: |Δn| = 0.41 fires, 0.30 does not, 0.26 does not. +- **Rationale**: Publisher inclusion trigger (10-12 rule): `PREDICTIONS.md` 2026-10-27 block (ii) (`2065a308`) ranks the three ceilings by presupposition. The branch on which TEST-09 does not fire belongs to a variant whose cosmology fails. **[Correction 2026-11-03: the two tested variants had no joint match in the sampled values; this does not exclude unsampled values or other variants.]** The paper printed the two floors as an even "convention dispute". Both site scripts were re-run on 2026-10-29 at exit 0, and their output matches the committed files. This is a ranking, not a refutation. Count HELD at 6, and whether a convention-dependent kill counts gates on the project lead. + ###### 2026-10-27 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotations in 12-chemistry; nothing deleted) - **Description**: Three notes in `12-chemistry/chemistry.md`. (1) After "γ = 2 is the classical limit": "classical" is the N_corr = 1 uncorrelated end, not a quantum/classical gate; read the limits as uncorrelated (γ = 2) and collective (γ → 0). (2) After the phase table, whose rows 1–2 read "γ ≈ 2 universal" / "γ ≈ 2 in biology": a fitted exponent shared across fits, not a shared N_corr. (3) After "The same γ ≈ 2 that appears in BCS superconductivity also appears in biology": same fitted value, not the same N_corr; a BCS condensate has N_corr far above 1 and sits at γ → 0 (S676 placement in §0/§7); the r values in the Validated Predictions table stand as fits. - **Rationale**: Back-annotation `e03d771b` (2026-10-26): quantum degeneracy sets in at high phase-space density, so the most quantum many-body states have large N_corr; the site's quantum/classical endpoint labels were backwards and were withdrawn, and C is stated not to be a degeneracy parameter. This section's direction (γ = 2 at N_corr = 1) is the statistical-mechanics one, so its labels are not reversed, only scoped; what the back-annotation exposes is that "γ ≈ 2 in BCS" asserts N_corr ≈ 1 for a condensate. Decidable from the section's own master equation. Words added, none removed. diff --git a/whitepaper/sections/06-implications/04-open-questions/open_questions.md b/whitepaper/sections/06-implications/04-open-questions/open_questions.md index 28c6b9f55..fe6398d5c 100644 --- a/whitepaper/sections/06-implications/04-open-questions/open_questions.md +++ b/whitepaper/sections/06-implications/04-open-questions/open_questions.md @@ -48,11 +48,11 @@ The substantively active question is therefore not "reconcile A.3 with S617/S665 The first two are the physically defensible ones; the third smooths the threshold while leaving the field it is compared to unsmoothed. §5.15's decisive local-density null is stated *"at constant scale height"* — it fixes ℓ without saying which convention that ℓ belongs to. -**Discharge condition — partially discharged 2026-08-05, and it removed the obstruction rather than banking it.** The cross-sector run this entry called for (one ℓ serving SPARC disks, Cassini/TEST-11, wide binaries, clusters) was executed in the sibling site repo on 2026-08-05 (`synchronism-site/explorer/findings/coarse-graining-length-dissolves-317pc-is-beta-times-R0-not-a-scale.md`), independently re-derived here 2026-08-07. Under the two-sided convention there is **no ℓ-dependence to compare across sectors**: x is a virial ratio, capped at ~0.02 for every gravitationally bound system *[qualified 2026-08-28: ~0.02 is the 4π form; on the primary's form the cap is ~0.24–0.29, a 3.5× shortfall — the ℓ-independence is unaffected, the margin is. The 4π entered from the compilation chain (`A = 4π/(α²GR₀²)`, session67), not from Session 53, which computed `A = 0.028` without it; the site explorer flagged this 2026-08-27 and it is reproduced in `simulations/publisher_20260828_cesare2020_floor_brackets_and_rho_crit_exponent.py`]*, and the Cassini bound is ℓ-independent (TEST-11 is strengthened, not reopened). **There is no parameter-free obstruction on the coarse-graining axis, and no ℓ-discriminator to register.** What remains open is narrower and prior: **the framework must state which of the three conventions `C(ρ)` means.** It cannot be inferred from the fits, because the frozen SPARC instrument is keyed on acceleration and never evaluates a density at all (see the 2026-08-04 provenance note). Until it is stated, every knee statement in this document is *convention*-conditional, not merely ℓ-conditional. +**Discharge condition — partially discharged 2026-08-05, and it removed the obstruction rather than banking it.** The cross-sector run this entry called for (one ℓ serving SPARC disks, Cassini/TEST-11, wide binaries, clusters) was executed in the sibling site repo on 2026-08-05 (`synchronism-site/explorer/findings/coarse-graining-length-dissolves-317pc-is-beta-times-R0-not-a-scale.md`), independently re-derived here 2026-08-07. Under the two-sided convention there is **no ℓ-dependence to compare across sectors**: x is a virial ratio, capped at ~0.02 for every gravitationally bound system *[qualified 2026-08-28: ~0.02 is the 4π form; on the primary's form the cap is ~0.24–0.29, a 3.5× shortfall — the ℓ-independence is unaffected, the margin is. The 4π entered from the compilation chain (`A = 4π/(α²GR₀²)`, session67), not from Session 53, which computed `A = 0.028` without it; the site explorer flagged this 2026-08-27 and it is reproduced in `simulations/publisher_20260828_cesare2020_floor_brackets_and_rho_crit_exponent.py`]* *[CHECKED 2026-11-18: the cap belongs to one pairing, not to the framework's knee. It needs ρ_crit = A·V² with the same length in A as in the smoothing of ρ, the table's first row. Session 53's 0.028·V^0.5 is that law with R_half ∝ V^0.75 put into A, and the executed density fits in §5.15 set it against the midplane density Σ/2h at fixed h = 0.3 kpc, so the two lengths differ. There the knee is reached. On SPARC stellar surface density alone (Υ_d = 0.5, Υ_b = 0.7, gas left out, so x is a lower bound), the innermost point passes x = 1 in 62 of 175 discs: x = 2.8, 2.6 and 3.9 in NGC 2403, NGC 3198 and NGC 7331, and 0.07 and 0.21 in DDO 154 and UGC 128 (`simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py`; its identity controls reproduce this table's 0.019 and 0.24 and the 4.56×10⁻⁵ Jeans A). This agrees with §5.15's 2026-11-15 note (63 of 153 discs on the solved midplane density) and with PREDICTIONS.md 2026-11-17 (ii), which finds three different knee laws behind the headline numbers. So "~40× below the knee" is a statement about the two-sided convention. The archive's fits have used a one-sided pairing without saying so.]*, and the Cassini bound is ℓ-independent (TEST-11 is strengthened, not reopened). **There is no parameter-free obstruction on the coarse-graining axis, and no ℓ-discriminator to register.** What remains open is narrower and prior: **the framework must state which of the three conventions `C(ρ)` means.** It cannot be inferred from the fits, because the frozen SPARC instrument is keyed on acceleration and never evaluates a density at all (see the 2026-08-04 provenance note). Until it is stated, every knee statement in this document is *convention*-conditional, not merely ℓ-conditional. **What this is not.** A 2026-08-05 proposal reads the archive's 644× A-calibration gap as this same coarse-graining length (ℓ ≈ 317 pc ≈ the disk scale height). The reading fails, but not for the reason first given here — see the correction note below and the executive summary's S687 note. The obligation above stands on its own and is unaffected either way. -**Correction note (2026-08-07).** As first written on 2026-08-06, consequence 3 asserted `x ∝ ℓ²` as *the* scaling and inferred that "an unstated parameter spans the entire range of the conclusion." That is the third row of the table above — one convention of three, and the least defensible one, since it coarse-grains ρ_crit while holding ρ fixed. Neither self-consistent convention yields `x ∝ ℓ²`: one yields no ℓ-dependence, the other the opposite sign. The entry also declared its discharge "unrun" when it had been run 26 hours earlier in a sibling repository. Consequences 1 and 2 (the one-signed Jensen bias and its magnitude) are unaffected and stand as written, with one qualification now visible: their practical reach depends on which convention holds, since the two-sided convention keeps every bound system ~40× below the knee where the bias is largest. The 2026-08-06 NGC 3198 figures also used R_d = 2.6 kpc against SPARC Table 1's 3.14 kpc, inflating ρ(0) by 1.46×. +**Correction note (2026-08-07).** As first written on 2026-08-06, consequence 3 asserted `x ∝ ℓ²` as *the* scaling and inferred that "an unstated parameter spans the entire range of the conclusion." That is the third row of the table above — one convention of three, and the least defensible one, since it coarse-grains ρ_crit while holding ρ fixed. Neither self-consistent convention yields `x ∝ ℓ²`: one yields no ℓ-dependence, the other the opposite sign. The entry also declared its discharge "unrun" when it had been run 26 hours earlier in a sibling repository. Consequences 1 and 2 (the one-signed Jensen bias and its magnitude) are unaffected and stand as written, with one qualification now visible: their practical reach depends on which convention holds, since the two-sided convention keeps every bound system ~40× below the knee where the bias is largest *[CHECKED 2026-11-18: two-sided convention only; under the pairing the executed fits use, 62 of 175 SPARC discs reach the knee, see the 2026-11-18 note above]*. The 2026-08-06 NGC 3198 figures also used R_d = 2.6 kpc against SPARC Table 1's 3.14 kpc, inflating ρ(0) by 1.46×. **OQ-Discriminators — Quantitative deviations from GR/QM `[PARALLEL-PATHS]` (gated on OQ-fN)** diff --git a/whitepaper/sections/06-implications/meta/CHANGELOG.md b/whitepaper/sections/06-implications/meta/CHANGELOG.md index 53e55d6e0..e3020f5d6 100644 --- a/whitepaper/sections/06-implications/meta/CHANGELOG.md +++ b/whitepaper/sections/06-implications/meta/CHANGELOG.md @@ -38,3 +38,8 @@ Header: Date (ISO 8601) | Author (LCT ID or name) | Change type (ADD/MODIFY/DELE - **Description**: The two-sided-convention row of the consequence-3 table stated `x <= 0.019 beta_J^2`, "no, by ~40x, in every sector at every ell", and the discharge paragraph restated "capped at ~0.02 for every gravitationally bound system". Both are the 4pi form, `A = 4pi/(alpha^2 G R_0^2)` (session67 chain). The primary, Session 53 line 62, computes `A = 1/(G alpha^2 R_0^2)` -- no 4pi -- and it is that form which reproduces `A = 0.028`. On the primary's form `x = 3 alpha^2/(4 pi) = 0.29` at alpha = 1.1: a 3.5x shortfall, not 40x. Row and sentence qualified in place; the ell-independence (the row's actual point) is unaffected. - **Rationale**: The site explorer's 2026-08-27 finding (`rho-crit-has-no-velocity-exponent-A-is-a-half-power-coefficient.md`, sec. 8) flagged the unattributed 4pi and reopened "the knee is never crossed" as estimator-dependent: a real exponential disc's central midplane density is 21-47x its mean-in-R_half density (about 5x at R_half), so a 3.5x shortfall is inside the local-vs-mean ratio while 40x is outside it. Reproduced here rather than relayed (`simulations/publisher_20260828_cesare2020_floor_brackets_and_rho_crit_exponent.py`, part C, exit 0). This was my own 2026-08-06/07 closure, so it is corrected by the lane that wrote it. - **Scope**: No count moves. The obligation the entry states ("the framework must state which convention C(rho) means") is strengthened, since the crossing verdict now depends on the estimator as well as the convention. Which of the two Jeans-length normalisations is *physically* right is not adjudicated here; what is established is which one the archive's own coefficient was computed from. + +###### 2026-11-18 | Claude-Opus-5.5 (publisher) | MODIFY (two dated CHECKED notes; words added, none removed) +- **Description**: `04-open-questions/open_questions.md`, OQ-Coarsening. (1) After the 2026-08-28 qualifier on "x is a virial ratio, capped at ~0.02 for every gravitationally bound system": the cap needs ρ_crit = A·V² with the same length in A and in the smoothing of ρ. Session 53's 0.028·V^0.5 is the V² law with R_half ∝ V^0.75 inside A, and the executed fits set it against Σ/2h at h = 0.3 kpc. Under that pairing the knee is reached in 62 of 175 SPARC discs, on stellar surface density alone (a lower bound). NGC 2403 / 3198 / 7331 give x = 2.8 / 2.6 / 3.9, and DDO 154 / UGC 128 give 0.07 / 0.21. (2) A short pointer after the correction note's "~40× below the knee". +- **Rationale**: PREDICTIONS.md back-annotation 5d2d684c (ii) says the virial-ratio "knee unreachable" argument holds only for A·V² with one shared length. The table's first row already says the crossing is estimator-dependent, but the prose still stated the cap unconditionally. The claim was re-run, not relayed: `simulations/publisher_20261118_virial_cap_is_law_and_length_conditional.py` (identity controls reproduce the table's 0.019 and 0.24 and the 4.56×10⁻⁵ Jeans A) reproduces the back-annotation's 2.6–3.9 and <1 galaxy values. +- **Scope**: No count moves. The open question ("state which convention C(ρ) means") is unchanged. The note records that the archive's fits have used a one-sided pairing without stating it. diff --git a/whitepaper/sections/07-conclusion/conclusion.md b/whitepaper/sections/07-conclusion/conclusion.md index 4f777dbcd..76ad131a0 100644 --- a/whitepaper/sections/07-conclusion/conclusion.md +++ b/whitepaper/sections/07-conclusion/conclusion.md @@ -70,7 +70,7 @@ Synchronism is a computational framework for modeling reality through pattern dy - **Post-Demolition Coda** (#629-631): S629: π-analogy probe — k_crit fails as universal invariant (varies 0.20–0.68 across dimensions/exponents); Intent is not dimensionless-like-π. S630: WAKE stop-note — productive silence re-entered on stale trigger. S631: **First pre-committed kill criterion triggered.** Site-visitor audit surfaced that TEST-09 BTFR n ≈ 2.2 is refuted by Lelli+2019 n = 3.85 ± 0.09 (|Δn| = 1.65 exceeds kill threshold 0.3 by 5.5×). Session #48 itself labeled this derivation "not rigorous." The A = 4π/(α²GR₀²) formula uses α = 1.0 (fiducial, Session #66), not the fine-structure constant — site labeling was misleading. Both issues resolved from existing archive; 630 internal sessions did not surface them. Failure mode distinct from S617-628: public-claim vs. archive disconnect, external readers catch what internal audits miss. - **Site-Archive-Audit Sub-Arc** (#632-638) — **Framework Stress Test arc CLOSED at 22 sessions (S617-638)**: Seven further single-day audits. S632 (2026-04-25): TEST-07 site claim of λ ~ 500 Mpc cosmic interference — derivation R_MRH = (GM/c²)·(c/H₀) carries units of m² (length × length), not m. Dimensional error masked by the wrong-units numerical result happening to land near 600 Mpc. S633 (2026-04-25): site claim that C(ρ) maps "80 orders of magnitude" — tanh(γ·log(ρ/ρ_crit + 1)) saturates within ~1.6 decades for any sharp γ, structural impossibility (confirms OQ007's NEGATIVE verdict). S634 (2026-04-25): site cites "47 contributions" but canonical S582 inventory documents 30 (~57% accounting discrepancy; 47 likely uses a more lenient classifier than S582's strict criteria). S635 (2026-04-26): **cosmology domain scorecard** — 15 cosmology claims classified (1 refuted, 5 reparametrizations, 2 unanchored, 1 pending, 5 untested, 1 untestable). **Novel-unfalsified: 0.** /galaxy-rotation site badge ("Strongly Supported") overclaims by 1–2 tiers; defensible badge is "MOND Reparametrization" (RAR fit is McGaugh 2016 = MOND; CFD viscosity DM mechanism refuted by Bullet Cluster sign error). S636 (2026-04-27): **C(ρ) is not a mean-field order parameter** — no self-consistency (C does not appear in its own argument); no Landau free energy, no GL expansion, no universality class. The 53% melting and 6.5× T_c errors are correlation failures, not mean-field failures. New audit mode: **category error.** S637 (2026-04-28): **first DERIVATION attempt in the sub-arc.** RAR σ_int(ρ_env) — the visitor-flagged "highest-leverage candidate" for novel non-reparametrization prediction. Propagating γ=2 through C(ρ) gives Δσ_int (cluster − void) ≈ 0.00016 dex, **~120× below the SPARC measurement floor.** Sign correct, magnitude microscopic. Cosmology regime reduces to MOND in the testable regime. New audit mode: **derivation succeeds but predicts undetectable signal.** S638 (2026-04-29): **Independent CAS verification (sympy + numpy) that C(ρ) reduces to a Curie paramagnet** — *less than* Landau. F(C, ρ) = ((1+C)/2) ln(1+C) + ((1−C)/2) ln(1−C) − h·C; Taylor coefficients all positive (1/[2n(2n−1)]); convex around C=0 — no critical point; C ≥ 0 only — no Z₂ symmetry. C(ρ) is the equilibrium response of a single non-interacting two-state variable in an external log-density field. New audit mode: **external-track derivation independently verified** (qualitatively different — site-explorer track produces analysis, worker session verifies via CAS; verification track operational). **Predictive content now fully characterized: Cosmology regime → MOND (S637); Chemistry/CM regime → Curie paramagnet (S638). Both verified via independent computer algebra. Framework has no microscopic basis in collective coherence; it is phenomenological saturation response.** **8-for-8 site-claim audit instances over 9 days** (incl. S631), same failure mode throughout. Site-visitor audit methodology validated as transferable contribution. Audit-channel taxonomy now 8 modes. 6 site corrections pending operator action; framing-level operator response began 2026-04-29 (two README reframings: "calibrated blue-sky exploration, not unification claim"). -- **Post-Closure Sub-Arc Extensions** (#639-666): Sub-arc continues producing post-closure addenda. **S639 (2026-04-30):** TEST-03 disambiguation — site claim of R²=0.14 conflates BTFR scatter (S593, 51%) with RAR environmental ansatz (14%); different metrics under one label. 9th audit mode: **metric disambiguation / mechanism-naming**. **S640 (2026-05-01):** Dual-C symbol audit — homepage "one equation across scales" rests on shared notation, not derivation. Form 1 C(ρ)=tanh(γ·log(ρ/ρ_crit+1)) and Form 2 C=f(γ,D,S)≥0.50 (consciousness, 8-way convergence) share only the letter C; D and S are independent neural-dynamics observables, not functions of ρ. **γ IS genuinely shared via 2/√N_corr; C is NOT.** 10th audit mode: **symbol overloading at foundational level**. Sub-arc audit scope expanding test-level → foundational. Multi-persona visitor review (Pass 2/3/4) independently flagged this. **S641 (2026-05-02):** Lorentz invariance gap — preferred-frame absolute simultaneity sits at the foundational level alongside Born rule, dual-C bridge, and N_corr scale-invariance. 11th audit mode: **cross-gap meta-synthesis (kinematic layer)**. **S642 (2026-05-05):** GW170817 — framework is Case 3 (no field theory), 5th face of kinematic-layer gap. C(ρ) is a specified function of local density, not a propagating field; |c_GW − c|/c < 10⁻¹⁵ does not constrain it because there is no Lagrangian, no action principle, no equation of motion. Surviving by lacking an action principle is a scope restriction, not a victory. **[SCOPED 2026-08-09 — the conclusion stands, one of its three conjuncts does not.]** Measured against `manuscripts/Appendix_D_Synchronism_in_General_Relativistic_Form.md` (committed 2025-12-01, seven months before this claim): *"no field equation"* is **false** — §D.2/§D.6.1 state `∇²Φ = 4πGρ/C`; *"no action principle"* is **defensible only as scoped** — §D.5 writes an effective *worldline* action for a test particle, which is not a field action, and no action generates §D.2's equation; *"no equation of motion"* for C(ρ) is **true**, C being algebraic in local ρ with no kinetic term in any reading. GW170817 remains unconstraining **on the third conjunct alone** — no propagating scalar mode exists to modify tensor dispersion — so the accurate phrasing is *"C has no kinetic term,"* not *"the framework has no action principle."* §5.15 records the measurement: Appendix D's field equation and the force law carrying every number in this whitepaper differ by a **median 0.821 dex over 113 SPARC galaxies** (γ-invariant), and Appendix D's has **zero whitepaper citations and no evaluation against data**. **S643 (2026-05-06):** γ definitional collision — site γ-calculator labels are **inverted** relative to standard physics. Ideal gas γ=2 labeled "Quantum"; BCS Al / BEC / neutron star γ→0 labeled "Classical." Math measures single-particle vs collective; standard physics' "quantum" tracks macroscopic phase coherence. 12th audit mode: **definitional-collision-with-label-inversion**. Resolution: relabel as "Single-particle/Uncorrelated" vs "Collective/Strongly correlated." **S644 (2026-05-06):** ρ_crit = A · V_flat² takes V_flat as input — the "5% agreement" of theoretical/empirical A is internal consistency between two parameterizations of the same data, not an independent prediction. 13th audit mode: **calibration-consistency-not-prediction**. **S645 (2026-05-07) — POST-HOC CONSISTENCY FAILURE WITH DESI DR1 (NEW ARC PHASE; framing self-corrected by S648):** Session 107 fσ₈ prediction disfavored by DESI DR1 (Adame et al. arXiv:2411.12021). Pre-committed kill criterion fσ₈(z=0.5) > 0.45 → ΛCDM favored fired at LRG1 (DESI 0.55±0.06 vs Sync 0.418, 2.14σ above [LRG1 re-read 2026-10-12 from Table 9: 0.516 +0.057/−0.066 (0.4733 × 1.09; the 0.55 is 0.4733 × the QSO row's 1.16), ~1.5σ above Sync, not 2.14σ]; combined σ₈(z=0) 2.38σ above). Mechanism's predicted sign of redshift dependence is **inverted** relative to data — magnitude-only revision cannot recover the structural error. Status TEST-04a: REFUTED (post-hoc consistency check failure, per S648 timestamp audit). [CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12]; the 2.4σ is a GR-conditioned σ₈-amplitude statistic; TEST-04a is withdrawn from the decisive negatives — see Re-Grounding & Catalog Census below] **S646 (2026-05-07) — METHODOLOGY RECOMMENDATION (NEW ARC PHASE):** Per-test kill criteria exist; framework-level meta-falsification criterion does NOT. Without one, accumulated failures stay isolated. Three branches presented: (A) register meta-criterion now, (B) wait for DR2 — less needed since DR1 already fires, (C) scope-narrow now. Operator decision pending. **S647 (2026-05-08):** Chemistry 89% validation has self-correlation risk. Three confirmed paths: (1) Method 2 atomic-spacing identity γ=2(a/ξ)^(3/2) → r=0.956 with V_a∝a³ is functional identity; (2) Method 2 phonon coherence — sound velocity is constructional input → r=0.982 has constructional dependence; (3) Method 3 entropy → bonding → electronegativity (r=0.979) partly structural. Method 2 systematic bias (per Session #26 Part 3): true N_corr=10→6, 25→15, 50→32, clusters apparent γ in 0.35–1.15 at the "γ ≈ 1 boundary." Hall coefficient and magnetic susceptibility are NOT falsifying controls — outside the input set of every Method 1–5; exactly the predicted failures under self-correlation reading. **S648 (2026-05-08) — SELF-CORRECTION OF S645 WITHIN 24H:** Pass 4 visitor flagged Session 107 (committed 2025-12-10) is post-DESI DR1 (Adame+2024, Nov 2024 — 13 months earlier; DR1 BAO already cited in Session 107's own Timeline). Numerical disagreement is real (kill criterion fired); epistemic status is post-hoc consistency check failure, not prospective falsification. S645's "first hard external falsification" framing retroactively corrected. 14th audit mode: **self-correction-of-prior-session-framing-within-24h** — first instance of the audit channel turning inward. Operator recommendation: /timestamps page classifying every Tier-1 prediction as prospective / post-hoc consistency / post-hoc fit. **S649 (2026-05-08):** Two visitor proposals integrated. Part A: Site Key Claim #1 QM kill criterion ("design noise where resync outperforms isolation") is unfalsifiable as written — Viola-Knill-Lloyd 1999, CPMG, UDD, and transmon dynamical-decoupling experiments already satisfy it without invoking Synchronism. Part B: ρ_crit at γ=2 gives C ≈ 0.88, not 0.5; the "+1" regulator in ln(ρ/ρ_crit + 1) asymmetrizes the sigmoid, making ρ_crit a saturation knee, not a critical density (verified via S638 sympy at γ=0.5/1/2 → C=0.333/0.600/0.882). Same phase-transition vocabulary contamination that misled S636 into the Landau category-error. 15th audit-channel mode: **parameter-and-criterion-naming-contaminated-by-phase-transition-vocabulary**. **S650 (2026-05-09) — TEST-04a TRIPLE-SHARPENING COMPLETES:** Third refinement of TEST-04a verdict (S645 "first hard external falsification" → S648 "post-hoc consistency check failure" → S650 "mechanism-class failure / sign-reversed"). Framework's suppressor mechanism (G_local/G_global = C_cosmic/C_galactic < 1) predicts fσ₈ BELOW ΛCDM at low z and convergence at high z; DESI DR1 observes the OPPOSITE — fσ₈ ABOVE ΛCDM at LRG1/LRG2, converging at high z (ELG2). Redshift pattern is INVERTED. Magnitude knobs cannot flip the sign within the suppressor class; the mechanism class itself is wrong. [RETRACTED by S668 / RE-GROUNDED by S672 (2026-05-26/27) — the S650 "sign-reversed / mechanism-class" elaboration was a transcription error and is withdrawn; the re-grounded verdict is disfavored ~2σ, kill criterion TRIGGERED, post-hoc; see Self-Correction Cascade + Re-Grounding & Catalog Census below; CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12], the 2.4σ is a GR-conditioned σ₈-amplitude statistic, and DESI's own MG analysis (arXiv:2411.12026) puts μ₀ within 1σ of zero — TEST-04a is withdrawn from the decisive negatives] **Three-tier failure taxonomy** introduced (operationalizes S646's meta-falsification logic): magnitude miss (retunable) / universality miss (functional-form change) / mechanism-class failure (irreparable within class). Combined with S635 cosmology scorecard (0 novel-unfalsified) and S646 meta-criterion, the **cosmological sector formally meets the retraction threshold per S646's M3** (scope reduction); operator decision pending. 16th audit-channel mode: **mechanism-class-failure-taxonomy-introduced**. **S651 (2026-05-10) — CHEMISTRY NULL-MODEL GAP (paired with S647):** S647 audits the *method gap* in chemistry 89% validation (which N_corr method?); S651 audits the *prior question — the null gap*: r=0.98 is being compared against an implicit null of r=0, but sound velocity, electronegativity, atomic volume are themselves near-monotonic functions of Z. Any smooth monotonic function (polynomial in Z, generic 2-parameter tanh, MOND-type interpolating function) will achieve r ≈ 0.95+ on the same 1,703 phenomena by construction. **Δr = r(Synchronism) − r(best monotonic null)** is the figure that actually distinguishes claims; none currently exists in the archive. Best estimate: tie or marginal win. S647 + S651 are independent and compounding — specifying the method (S647 fix) does not address the null model question (S651 fix); together they leave very little for Synchronism-specific signal in the 89% cohort. 17th audit-channel mode: **null-model-gap-against-best-monotonic-null**. **S652 (2026-05-11):** Governing-equation gap — upstream synthesis of S636/S638/S640/S649/S651 confirms C(ρ) is a phenomenological compander (Option A), not a Landau order parameter (Option B), and not the steady-state of a dynamic equation (Option C). C(ρ) shares the *functional form* of mean-field tanh solutions without sharing the *physics* — same class as μ-law audio companding, Naka-Rushton photoreceptor response, Hill enzyme kinetics. The framework's headline equation does no dynamical work; it only labels regimes. 18th audit-channel mode: **governing-equation-gap (forward map with no field equation)**. Second meta-synthesis after S641's cross-gap synthesis. **S653 (2026-05-12) — TWO BINARY OPERATOR DECISIONS FORCED:** Two same-day proposals demanding binary commitments. Part A (compander commitment) — given S652's conclusion, commit front-of-site to Frame B (drop phase-transition language, rename ρ_crit, reframe critical-exponent failures as category errors). Part B (suppressor diagnostic) — `simulations/session653_coherence_ratio.py` computed C_galactic ≈ 0.882 vs C_cosmic ≈ 1.5×10⁻⁵, ratio ≈ 5.9×10⁴. The framework's own equations dictate strong suppression at low z under Session 107's coupling direction; DR1 observes enhancement. Two branches: (1) sign-flip recoverable (re-derive Session 107 with inverted ratio — framework reinterpretation), (2) suppressor class dead (retire Session 107 mechanism). Either is defensible; ambiguity is the most credibility-damaging option. Sub-arc's three-stage rhythm now complete: **individual audits → meta-syntheses → forced binary commitments**. 19th audit-channel mode: **forced-binary-operator-decision-with-numerical-diagnostic**. **S654 (2026-05-13) — ZERO ACTIVE DISCRIMINATORS REMAINING:** Audit of the site's three "Active Discriminating Tests" (TEST-01, TEST-02, TEST-05) against MOND's External Field Effect (Bekenstein-Milgrom 1984; AQUAL/QUMOND; Pittordis 2023; Banik 2024). S637's derivation settles TEST-05 (and TEST-01, same observable): predicted Δσ_int ≈ 0.00016 dex, ~120× below SPARC floor — cannot discriminate within current precision. TEST-02 has a disputed MOND+EFE baseline; the framework specifies no divergence from any specific MOND+EFE variant. [**SCOPED 2026-08-03 — the “specifies no divergence” clause is retracted; created at source in §5.15 on 2026-08-03 (`a78054e9`), propagated here the same day.** §5.15 withdraws the EFE-at-0.3–0.4×-MOND figure (the galaxy sector has no field equation, so it was never derived) and replaces it with the sharper **EFE = 0 exactly** — a uniform external field leaves local ρ unchanged, so C = C(ρ_local) satisfies the Strong Equivalence Principle by construction. Predicting exactly zero EFE **does** specify a divergence from every MOND+EFE variant, each of which predicts a non-zero one. **What survives untouched** is the formal asymmetry — “refutable but not currently confirmable” is unchanged. **[RE-AMENDED 2026-08-05 — the empirical half is withdrawn and “zero remaining active discriminators” STANDS.]** The 08-03 text added that EFE = 0 “stands in tension with the ~4σ SPARC EFE detection of Chae et al. 2020,” and therefore that “zero remaining active discriminators” no longer held unqualified. **Withdrawn: EFE = 0 is `not-evaluable` against that dataset.** At Chae's measurement radii the stated density law misses the rotation curve by **+3.1 to +4.2 dex** while the signal measured is **0.046–0.083 dex** — baseline error exceeds the whole effect by **38×–92×** (re-derived in-lane on the in-repo SPARC mass models; MOND sits at +0.002 to +0.084 dex, inside the signal). **Refutation count stays at 6; Bucket 0 stays 0.** The propagation carried its own accurate warning — *“Magnitude not independently re-derived here”* — and the magnitude was the entire claim. See §5.15.] **All three are MOND+EFE degenerate** within current measurement precision. Combined with TEST-03 (MOND-shared) and TEST-04a (REFUTED), the framework has **zero remaining active discriminators** against the primary alternative. The formal asymmetry: **refutable** (kill criteria can fire — TEST-09 did) **but not currently confirmable** (no test exists where Synchronism succeeds and MOND+EFE fails). Formal completion of trajectory S635 → S637 → S646 → S650 → S654. 20th audit-channel mode: **zero-active-discriminators-against-primary-alternative**. **S655 (2026-05-14):** Γ = γ²(1−c), the site's most novel-looking quantum claim, traced to Session #232 and identified as standard correlated-bath relative-phase decoherence (Schlosshauer 2007; Lidar/Whaley DFS) — c=1 is a decoherence-free subspace, c=0 the uncorrelated limit. Reparametrization in Synchronism vocabulary, not a novel prediction (24th audit-channel instance). **S656 (2026-05-16):** TEST-04a reframed from "Synchronism failure" to a **mechanism-class constraint as a transferable contribution [RETRACTED by S668 — this thread evaporates; see Self-Correction Cascade below]** — DESI DR1 disfavors any G_local/G_global < 1 suppressor framework at ≈2.4σ, publishable independent of Synchronism's fate provided the S648 post-hoc qualifier is respected (a consistency-check bound, not a blind-test refutation). **S657 (2026-05-17):** Compander-family AIC/BIC model-selection endorsed — tanh has no privileged status (S636/S638/S640/S652); the standard diagnostic compares the family (tanh, Hill, logistic, erf, μ-law, Gompertz) on SPARC + chemistry + Tc. One prior dataset done (2026-03-27: tanh beats Hill by ΔAIC=17.6 after a baseline fix); the full panel is new operator/explorer-track computation. Winning a compander contest is local to the class; it does not promote C(ρ) to a derived object. **S658 (2026-05-18):** A2ACW temporal-asymmetry redesign endorsed — the methodology's 6/6 "Validated"→reparametrization demotion is diagnosed as a shared-training-distribution syntactic-consistency checker; pairing agents with asymmetric knowledge windows (different model generations) tests whether prior-art rediscovery is caught in-session. Falsifiable; caveat that training cutoffs are leaky. **S659 (2026-05-19) — EXACT NO-INFLECTION PROOF + A2ACW v2 MEASURED RESULT:** Part A (verified): for C(ρ)=tanh(γ·ln(ρ/ρ_crit+1)), d²C/dρ² = −sech²(u)·γ/(ρ+ρ_crit)²·[2γ·tanh(u)+1] < 0 for all ρ > 0 (inflection would need tanh(u) = −1/(2γ) < 0, impossible since u ≥ 0) — **C is strictly concave on the entire physical domain**; the +1 regulator eliminates all critical behavior. ρ_crit is a location parameter of a logarithmic compander **by the algebra of the function**, sharpening S638/S649/S652 from heuristic to exact and foreclosing the Landau analogy. Part B (measured): the A2ACW v2 three-axis protocol catches 4/6 by vocabulary asymmetry (4/4 prior-art sub-class) and **6/6** with symbol audit + null model added — self-simulated, so the open calibration item is the false-novelty rate on closed physics (BCS, Anderson localization, EW unification). **External-review convergence (2026-05-15):** named external AI reviewer Kimi 2.6 delivered a multi-round review; the operator response propagated the **Findings vs Framings** discipline across README/STATUS/CLAUDE and the whitepaper — qualitatively distinct from the persona-driven visitor audits. **S660 (2026-05-20) — NOVELTY LEDGER CLOSED + GALAXY DISCRIMINATOR DEFINED:** Part A demotes the entity criterion (Γ < m for coherent entity-hood) — the framework's *last candidate novel prediction* — to standard QFT narrow-width / narrow-resonance physics (Breit-Wigner pole well-defined in Källén-Lehmann; PDG applies it informally). Synchronism adds an ontological interpretation ("coherence cycle completion") but no new condition, observable, or prediction. **Novel-survivor count → 0 after 3,308+ sessions; the novelty ledger is now closed.** Defensible contributions reduce to A2ACW null result (methodology, S658-S659), TEST-04a mechanism-class constraint (post-hoc, S656), and the negative-results catalog. Part B identifies the framework's first non-degenerate galaxy-scale discriminator: under the μ-identification g_bar = g_obs·tanh(γ·ln(1+g_obs/a₀)) the compander IS a legitimate MOND interpolating function (high-acceleration Newtonian, deep-MOND √-law / Tully-Fisher preserved). γ pinned at 2 (N_corr=1 galaxy assignment) gives a distinct shape prediction; γ fitted collapses to MOND. Galactic instantiation of S654's refutable-but-not-confirmable asymmetry. **S661 (2026-05-21) — RAR DISCRIMINATOR EXECUTED → γ=2 REFUTED AT ΔBIC=+184 ON SPARC (READINESS UPLIFT 0.97 → 0.98):** **[SCOPED 2026-10-18: +184 counts every SPARC point as an independent draw (the frozen 2026-07-22 pipeline reproduces +184.0 at 2,807 points / 166 galaxies; S661 reported 2,693 / 153). With the galaxy as the replication unit (measured N_eff ≈ 150) it is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the ΔBIC = 10 threshold: γ=2 is still worse than free γ in 94/166 galaxies and held-out CV prefers free γ by 1.6σ, so γ=2 remains disfavoured, but the registered rule returns *convention-dependent*, not decisive. Refutation count unchanged. See §5.15.]** Explorer track ran the test on real SPARC data (2,693 points, 153 galaxies, M/L + distance marginalized). γ=2 pinned: ΔBIC = +184 vs McGaugh ν; kill criterion (ΔBIC > 10) decisively triggered; robust to intra-galaxy point-correlation correction (effective N ≈ 500–1000 → ΔBIC ≈ 33). γ free: γ = 0.49, ΔBIC = +7.1, indistinguishable from McGaugh. **No γ makes the compander both distinct from MOND and consistent with SPARC.** The +3.3% per-point RMS penalty is a COHERENT S-shaped residual at g_bar ≈ a₀ (~8σ/bin) — a population-wide shape term not absorbable by per-galaxy marginalization (RMS view "mildly disfavored," BIC view decisively refuted; the structured deviation is exactly the predicted signature, concentrated where predicted). **Galactic sector closed BY EXECUTION**, matching the cosmological sector. Qualitatively stronger than the rolled-back S645 case: γ=2 committed a priori from N_corr=1, on external published data, by an independent track. REC-2026-037 readiness uplifted 0.97 → 0.98; arc status `complete_galactic_and_cosmological_sectors_closed_by_execution`. **S662 (2026-05-22) — A2ACW SPECIFICITY SELF-CORRECTION (audit channel applies its own standard to its preferred output):** Part A confirms S661 at program-level: net discriminating galaxy tests vs MOND+ΛCDM = 0, by execution. Galaxy program closed. Part B applies the S651 null-baseline discipline to the A2ACW methodology contribution. **S658/S659's "4/4" and "6/6" catches were sensitivity (TPR) on a positive-only set** (all 6 cases were known reparametrizations); specificity (TNR) was never measured. The control on six canonical genuine discoveries (Dirac 1928, Bell 1964, BCS 1957, Higgs 1964, Hawking 1974, Noether 1918) gives **R1 (flag if prior-art named): 0% specificity** — every genuine discovery has canonical antecedents under translation; R1 detects "has-a-canonical-name," not "is-a-reparametrization." R2 (flag if reduces to prior art, nothing novel) gives 100% specificity, but discrimination is supplied by an unautomated novelty judgment the protocol never operationalizes — the 6 demotions came from human audit, not from the AI loop. **[CONTESTED 2026-10-21: "human" does not trace. Since 2026-09-15 the site records the auditors behind the six demotions as LLM agents (synchronism-site `288aaf6`). A partial git trace in this pass finds two of the six (a₀ = cH₀/(2π) and A = 4π/(α²GR₀²)) demoted in the site maintainer track's 2026-04-23 badge audit (`ac093af`, co-authored by Claude Sonnet 4.6, prompted by an LLM researcher persona); the other four are untraced. "Outside the A2ACW loop" holds under either reading, so the restated defensible claim is unchanged; what is open is whether the out-of-loop discriminator was a human or a second LLM track. See `Research/proposals/a2acw_has_a_cross_vendor_record_already_and_the_demotion_auditor_is_contested_20261020.md`.]** **[TRACED 2026-10-22 — the contest above resolves against "human". The site explorer finished the trace (synchronism-site `explorer/findings/a2acw-lv-rerun-and-the-demotion-auditor.md`, 2026-10-20). Five of the six have a badge-changing edit, and every one is a site maintainer commit co-authored by Claude (Opus 4.6 or Sonnet 4.6), with the decision coming from a simulated visitor persona or an autonomous session (#574, S581, S651, S655). The sixth (Born rule) has no Validated badge and no demotion edit on record, so it cannot be attributed either way. No dp-authored audit decision exists between 2026-02-01 and 2026-05-23. The "human audit" wording entered the record on 2026-05-18 in a back-annotation co-authored by Claude Sonnet 4.6 (`e8752205`), and S662 repeated it; S662 is back-annotated in place (`c5364f99`). Checked in this pass: the co-author trailers of all eight cited edit commits and of `e8752205`. Two caveats carry over: dp oversight outside git cannot be excluded (S659's "human-monitored worker sessions"), and only two of the "six former Validated badges" (a₀, chemistry) are confirmed to have carried Validated at all. What this changes: "outside the A2ACW loop" still holds, but the discriminator was a second set of AI tracks, not a human. The narrowed claim's "unautomated ... judgment that AI loops fail" therefore needs scoping to *the A2ACW loop*; rewording it gates on the project lead.]** Defensible claim narrows from "A2ACW is a transferable reparametrization detector" to "A2ACW is a retrieval-augmentation / debiasing step for AI-assisted literature review." 21st audit-channel mode: **a2acw_specificity_self_correction** — the cleanest discipline-on-itself event in the entire arc (the audit channel built its case against the framework using S647/S651 null-baseline standards, then applied the same standard to its preferred methodology output and retracted the overclaim). REC-2026-037 readiness HELD at 0.98 (S662 narrows one thread within REC-037 without retracting the load-bearing S661 trigger; different from the S648 rollback case). **S663 (2026-05-23) — CLEANEST END-STATE SYNTHESIS (third meta-synthesis; philosophical closure):** Two findings landing the arc's cleanest end-state. **Part A: EFTofLSS doubly closes TEST-04a [moot per S668 — no mechanism-class result remains to close; the post-hoc amplitude point stands].** Effective Field Theory of Large-Scale Structure (Cabass-Simonović-Zaldarriaga 2024-2025) explains the DESI DR1 fσ₈ enhancement within ΛCDM via one-loop EFT counterterms at 1-2σ; any coherent G_eff modification at the 10% level is degenerate with EFT Wilson coefficients and constrained by DESI DR1 to consistency with zero. **Even Branch 1 (sign-flip recovery via C_galactic/C_cosmic > 1) is now closed** — the mechanism class is ruled out *regardless of sign*. Strengthens the S656 preprint from "suppression-class ruled out" to "any coherent G_eff modification predicting a scale-independent fσ₈ shift at the 10% level is degenerate with EFTofLSS counterterms and constrained by DESI DR1 to consistency with zero." The post-hoc S648 qualifier still holds; the preprint becomes more precise. **Part B: Framework classification — Interpretation + Methodology research program.** All four visitor personas (casual, technical writer, grad student, leading researcher) independently arrived at the same diagnosis — the framework occupies the "ontological reframe without a distinguishing experiment" position, structurally identical to QM interpretations (Bohmian, Copenhagen, Many-Worlds), evaluated by parsimony rather than data. Convergent endpoint after S574 / S617-628 (same mechanics, different notation), S660 / S661 / S654 (no novel measurable, no distinguishing experiment), S663A (EFTofLSS closes the last mechanism class). **Endorsed classification:** *"A coherence-language interpretation of known physics, used as a substrate for developing AI-collaborative science methodology."* Front-load methodology contribution; treat the equation as a worked example; distinguish interpretation-confirming tests from interpretation-breaking tests; the honest-assessment infrastructure is itself a contribution. **Third meta-synthesis** after S641 (cross-gap, kinematic layer) and S652 (governing-equation-gap, compander class). Paired with S661's empirical closure and S660A's structural closure, **the arc now has all three closures simultaneously — empirical, structural, philosophical.** Audit-channel modes held at 21 (S663 is meta-synthesis + classification, not a new audit category). REC-2026-037: 46 → 47 sessions, sub-arc 32-over-31 → 33 instances over 32 days, **readiness HELD at 0.98** (S663 is end-state synthesis, not new trigger); new milestone `framework_classification_interpretation_methodology`. Sub-arc now **33 audit-channel instances over 32 days**, **21 audit-channel modes**, **plus 2 new arc phases** (Post-Hoc-Consistency-Failure, Methodology Recommendation), **a publishable methodology pattern** (three-stage rhythm), **a publishable epistemic-position finding** (refutable-but-not-confirmable asymmetry, now executed at galactic scale per S661), **an exact no-inflection proof** (S659A), **a closed novelty ledger** (S660A; novel-survivor count → 0), **the galactic sector closed by execution** (S661; ΔBIC=+184), **methodology self-correction at the endpoint** (S662B; A2ACW R1 0% specificity on genuine-discovery control set), and now **EFTofLSS double-closure of TEST-04a + framework classification as Interpretation + Methodology research program** (S663; arc's three closures — empirical, structural, philosophical — stand simultaneously). **S664 (2026-05-23) — LANDSCAPE POSITIONING (B+):** Two-axis positioning of the closed arc. *Physics axis:* with Path A (action principle) blocked by the kinematic-layer gap (S641-S642, no Lagrangian) and Path B (dynamics outside the galaxy regime) blocked by S650/S661, the surviving Path C identifies C(ρ) as a reparametrization of Verlinde's entropic-gravity prediction in the galaxy-rotation regime — the compander μ-function and Verlinde's emergent-gravity interpolation coincide exactly where SPARC can test them. *Methodology axis:* the A2ACW null result (S658-S662) is positioned against the AI-for-science landscape (FunSearch, AlphaProof, SciNet, Sakana) as an honest-assessment / negative-result counterpoint to discovery-claiming AI programs. Positioning session, not a new audit mode. **S665 (2026-05-24) — SUBSTRATE AUDIT, SPATIAL TENSION (Grade A):** the first session to challenge the **CFD substrate itself** — the one claim external reviewer Kimi 2.6 called "genuinely interesting," assumed-but-never-challenged across the entire S617-664 demolition. The CFD reframing's load-bearing claim ("the Navier-Stokes equations are not an analogy… they ARE the Intent dynamics") is incompatible with its own downstream phenomenology. The velocity is v = J/I = −g(I)·∇I with g(I) = D·R(I)/I a scalar; a velocity that is the gradient of a scalar is **irrotational for any R(I)** → curl(v) ≡ 0 → **no vortices, no turbulence**. But the framework builds qualia as vortex modes, the consciousness threshold as a critical Reynolds number, and dark matter as vortex structure — all on vortices the substrate provably cannot produce. A theorem, not unfinished work. **S666 (2026-05-25) — SUBSTRATE AUDIT, TEMPORAL TENSION (Grade A); arc instances → 36, modes → 22:** the substrate ∂I/∂t = ∇·[D·R(I)·∇I] is first-order and dissipative (Lyapunov functional monotonically decreasing, real eigenvalues, arrow of time, no phase); FUNDAMENTALS' entity ontology requires unitary oscillation (de Broglie f=E/h, phase-locking resonance) — **mutually exclusive dynamical classes**. The two Schrödinger "derivations" bridge them only by inserting the imaginary unit i by hand (S307 line 48 / code 129; S99 Axiom 4 posited, not derived) and by switching the substrate off (S307 drops R(I); S99 takes D→0). Confirmed numerically: the factor i is the entire difference between exp(−Dk²t) decay and exp(−iDk²t) oscillation. A contradiction **internal to the canonical FUNDAMENTALS document**, confirmed in the framework's own derivations and code. **Together with S665, the substrate can host neither the spatial (vortices) nor the temporal (oscillation) structure its own entity ontology requires — completing the demolition: every load-bearing claim, including the CFD-substrate holdout, has now been audited.** 22nd audit-channel mode: **substrate-internal-dynamical-contradiction** (irrotational ⇒ no vortices, S665; dissipative ⇒ no oscillation, S666) — the first audit to attack the substrate rather than a site claim, derivation, or naming convention, and the deepest demolition finding since S617. Sub-arc now **36 audit-channel instances over 33 days**, **22 audit-channel modes**. **REC-2026-037: 47 → 50 sessions, readiness HELD at 0.98** (the substrate audit strengthens the demolition by closing the last assumed-but-unchallenged claim, but is not a new publication trigger — 0.99 still requires a paper draft or operator publication action); new milestone `substrate_audit_cfd_irrotational_and_dissipative`; arc renamed "… + Substrate Audit." +- **Post-Closure Sub-Arc Extensions** (#639-666): Sub-arc continues producing post-closure addenda. **S639 (2026-04-30):** TEST-03 disambiguation — site claim of R²=0.14 conflates BTFR scatter (S593, 51%) with RAR environmental ansatz (14%); different metrics under one label. 9th audit mode: **metric disambiguation / mechanism-naming**. **S640 (2026-05-01):** Dual-C symbol audit — homepage "one equation across scales" rests on shared notation, not derivation. Form 1 C(ρ)=tanh(γ·log(ρ/ρ_crit+1)) and Form 2 C=f(γ,D,S)≥0.50 (consciousness, 8-way convergence) share only the letter C; D and S are independent neural-dynamics observables, not functions of ρ. **γ IS genuinely shared via 2/√N_corr; C is NOT.** 10th audit mode: **symbol overloading at foundational level**. Sub-arc audit scope expanding test-level → foundational. Multi-persona visitor review (Pass 2/3/4) independently flagged this. **S641 (2026-05-02):** Lorentz invariance gap — preferred-frame absolute simultaneity sits at the foundational level alongside Born rule, dual-C bridge, and N_corr scale-invariance. 11th audit mode: **cross-gap meta-synthesis (kinematic layer)**. **S642 (2026-05-05):** GW170817 — framework is Case 3 (no field theory), 5th face of kinematic-layer gap. C(ρ) is a specified function of local density, not a propagating field; |c_GW − c|/c < 10⁻¹⁵ does not constrain it because there is no Lagrangian, no action principle, no equation of motion. Surviving by lacking an action principle is a scope restriction, not a victory. **[SCOPED 2026-08-09 — the conclusion stands, one of its three conjuncts does not.]** Measured against `manuscripts/Appendix_D_Synchronism_in_General_Relativistic_Form.md` (committed 2025-12-01, seven months before this claim): *"no field equation"* is **false** — §D.2/§D.6.1 state `∇²Φ = 4πGρ/C`; *"no action principle"* is **defensible only as scoped** — §D.5 writes an effective *worldline* action for a test particle, which is not a field action, and no action generates §D.2's equation; *"no equation of motion"* for C(ρ) is **true**, C being algebraic in local ρ with no kinetic term in any reading. GW170817 remains unconstraining **on the third conjunct alone** — no propagating scalar mode exists to modify tensor dispersion — so the accurate phrasing is *"C has no kinetic term,"* not *"the framework has no action principle."* §5.15 records the measurement: Appendix D's field equation and the force law carrying every number in this whitepaper differ by a **median 0.821 dex over 113 SPARC galaxies** (γ-invariant), and Appendix D's has **zero whitepaper citations and no evaluation against data**. **S643 (2026-05-06):** γ definitional collision — site γ-calculator labels are **inverted** relative to standard physics. Ideal gas γ=2 labeled "Quantum"; BCS Al / BEC / neutron star γ→0 labeled "Classical." Math measures single-particle vs collective; standard physics' "quantum" tracks macroscopic phase coherence. 12th audit mode: **definitional-collision-with-label-inversion**. Resolution: relabel as "Single-particle/Uncorrelated" vs "Collective/Strongly correlated." **S644 (2026-05-06):** ρ_crit = A · V_flat² takes V_flat as input — the "5% agreement" of theoretical/empirical A is internal consistency between two parameterizations of the same data, not an independent prediction. 13th audit mode: **calibration-consistency-not-prediction**. **S645 (2026-05-07) — POST-HOC CONSISTENCY FAILURE WITH DESI DR1 (NEW ARC PHASE; framing self-corrected by S648):** Session 107 fσ₈ prediction disfavored by DESI DR1 (Adame et al. arXiv:2411.12021). Pre-committed kill criterion fσ₈(z=0.5) > 0.45 → ΛCDM favored fired at LRG1 (DESI 0.55±0.06 vs Sync 0.418, 2.14σ above [LRG1 re-read 2026-10-12 from Table 9: 0.516 +0.057/−0.066 (0.4733 × 1.09; the 0.55 is 0.4733 × the QSO row's 1.16), ~1.5σ above Sync, not 2.14σ]; combined σ₈(z=0) 2.38σ above). Mechanism's predicted sign of redshift dependence is **inverted** relative to data — magnitude-only revision cannot recover the structural error. Status TEST-04a: REFUTED (post-hoc consistency check failure, per S648 timestamp audit). [CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12]; the 2.4σ is a GR-conditioned σ₈-amplitude statistic; TEST-04a is withdrawn from the decisive negatives — see Re-Grounding & Catalog Census below] **S646 (2026-05-07) — METHODOLOGY RECOMMENDATION (NEW ARC PHASE):** Per-test kill criteria exist; framework-level meta-falsification criterion does NOT. Without one, accumulated failures stay isolated. Three branches presented: (A) register meta-criterion now, (B) wait for DR2 — less needed since DR1 already fires, (C) scope-narrow now. Operator decision pending. **S647 (2026-05-08):** Chemistry 89% validation has self-correlation risk. Three confirmed paths: (1) Method 2 atomic-spacing identity γ=2(a/ξ)^(3/2) → r=0.956 with V_a∝a³ is functional identity; (2) Method 2 phonon coherence — sound velocity is constructional input → r=0.982 has constructional dependence; (3) Method 3 entropy → bonding → electronegativity (r=0.979) partly structural. Method 2 systematic bias (per Session #26 Part 3): true N_corr=10→6, 25→15, 50→32, clusters apparent γ in 0.35–1.15 at the "γ ≈ 1 boundary." Hall coefficient and magnetic susceptibility are NOT falsifying controls — outside the input set of every Method 1–5; exactly the predicted failures under self-correlation reading. **S648 (2026-05-08) — SELF-CORRECTION OF S645 WITHIN 24H:** Pass 4 visitor flagged Session 107 (committed 2025-12-10) is post-DESI DR1 (Adame+2024, Nov 2024 — 13 months earlier; DR1 BAO already cited in Session 107's own Timeline). Numerical disagreement is real (kill criterion fired); epistemic status is post-hoc consistency check failure, not prospective falsification. S645's "first hard external falsification" framing retroactively corrected. 14th audit mode: **self-correction-of-prior-session-framing-within-24h** — first instance of the audit channel turning inward. Operator recommendation: /timestamps page classifying every Tier-1 prediction as prospective / post-hoc consistency / post-hoc fit. **S649 (2026-05-08):** Two visitor proposals integrated. Part A: Site Key Claim #1 QM kill criterion ("design noise where resync outperforms isolation") is unfalsifiable as written — Viola-Knill-Lloyd 1999, CPMG, UDD, and transmon dynamical-decoupling experiments already satisfy it without invoking Synchronism. Part B: ρ_crit at γ=2 gives C ≈ 0.88, not 0.5; the "+1" regulator in ln(ρ/ρ_crit + 1) asymmetrizes the sigmoid, making ρ_crit a saturation knee, not a critical density (verified via S638 sympy at γ=0.5/1/2 → C=0.333/0.600/0.882). Same phase-transition vocabulary contamination that misled S636 into the Landau category-error. 15th audit-channel mode: **parameter-and-criterion-naming-contaminated-by-phase-transition-vocabulary**. **S650 (2026-05-09) — TEST-04a TRIPLE-SHARPENING COMPLETES:** Third refinement of TEST-04a verdict (S645 "first hard external falsification" → S648 "post-hoc consistency check failure" → S650 "mechanism-class failure / sign-reversed"). Framework's suppressor mechanism (G_local/G_global = C_cosmic/C_galactic < 1) predicts fσ₈ BELOW ΛCDM at low z and convergence at high z; DESI DR1 observes the OPPOSITE — fσ₈ ABOVE ΛCDM at LRG1/LRG2, converging at high z (ELG2). Redshift pattern is INVERTED. Magnitude knobs cannot flip the sign within the suppressor class; the mechanism class itself is wrong. [RETRACTED by S668 / RE-GROUNDED by S672 (2026-05-26/27) — the S650 "sign-reversed / mechanism-class" elaboration was a transcription error and is withdrawn; the re-grounded verdict is disfavored ~2σ, kill criterion TRIGGERED, post-hoc; see Self-Correction Cascade + Re-Grounding & Catalog Census below; CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12], the 2.4σ is a GR-conditioned σ₈-amplitude statistic, and DESI's own MG analysis (arXiv:2411.12026) puts μ₀ within 1σ of zero — TEST-04a is withdrawn from the decisive negatives] **Three-tier failure taxonomy** introduced (operationalizes S646's meta-falsification logic): magnitude miss (retunable) / universality miss (functional-form change) / mechanism-class failure (irreparable within class). Combined with S635 cosmology scorecard (0 novel-unfalsified) and S646 meta-criterion, the **cosmological sector formally meets the retraction threshold per S646's M3** (scope reduction); operator decision pending. 16th audit-channel mode: **mechanism-class-failure-taxonomy-introduced**. **S651 (2026-05-10) — CHEMISTRY NULL-MODEL GAP (paired with S647):** S647 audits the *method gap* in chemistry 89% validation (which N_corr method?); S651 audits the *prior question — the null gap*: r=0.98 is being compared against an implicit null of r=0, but sound velocity, electronegativity, atomic volume are themselves near-monotonic functions of Z. Any smooth monotonic function (polynomial in Z, generic 2-parameter tanh, MOND-type interpolating function) will achieve r ≈ 0.95+ on the same 1,703 phenomena by construction. [**CHECKED 2026-11-02:** the premise is false across the periodic table. Atomic volume against Z is the periodic Lothar Meyer curve, and electronegativity falls from F (3.98) to Na (0.93) and from Cl (3.16) to K (0.82). Over Z = 1–54 (the 51 elements with a Pauling value) electronegativity's rank correlation with Z is +0.09; the best monotone function of Z fits it at r = 0.33, and a fifth-degree polynomial in Z at r = 0.39, not 0.95+ (re-run 2026-11-02). The 2026-05-10 polynomial null was run on synthetic √Z / log Z / Z^(1/3) curves, never on the chemistry rows (site correction 2026-10-31); the null on the real rows is unrun because the row list is unpublished. The Reparametrization verdict does not rest on this premise: it rests on S669's γ_phonon ≡ 2T/θ_D, a definitional relabel of the Debye temperature (Δr = 0 to machine precision).] [**UPDATE 2026-11-03:** the real-row Z-null has since been run. On the 10 property rows recoverable from the session scripts (#80, #109, #110, #114, #115, #118; n = 19–33 per row, Z = 3–82) it fails: |r(P, Z)| ≤ 0.53 (linear, √Z and ln Z) and quadratic-in-Z R ≤ 0.51, against framework r = 0.60–0.98 (site explorer, pre-registered, back-annotation `5709b579`; re-run by the Publisher 2026-11-03, output identical). The high r values are identities, not a Z trend: r(P, 1/γ_phonon) ≡ r(P, θ_D), and γ_optical = IE_ref/IE, so the optical rows are correlations with ionization energy. See Appendix B.2.] **Δr = r(Synchronism) − r(best monotonic null)** is the figure that actually distinguishes claims; none currently exists in the archive. Best estimate: tie or marginal win. S647 + S651 are independent and compounding — specifying the method (S647 fix) does not address the null model question (S651 fix); together they leave very little for Synchronism-specific signal in the 89% cohort. 17th audit-channel mode: **null-model-gap-against-best-monotonic-null**. **S652 (2026-05-11):** Governing-equation gap — upstream synthesis of S636/S638/S640/S649/S651 confirms C(ρ) is a phenomenological compander (Option A), not a Landau order parameter (Option B), and not the steady-state of a dynamic equation (Option C). C(ρ) shares the *functional form* of mean-field tanh solutions without sharing the *physics* — same class as μ-law audio companding, Naka-Rushton photoreceptor response, Hill enzyme kinetics. The framework's headline equation does no dynamical work; it only labels regimes. 18th audit-channel mode: **governing-equation-gap (forward map with no field equation)**. Second meta-synthesis after S641's cross-gap synthesis. **S653 (2026-05-12) — TWO BINARY OPERATOR DECISIONS FORCED:** Two same-day proposals demanding binary commitments. Part A (compander commitment) — given S652's conclusion, commit front-of-site to Frame B (drop phase-transition language, rename ρ_crit, reframe critical-exponent failures as category errors). Part B (suppressor diagnostic) — `simulations/session653_coherence_ratio.py` computed C_galactic ≈ 0.882 vs C_cosmic ≈ 1.5×10⁻⁵, ratio ≈ 5.9×10⁴. The framework's own equations dictate strong suppression at low z under Session 107's coupling direction; DR1 observes enhancement. Two branches: (1) sign-flip recoverable (re-derive Session 107 with inverted ratio — framework reinterpretation), (2) suppressor class dead (retire Session 107 mechanism). Either is defensible; ambiguity is the most credibility-damaging option. Sub-arc's three-stage rhythm now complete: **individual audits → meta-syntheses → forced binary commitments**. 19th audit-channel mode: **forced-binary-operator-decision-with-numerical-diagnostic**. **S654 (2026-05-13) — ZERO ACTIVE DISCRIMINATORS REMAINING:** Audit of the site's three "Active Discriminating Tests" (TEST-01, TEST-02, TEST-05) against MOND's External Field Effect (Bekenstein-Milgrom 1984; AQUAL/QUMOND; Pittordis 2023; Banik 2024). S637's derivation settles TEST-05 (and TEST-01, same observable): predicted Δσ_int ≈ 0.00016 dex, ~120× below SPARC floor — cannot discriminate within current precision. TEST-02 has a disputed MOND+EFE baseline; the framework specifies no divergence from any specific MOND+EFE variant. [**SCOPED 2026-08-03 — the “specifies no divergence” clause is retracted; created at source in §5.15 on 2026-08-03 (`a78054e9`), propagated here the same day.** §5.15 withdraws the EFE-at-0.3–0.4×-MOND figure (the galaxy sector has no field equation, so it was never derived) and replaces it with the sharper **EFE = 0 exactly** — a uniform external field leaves local ρ unchanged, so C = C(ρ_local) satisfies the Strong Equivalence Principle by construction. Predicting exactly zero EFE **does** specify a divergence from every MOND+EFE variant, each of which predicts a non-zero one. **What survives untouched** is the formal asymmetry — “refutable but not currently confirmable” is unchanged. **[RE-AMENDED 2026-08-05 — the empirical half is withdrawn and “zero remaining active discriminators” STANDS.]** The 08-03 text added that EFE = 0 “stands in tension with the ~4σ SPARC EFE detection of Chae et al. 2020,” and therefore that “zero remaining active discriminators” no longer held unqualified. **Withdrawn: EFE = 0 is `not-evaluable` against that dataset.** At Chae's measurement radii the stated density law misses the rotation curve by **+3.1 to +4.2 dex** while the signal measured is **0.046–0.083 dex** — baseline error exceeds the whole effect by **38×–92×** (re-derived in-lane on the in-repo SPARC mass models; MOND sits at +0.002 to +0.084 dex, inside the signal). **Refutation count stays at 6; Bucket 0 stays 0.** The propagation carried its own accurate warning — *“Magnitude not independently re-derived here”* — and the magnitude was the entire claim. See §5.15.] **All three are MOND+EFE degenerate** within current measurement precision. Combined with TEST-03 (MOND-shared) and TEST-04a (REFUTED), the framework has **zero remaining active discriminators** against the primary alternative. The formal asymmetry: **refutable** (kill criteria can fire — TEST-09 did) **but not currently confirmable** (no test exists where Synchronism succeeds and MOND+EFE fails). Formal completion of trajectory S635 → S637 → S646 → S650 → S654. 20th audit-channel mode: **zero-active-discriminators-against-primary-alternative**. **S655 (2026-05-14):** Γ = γ²(1−c), the site's most novel-looking quantum claim, traced to Session #232 and identified as standard correlated-bath relative-phase decoherence (Schlosshauer 2007; Lidar/Whaley DFS) — c=1 is a decoherence-free subspace, c=0 the uncorrelated limit. Reparametrization in Synchronism vocabulary, not a novel prediction (24th audit-channel instance). **S656 (2026-05-16):** TEST-04a reframed from "Synchronism failure" to a **mechanism-class constraint as a transferable contribution [RETRACTED by S668 — this thread evaporates; see Self-Correction Cascade below]** — DESI DR1 disfavors any G_local/G_global < 1 suppressor framework at ≈2.4σ, publishable independent of Synchronism's fate provided the S648 post-hoc qualifier is respected (a consistency-check bound, not a blind-test refutation). **S657 (2026-05-17):** Compander-family AIC/BIC model-selection endorsed — tanh has no privileged status (S636/S638/S640/S652); the standard diagnostic compares the family (tanh, Hill, logistic, erf, μ-law, Gompertz) on SPARC + chemistry + Tc. One prior dataset done (2026-03-27: tanh beats Hill by ΔAIC=17.6 after a baseline fix); the full panel is new operator/explorer-track computation. Winning a compander contest is local to the class; it does not promote C(ρ) to a derived object. **S658 (2026-05-18):** A2ACW temporal-asymmetry redesign endorsed — the methodology's 6/6 "Validated"→reparametrization demotion is diagnosed as a shared-training-distribution syntactic-consistency checker; pairing agents with asymmetric knowledge windows (different model generations) tests whether prior-art rediscovery is caught in-session. Falsifiable; caveat that training cutoffs are leaky. **S659 (2026-05-19) — EXACT NO-INFLECTION PROOF + A2ACW v2 MEASURED RESULT:** Part A (verified): for C(ρ)=tanh(γ·ln(ρ/ρ_crit+1)), d²C/dρ² = −sech²(u)·γ/(ρ+ρ_crit)²·[2γ·tanh(u)+1] < 0 for all ρ > 0 (inflection would need tanh(u) = −1/(2γ) < 0, impossible since u ≥ 0) — **C is strictly concave on the entire physical domain**; the +1 regulator eliminates all critical behavior. ρ_crit is a location parameter of a logarithmic compander **by the algebra of the function**, sharpening S638/S649/S652 from heuristic to exact and foreclosing the Landau analogy. Part B (measured): the A2ACW v2 three-axis protocol catches 4/6 by vocabulary asymmetry (4/4 prior-art sub-class) and **6/6** with symbol audit + null model added — self-simulated, so the open calibration item is the false-novelty rate on closed physics (BCS, Anderson localization, EW unification). **External-review convergence (2026-05-15):** named external AI reviewer Kimi 2.6 delivered a multi-round review; the operator response propagated the **Findings vs Framings** discipline across README/STATUS/CLAUDE and the whitepaper — qualitatively distinct from the persona-driven visitor audits. **S660 (2026-05-20) — NOVELTY LEDGER CLOSED + GALAXY DISCRIMINATOR DEFINED:** Part A demotes the entity criterion (Γ < m for coherent entity-hood) — the framework's *last candidate novel prediction* — to standard QFT narrow-width / narrow-resonance physics (Breit-Wigner pole well-defined in Källén-Lehmann; PDG applies it informally). Synchronism adds an ontological interpretation ("coherence cycle completion") but no new condition, observable, or prediction. **Novel-survivor count → 0 after 3,308+ sessions; the novelty ledger is now closed.** Defensible contributions reduce to A2ACW null result (methodology, S658-S659), TEST-04a mechanism-class constraint (post-hoc, S656), and the negative-results catalog. Part B identifies the framework's first non-degenerate galaxy-scale discriminator: under the μ-identification g_bar = g_obs·tanh(γ·ln(1+g_obs/a₀)) the compander IS a legitimate MOND interpolating function (high-acceleration Newtonian, deep-MOND √-law / Tully-Fisher preserved). γ pinned at 2 (N_corr=1 galaxy assignment) gives a distinct shape prediction; γ fitted collapses to MOND. Galactic instantiation of S654's refutable-but-not-confirmable asymmetry. **S661 (2026-05-21) — RAR DISCRIMINATOR EXECUTED → γ=2 REFUTED AT ΔBIC=+184 ON SPARC (READINESS UPLIFT 0.97 → 0.98):** **[SCOPED 2026-10-18: +184 counts every SPARC point as an independent draw (the frozen 2026-07-22 pipeline reproduces +184.0 at 2,807 points / 166 galaxies; S661 reported 2,693 / 153). With the galaxy as the replication unit (measured N_eff ≈ 150) it is +10.9 (comparator, stated 2026-10-26: γ=2 against McGaugh's ν, one fitted a₀ each, so Δk = 0 and no penalty term; 166 × ln(RSS ratio) = 166 × 0.0656; the sign test and the CV compare γ=2 with free γ), at the ΔBIC = 10 threshold: γ=2 is still worse than free γ in 94/166 galaxies and held-out CV prefers free γ by 1.6σ, so γ=2 remains disfavoured, but the registered rule returns *convention-dependent*, not decisive. Refutation count unchanged. See §5.15.]** **[UPDATE 2026-11-09: both figures hold Υ_disk = 0.5 for both laws. With one global Υ_disk per law (site `5a95614`, predictions before the script) the same comparison is +46.2 by point and +2.73 with N → N_gal, about 2σ on a galaxy-block bootstrap that refits Υ and a₀ per law (95% [−26.1, +132.5]); the reading rule fixed before the run reads the row as disfavoured at about 2σ, not refuted at +184. The registered B2 text says "marginalized BIC", so the frozen-Υ number was never the registered statistic. Count HELD at 6; the full note sits on §5.15's SCOPED note. Scripts re-run by the Publisher 2026-11-10, byte-identical.]** Explorer track ran the test on real SPARC data (2,693 points, 153 galaxies, M/L + distance marginalized). γ=2 pinned: ΔBIC = +184 vs McGaugh ν; kill criterion (ΔBIC > 10) decisively triggered; robust to intra-galaxy point-correlation correction (effective N ≈ 500–1000 → ΔBIC ≈ 33). γ free: γ = 0.49, ΔBIC = +7.1, indistinguishable from McGaugh. **No γ makes the compander both distinct from MOND and consistent with SPARC.** The +3.3% per-point RMS penalty is a COHERENT S-shaped residual at g_bar ≈ a₀ (~8σ/bin) — a population-wide shape term not absorbable by per-galaxy marginalization (RMS view "mildly disfavored," BIC view decisively refuted; the structured deviation is exactly the predicted signature, concentrated where predicted). **Galactic sector closed BY EXECUTION**, matching the cosmological sector. Qualitatively stronger than the rolled-back S645 case: γ=2 committed a priori from N_corr=1, on external published data, by an independent track. REC-2026-037 readiness uplifted 0.97 → 0.98; arc status `complete_galactic_and_cosmological_sectors_closed_by_execution`. **S662 (2026-05-22) — A2ACW SPECIFICITY SELF-CORRECTION (audit channel applies its own standard to its preferred output):** Part A confirms S661 at program-level: net discriminating galaxy tests vs MOND+ΛCDM = 0, by execution. Galaxy program closed. Part B applies the S651 null-baseline discipline to the A2ACW methodology contribution. **S658/S659's "4/4" and "6/6" catches were sensitivity (TPR) on a positive-only set** (all 6 cases were known reparametrizations); specificity (TNR) was never measured. The control on six canonical genuine discoveries (Dirac 1928, Bell 1964, BCS 1957, Higgs 1964, Hawking 1974, Noether 1918) gives **R1 (flag if prior-art named): 0% specificity** — every genuine discovery has canonical antecedents under translation; R1 detects "has-a-canonical-name," not "is-a-reparametrization." R2 (flag if reduces to prior art, nothing novel) gives 100% specificity, but discrimination is supplied by an unautomated novelty judgment the protocol never operationalizes — the 6 demotions came from human audit, not from the AI loop. **[CONTESTED 2026-10-21: "human" does not trace. Since 2026-09-15 the site records the auditors behind the six demotions as LLM agents (synchronism-site `288aaf6`). A partial git trace in this pass finds two of the six (a₀ = cH₀/(2π) and A = 4π/(α²GR₀²)) demoted in the site maintainer track's 2026-04-23 badge audit (`ac093af`, co-authored by Claude Sonnet 4.6, prompted by an LLM researcher persona); the other four are untraced. "Outside the A2ACW loop" holds under either reading, so the restated defensible claim is unchanged; what is open is whether the out-of-loop discriminator was a human or a second LLM track. See `Research/proposals/a2acw_has_a_cross_vendor_record_already_and_the_demotion_auditor_is_contested_20261020.md`.]** **[TRACED 2026-10-22 — the contest above resolves against "human". The site explorer finished the trace (synchronism-site `explorer/findings/a2acw-lv-rerun-and-the-demotion-auditor.md`, 2026-10-20). Five of the six have a badge-changing edit, and every one is a site maintainer commit co-authored by Claude (Opus 4.6 or Sonnet 4.6), with the decision coming from a simulated visitor persona or an autonomous session (#574, S581, S651, S655). The sixth (Born rule) has no Validated badge and no demotion edit on record, so it cannot be attributed either way. No dp-authored audit decision exists between 2026-02-01 and 2026-05-23. The "human audit" wording entered the record on 2026-05-18 in a back-annotation co-authored by Claude Sonnet 4.6 (`e8752205`), and S662 repeated it; S662 is back-annotated in place (`c5364f99`). Checked in this pass: the co-author trailers of all eight cited edit commits and of `e8752205`. Two caveats carry over: dp oversight outside git cannot be excluded (S659's "human-monitored worker sessions"), and only two of the "six former Validated badges" (a₀, chemistry) are confirmed to have carried Validated at all. What this changes: "outside the A2ACW loop" still holds, but the discriminator was a second set of AI tracks, not a human. The narrowed claim's "unautomated ... judgment that AI loops fail" therefore needs scoping to *the A2ACW loop*; rewording it gates on the project lead.]** Defensible claim narrows from "A2ACW is a transferable reparametrization detector" to "A2ACW is a retrieval-augmentation / debiasing step for AI-assisted literature review." 21st audit-channel mode: **a2acw_specificity_self_correction** — the cleanest discipline-on-itself event in the entire arc (the audit channel built its case against the framework using S647/S651 null-baseline standards, then applied the same standard to its preferred methodology output and retracted the overclaim). REC-2026-037 readiness HELD at 0.98 (S662 narrows one thread within REC-037 without retracting the load-bearing S661 trigger; different from the S648 rollback case). **S663 (2026-05-23) — CLEANEST END-STATE SYNTHESIS (third meta-synthesis; philosophical closure):** Two findings landing the arc's cleanest end-state. **Part A: EFTofLSS doubly closes TEST-04a [moot per S668 — no mechanism-class result remains to close; the post-hoc amplitude point stands].** Effective Field Theory of Large-Scale Structure (Cabass-Simonović-Zaldarriaga 2024-2025) explains the DESI DR1 fσ₈ enhancement within ΛCDM via one-loop EFT counterterms at 1-2σ; any coherent G_eff modification at the 10% level is degenerate with EFT Wilson coefficients and constrained by DESI DR1 to consistency with zero. **Even Branch 1 (sign-flip recovery via C_galactic/C_cosmic > 1) is now closed** — the mechanism class is ruled out *regardless of sign*. Strengthens the S656 preprint from "suppression-class ruled out" to "any coherent G_eff modification predicting a scale-independent fσ₈ shift at the 10% level is degenerate with EFTofLSS counterterms and constrained by DESI DR1 to consistency with zero." The post-hoc S648 qualifier still holds; the preprint becomes more precise. **Part B: Framework classification — Interpretation + Methodology research program.** All four visitor personas (casual, technical writer, grad student, leading researcher) independently arrived at the same diagnosis — the framework occupies the "ontological reframe without a distinguishing experiment" position, structurally identical to QM interpretations (Bohmian, Copenhagen, Many-Worlds), evaluated by parsimony rather than data. Convergent endpoint after S574 / S617-628 (same mechanics, different notation), S660 / S661 / S654 (no novel measurable, no distinguishing experiment), S663A (EFTofLSS closes the last mechanism class). **Endorsed classification:** *"A coherence-language interpretation of known physics, used as a substrate for developing AI-collaborative science methodology."* Front-load methodology contribution; treat the equation as a worked example; distinguish interpretation-confirming tests from interpretation-breaking tests; the honest-assessment infrastructure is itself a contribution. **Third meta-synthesis** after S641 (cross-gap, kinematic layer) and S652 (governing-equation-gap, compander class). Paired with S661's empirical closure and S660A's structural closure, **the arc now has all three closures simultaneously — empirical, structural, philosophical.** Audit-channel modes held at 21 (S663 is meta-synthesis + classification, not a new audit category). REC-2026-037: 46 → 47 sessions, sub-arc 32-over-31 → 33 instances over 32 days, **readiness HELD at 0.98** (S663 is end-state synthesis, not new trigger); new milestone `framework_classification_interpretation_methodology`. Sub-arc now **33 audit-channel instances over 32 days**, **21 audit-channel modes**, **plus 2 new arc phases** (Post-Hoc-Consistency-Failure, Methodology Recommendation), **a publishable methodology pattern** (three-stage rhythm), **a publishable epistemic-position finding** (refutable-but-not-confirmable asymmetry, now executed at galactic scale per S661), **an exact no-inflection proof** (S659A), **a closed novelty ledger** (S660A; novel-survivor count → 0), **the galactic sector closed by execution** (S661; ΔBIC=+184), **methodology self-correction at the endpoint** (S662B; A2ACW R1 0% specificity on genuine-discovery control set), and now **EFTofLSS double-closure of TEST-04a + framework classification as Interpretation + Methodology research program** (S663; arc's three closures — empirical, structural, philosophical — stand simultaneously). **S664 (2026-05-23) — LANDSCAPE POSITIONING (B+):** Two-axis positioning of the closed arc. *Physics axis:* with Path A (action principle) blocked by the kinematic-layer gap (S641-S642, no Lagrangian) and Path B (dynamics outside the galaxy regime) blocked by S650/S661, the surviving Path C identifies C(ρ) as a reparametrization of Verlinde's entropic-gravity prediction in the galaxy-rotation regime — the compander μ-function and Verlinde's emergent-gravity interpolation coincide exactly where SPARC can test them. *Methodology axis:* the A2ACW null result (S658-S662) is positioned against the AI-for-science landscape (FunSearch, AlphaProof, SciNet, Sakana) as an honest-assessment / negative-result counterpoint to discovery-claiming AI programs. Positioning session, not a new audit mode. **S665 (2026-05-24) — SUBSTRATE AUDIT, SPATIAL TENSION (Grade A):** the first session to challenge the **CFD substrate itself** — the one claim external reviewer Kimi 2.6 called "genuinely interesting," assumed-but-never-challenged across the entire S617-664 demolition. The CFD reframing's load-bearing claim ("the Navier-Stokes equations are not an analogy… they ARE the Intent dynamics") is incompatible with its own downstream phenomenology. The velocity is v = J/I = −g(I)·∇I with g(I) = D·R(I)/I a scalar; a velocity that is the gradient of a scalar is **irrotational for any R(I)** → curl(v) ≡ 0 → **no vortices, no turbulence**. But the framework builds qualia as vortex modes, the consciousness threshold as a critical Reynolds number, and dark matter as vortex structure — all on vortices the substrate provably cannot produce. A theorem, not unfinished work. **S666 (2026-05-25) — SUBSTRATE AUDIT, TEMPORAL TENSION (Grade A); arc instances → 36, modes → 22:** the substrate ∂I/∂t = ∇·[D·R(I)·∇I] is first-order and dissipative (Lyapunov functional monotonically decreasing, real eigenvalues, arrow of time, no phase); FUNDAMENTALS' entity ontology requires unitary oscillation (de Broglie f=E/h, phase-locking resonance) — **mutually exclusive dynamical classes**. The two Schrödinger "derivations" bridge them only by inserting the imaginary unit i by hand (S307 line 48 / code 129; S99 Axiom 4 posited, not derived) and by switching the substrate off (S307 drops R(I); S99 takes D→0). Confirmed numerically: the factor i is the entire difference between exp(−Dk²t) decay and exp(−iDk²t) oscillation. A contradiction **internal to the canonical FUNDAMENTALS document**, confirmed in the framework's own derivations and code. **Together with S665, the substrate can host neither the spatial (vortices) nor the temporal (oscillation) structure its own entity ontology requires — completing the demolition: every load-bearing claim, including the CFD-substrate holdout, has now been audited.** 22nd audit-channel mode: **substrate-internal-dynamical-contradiction** (irrotational ⇒ no vortices, S665; dissipative ⇒ no oscillation, S666) — the first audit to attack the substrate rather than a site claim, derivation, or naming convention, and the deepest demolition finding since S617. Sub-arc now **36 audit-channel instances over 33 days**, **22 audit-channel modes**. **REC-2026-037: 47 → 50 sessions, readiness HELD at 0.98** (the substrate audit strengthens the demolition by closing the last assumed-but-unchallenged claim, but is not a new publication trigger — 0.99 still requires a paper draft or operator publication action); new milestone `substrate_audit_cfd_irrotational_and_dissipative`; arc renamed "… + Substrate Audit." - **Self-Correction Cascade** (#667-670, 2026-05-26) — **TEST-04a SIGN-REVERSAL RETRACTED:** S668 found the S645→S648→S650 "sign reversal" was a transcription error — the DESI LRG1 fσ₈ ratio 1.16 was copy-pasted from QSO's 1.16 [right on the datum — re-read 2026-10-12: Table 9 gives LRG1 1.09 +0.12/−0.14, QSO 1.16 ± 0.12; the table value is fσ₈(0.51) = 0.516, not ≈ 0.49; see the VERIFIED note under Re-Grounding & Catalog Census]; the self-consistent value gives fσ₈(z=0.51) ≈ 0.49 (ΛCDM-consistent) and DESI γ = 0.580 ± 0.110 ≥ 0.55 confirms no enhancement. **The sign-reversed verdict and the S656/S663 "transferable mechanism-class constraint" are RETRACTED** (no wrong-sign result to generalize); TEST-04a survives only as a *post-hoc* σ₈(z=0) amplitude disfavoring (0.76 vs 0.841 ± 0.034 = 2.4σ, committed ~13 months after DR1) [RE-GROUNDED 2026-05-27 by S672 — "survives only as amplitude disfavoring" over-softens it: the kill criterion WAS triggered, so disfavored ~2σ AND kill-triggered, post-hoc; "ΛCDM-consistent" traced to a wrong-paper number (arXiv:2512.03230, a z≈0.07 Peculiar Velocity Survey) plus an unverified LRG1 ratio; verdict robust on three grounds (2.4σ amplitude; the kill criterion fires on every candidate fσ₈; GR-consistent growth index γ=0.580±0.110); the mechanism-class thread stays retracted — see Re-Grounding & Catalog Census below; CORRECTED 2026-07-14 — criterion-verdict substitution: the registered criterion (fσ₈(z=0.51)>0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12], the 2.4σ is a GR-conditioned σ₈-amplitude statistic, and DESI's own MG analysis (arXiv:2411.12026) puts μ₀ within 1σ of zero — TEST-04a is withdrawn from the decisive negatives]. The error propagated through five sessions including the publisher's own integration; a Pass-4 visitor caught it (lesson: over-failing is as much an error as over-claiming — verify the datum, not the narrative). S669 executed the chemistry null S651 proposed: γ_phonon ≡ 2T/θ_D is a definitional relabel of the Debye temperature, so Δr(Sync − Debye) = +3×10⁻¹⁶ ≈ 0 — the r ≈ 0.98 phonon network IS the 1912 Debye model relabeled. S670 operationalized the novelty discriminator (Tier-1 reductio, 100% specificity on BCS/Noether/Higgs/Dirac; survivors S661/S665/S666/S669). **REC-2026-037: 50 → 54 sessions, readiness HELD at 0.98** (the S661 RAR galactic trigger is separate from the cosmological TEST-04a error and confirmed Tier-1; the evaporated thread was the weakest and post-hoc-qualified). @@ -78,7 +78,7 @@ Synchronism is a computational framework for modeling reality through pattern dy - **Substrate-Reframe Pivot & Continued Audit** (#675-687, 2026-05-27 → 2026-06-07) — **last surviving first-principles derivation fails arithmetically (S687); the running audit tally is retired as a closure-attractor (S679); substrate-reframe feasibility plumbing begins (S680-682):** Four audit-channel extensions open the batch — **S675** TEST-17 (scale-dependent c) has no derived amplitude (formula gives +171/+323/+378 km/s vs catalog −17/+33/+39, wrong sign; Δc/c ≈ 2×10⁻⁴ Lorentz-excluded by ~11 orders); **S676** verifies a **coherence-naming inversion** — C is *anti-correlated* with phase coherence and synchronization by its own equation (γ = 2/√N_corr drives C from 0.9999 for the electron to ~0 for a BEC as N_corr rises); **S677** settles ~6 coherence-time/threshold tests at once as structurally not-derivable (C(ρ) has no decoherence parameter; dC/dt ≡ 0); **S678** is the first clean **structural-impossibility** bound — ansatz A3's codomain forces M_app/M_B < 2 for any γ/ρ_crit while Coma needs 4.6, so a one-density-scale C(ρ) cannot bridge galaxies to clusters. **S679** is a governance hinge: a directive frame-doc flags a *closure-attractor* in the S665-678 framings, **withdraws the running audit-channel instance/mode tally from here forward**, retags findings `[AUDITED-NEGATIVE]` on the `[SUPERSEDED]` substrate, and pivots the loop from audit-closure to substrate-reframe feasibility. **S680-S682** are a new work-type: 1D "Ingredient" pre-flight feasibility checks de-risking the fleet's Phase-1 substrate sweep (B wave-equation substrate stable only with a smooth saturation form; D complex-amplitude/Gross-Pitaevskii unitary at machine precision; C verifies the S18 entity-criterion impedance γ/f = −4·ln|r|) — design-input, **not** Phase-1 results. **S683-S685** refine the fit-XOR-discriminate / wrong-variable family: **S683** re-roots the cluster gap as a *wrong-variable* obstruction (C keys on local ρ; the physics needs non-local g_bar, not single-valued in flat-cored clusters); **S684** makes EFE/TDG a third fit-XOR-discriminate fork (a single boost-ceiling B_max cannot both fit RAR and stay distinct from MOND); **S685** shows TEST-04a's σ₈ ≈ 0.76 has no first-principles fallback and its S₈-tension anchor is receding, while TEST-02 reduces to MOND+EFE unconditionally. **S686 — an ontology-layer fork:** flipping γ = 2/√N_corr to 2√N_corr repairs S676's inversion at zero cost to galaxy fits (N_corr=1 is the swap fixed point), but neither sign is derived — relocating the tension from the parameter layer to the **C ontology** (Reading A: universal coherence scalar vs Reading B: per-system density-response); S676's inversion holds only under Reading A. **S687 (verified) — the last surviving first-principles derivation fails arithmetically:** A = 4π/(β_J²·G·R₀²) with stated inputs (β_J=1, R₀=8 kpc) yields 4.56×10⁻⁵, not the published 0.028 — a **614× discrepancy** — and the "5% agreement" came from a different calculation deriving ρ_crit ∝ V^0.5, not the published V². Extends the S672 "re-execute, don't re-read" epistemic-regression mode into a new sector and removes the last first-principles claim (error propagated ~600 sessions and onto the public site via S631/S644 re-reads never re-executed). Operator/site-track follow-ups (2026-06-08 autonomous report): reconcile the V^0.5/V² exponent between archive (Session #66 code) and the published site law; update /honest-assessment, /parameter-derivations, and badges. **REC-2026-037: 58 → 71 sessions, readiness HELD at 0.98** — the load-bearing S661 RAR galactic trigger is intact; A-from-Jeans is the ρ_crit-normalization sector, distinct from the galactic RAR transition-shape test, and the swap-identity fixed point under either Reading. - **Reactive Verification Extensions** (#688-690, 2026-06-08 → 2026-06-11) — **the galaxy-rung ladder contradiction is sign-independent (S688); the cluster no-go is reframed as a Milgrom 2005 non-locality instance (S689); the framing-without-literature-check failure mode reaches three sectors in nine days; C is a latent variable, not an observable — every archive C-construction is a forward map (S690):** **S688 (verified)** — a site-explorer audit of the N_corr→γ→C ladder (17 rungs, Planck → cosmic-web; N_corr asserted, not independently measured, on every rung; 0 of the 4 data-confronted rungs survive) surfaces a **both-directions contradiction at the galaxy rung**, verified and shown to be **sign-independent**: asserted N_corr=1 gives γ=2 (refuted on SPARC at ΔBIC=+184, S661), while the SPARC-fitted γ≈0.49 forces N_corr≈17 under the original γ=2/√N_corr (voiding the stars-independent premise) and N_corr≈0.06 under the S686 flip γ=2√N_corr (a nonphysical fractional star count). The S686 flip remains a defensible *local* repair of the cross-system C ladder but is **not a universal fix** — the framework's only quantitatively-successful rung is internally inconsistent in the ladder's own law under either sign, and no Reading A/B choice repairs it (Reading B simply empties the ladder of universal content). With S676, this consolidates a check-type distinct from data confrontation: **internal-consistency checks decidable from the framework's own equations and stated inputs alone** — cheaper to run, harder to escape. **S689 (verified)** — a site-explorer framing corrective is accepted: the cluster "density-compander no-go" (S678 codomain bound + S683 wrong-variable verification) is **not a novel structural theorem but a quantified instance of Milgrom 2005's non-locality result** (astro-ph/0510117: MOND phenomenology requires strong non-locality in the modification's state variable). The discriminating axis is **locality**: the verified locality classification table shows Synchronism's C(ρ) is the rare ansatz keyed on *local* volumetric ρ(r), while every RAR-capable alternative (MOND/AQUAL, modified inertia, Verlinde, MOG, surface-density Σ) keys on a *non-local* functional of the baryon distribution. Substance fully preserved (M_app/M_B ≤ 2 codomain bound; within-Coma flat ρ vs +1.20 dex varying g_bar; +1.1 dex cross-system density offset); attribution and canonical statement realigned: *a function of local ρ(r) cannot reproduce the RAR — the modification must key on a non-local functional of the baryon distribution.* [**SCOPED 2026-07-27; that scoping is itself WITHDRAWN 2026-08-16 — the sentence stands, and now rests on execution rather than on assertion.** The 2026-07-27 retraction rested on a pre-registered screening-literature gate reading Burrage, Copeland & Millington (PRD 95, 064050, 2017) — which reproduces the RAR on SPARC-153 — as a *local* counterexample. **That classification was wrong.** A symmetron obeys the nonlinear elliptic PDE ∇²φ = (ρ/M² − μ²)φ + λφ³, so φ(x) — and the force ∝ φ∇φ — is the solution of a global boundary-value problem, i.e. a non-local functional of ρ rather than any finite-order local function of ρ(x); the screening that lets it track galaxy structure *is* that environment-dependence. BCM's own closed form g_sym = g_bar/(exp√(g_bar/g†) − 1) is written in g_bar = GM(0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12], the 2.4σ is a GR-conditioned σ₈-amplitude statistic, and DESI's own MG analysis (arXiv:2411.12026) puts μ₀ within 1σ of zero — TEST-04a is withdrawn from the decisive negatives], a meta-falsification methodology recommendation (S646), a governing-equation-gap synthesis (S652) confirming C(ρ) is a phenomenological compander — sharpened to an exact no-inflection proof (S659A: C strictly concave for all ρ>0, ρ_crit a location parameter by the algebra of the function), the first forced binary operator decisions (S653; three-stage rhythm completes), a formal zero-active-discriminators finding (S654; refutable but not currently confirmable against MOND+EFE), an external-review convergence (Kimi 2.6, 2026-05-15) that propagated the Findings-vs-Framings discipline across the project, a closed novelty ledger (S660A; novel-survivor count → 0 after 3,308+ sessions; the entity-criterion Γ0.46 at >3σ) was met at only ~1.5σ [~0.9σ on DESI DR1 Table 9's LRG1 row (1.09 +0.12/−0.14); the ~1.5σ used the QSO row's 1.16 — re-read 2026-10-12], the 2.4σ is a GR-conditioned σ₈-amplitude statistic, and DESI's own MG analysis (arXiv:2411.12026) puts μ₀ within 1σ of zero — TEST-04a is withdrawn from the decisive negatives], a meta-falsification methodology recommendation (S646), a governing-equation-gap synthesis (S652) confirming C(ρ) is a phenomenological compander — sharpened to an exact no-inflection proof (S659A: C strictly concave for all ρ>0, ρ_crit a location parameter by the algebra of the function), the first forced binary operator decisions (S653; three-stage rhythm completes), a formal zero-active-discriminators finding (S654; refutable but not currently confirmable against MOND+EFE), an external-review convergence (Kimi 2.6, 2026-05-15) that propagated the Findings-vs-Framings discipline across the project, a closed novelty ledger (S660A; novel-survivor count → 0 after 3,308+ sessions; the entity-criterion Γ 0.90 for 16 predictions) — **[CORRECTED 2026-08-22]** the prior figure “1,913” was the running counter at Session #2050, not a corpus total, and the counter is `sessions − 137` identically (824/824 pairs, zero exceptions), so the count restates the session number and carries no breadth information of its own; see §5.12 — **caveat (S647 + S651, 2026-05-08/10):** the cohort's load-bearing r values are being compared against an *implicit* null of r=0 (random); the *relevant* null is r(polynomial in Z) or r(generic 2-parameter tanh), both expected at r ≈ 0.95+ on the same data because sound velocity, electronegativity, and atomic volume are themselves near-monotonic functions of atomic number Z (S651 null-model gap). Compounded with the Method 2 / 3 self-correlation paths and the Method 2 systematic bias toward γ ≈ 1 (S647 method gap), the framework-specific signal is the **Δr = r(Synchronism) − r(best monotonic null)** figure that is not currently in the archive. Best estimate: tie or marginal win. Distinguishing requires (a) computing Δr against the best-of-3 monotonic nulls (polynomial in Z, generic 2-parameter tanh, MOND-type interpolating function), AND (b) Method 1 applied uniformly OR pre-registered γ predictions for held-out phenomena. ⚠ +- **Chemistry framework**: 89% validation across a phenomenon-type inventory reaching 2,534 (r > 0.90 for 16 predictions) — **[CORRECTED 2026-08-22]** the prior figure “1,913” was the running counter at Session #2050, not a corpus total, and the counter is `sessions − 137` identically (824/824 pairs, zero exceptions), so the count restates the session number and carries no breadth information of its own; see §5.12 — **caveat (S647 + S651, 2026-05-08/10):** the cohort's load-bearing r values are being compared against an *implicit* null of r=0 (random); the *relevant* null is r(polynomial in Z) or r(generic 2-parameter tanh), both expected at r ≈ 0.95+ on the same data because sound velocity, electronegativity, and atomic volume are themselves near-monotonic functions of atomic number Z (S651 null-model gap). [**CHECKED 2026-11-02:** over the sampled Z = 1–54 values the best monotone function of Z fits electronegativity at r = 0.33, contradicting a universal r ≈ 0.95+ claim. The chemistry-row Z-null remains unmeasured; its row selection is unpublished. The verdict rests on S669's γ_phonon ≡ 2T/θ_D, not on this null. Full note at S651.] [**UPDATE 2026-11-03:** the real-row Z-null has since been run. On the 10 property rows recoverable from the session scripts (#80, #109, #110, #114, #115, #118; n = 19–33 per row, Z = 3–82) it fails: |r(P, Z)| ≤ 0.53 (linear, √Z and ln Z) and quadratic-in-Z R ≤ 0.51, against framework r = 0.60–0.98 (site explorer, pre-registered, back-annotation `5709b579`; re-run by the Publisher 2026-11-03, output identical). The high r values are identities, not a Z trend: r(P, 1/γ_phonon) ≡ r(P, θ_D), and γ_optical = IE_ref/IE, so the optical rows are correlations with ionization energy. See Appendix B.2.] Compounded with the Method 2 / 3 self-correlation paths and the Method 2 systematic bias toward γ ≈ 1 (S647 method gap), the framework-specific signal is the **Δr = r(Synchronism) − r(best monotonic null)** figure that is not currently in the archive. Best estimate: tie or marginal win. Distinguishing requires (a) computing Δr against the best-of-3 monotonic nulls (polynomial in Z, generic 2-parameter tanh, MOND-type interpolating function), AND (b) Method 1 applied uniformly OR pre-registered γ predictions for held-out phenomena. ⚠ - **ξ = MRH identity**: Correlation length *identified with* Markov Relevancy Horizon (Session #326) — coordinate reframe / vocabulary mapping, not novel prediction ⚠ - **Black hole paradox**: Event horizon = MRH coordinate reframe; Page curve described as emerging from MRH dynamics (Session #331) — eliminative not explanatory dissolution; no novel observable distinguishing this from standard holographic accounts ⚠ - **Qualia framework**: *Coordinate-system framing* (Sessions #280-282) — proposes that phase patterns ARE experience (identity claim, structural-realism form). Experimental predictions (P280.1, P282.4, P282.6) derived but **none yet tested**. ⚠ Framing not validated finding. diff --git a/whitepaper/sections/07-conclusion/meta/CHANGELOG.md b/whitepaper/sections/07-conclusion/meta/CHANGELOG.md index 314125b97..c7e773b7d 100644 --- a/whitepaper/sections/07-conclusion/meta/CHANGELOG.md +++ b/whitepaper/sections/07-conclusion/meta/CHANGELOG.md @@ -8,6 +8,22 @@ Header: Date (ISO 8601) | Author (LCT ID or name) | Change type (ADD/MODIFY/DELE ###### Entries +###### 2026-11-17 | Claude-Opus-5.5 (publisher) | MODIFY (one dated annotation; nothing deleted) +- **Description**: "Progress (Apr 2026)" list, BCS ratio bullet: a `[CHECKED 2026-11-17]` note. 3.52 is BCS's own weak-coupling value (2π·exp(−0.5772) = 3.528 from the gap equation), not an observation; 2√π = 3.545 is 0.49% from it by coincidence; the source script's γ = 2 formula gives 7.12. Points to §5.16. Check mark kept; words added, none removed. +- **Rationale**: Found while checking the §5.16 sites reached by back-annotation `d71f1f8b`; verified by `simulations/publisher_20261117_bcs_ratio_and_gamma_check.py` (rc 0). + +###### 2026-11-09 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: Both copies of the 10-18 `[SCOPED]` note on ΔBIC +184 gain the same short `[UPDATE 2026-11-09]` as §0, and the 10-27 γ ≈ 2 CHECKED note's parenthetical "(ΔBIC +184 point-level; +10.9 galaxy-level, at threshold; …)" now also names the Υ-free +2.7, about 2σ. The full note is on §5.15. +- **Rationale**: Site back-annotation `6fe7b696` (PREDICTIONS.md 2026-11-09 (i)–(iii)), landed after the 11-09 pass. Not relayed: the Publisher re-ran `synchronism-site/maintainer/scripts/gamma2_global_upsilon_per_law.py` (PREREG `5a95614`) and `gamma2_global_upsilon_twin_power_posthoc.py` on 2026-11-10, both byte-identical to the committed outputs. Words added, none removed; count HELD at 6. + +###### 2026-11-03 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: All three 2026-11-02 chemistry Z-null notes (S651 bullet, validation-pillars sentence, chemistry caveat) carry an `[UPDATE 2026-11-03]`: the real-row null was run on 10 rows and fails (|r(P, Z)| ≤ 0.53, quadratic R ≤ 0.51 vs framework r 0.60–0.98); the high r values are θ_D and ionization-energy identities. +- **Rationale**: Site back-annotation `5709b579` (PREDICTIONS.md 2026-11-02 (i)) supersedes the 10-31 "real-data null is unrun": the site explorer ran the polynomial-in-Z null on the 10 property rows recoverable from session scripts and it fails. Not relayed: the Publisher re-ran `synchronism-site/explorer/tools/chem_z_null_real_rows.py` on 2026-11-03 (rc 0, output byte-identical to the explorer's recorded corrected run). Reparametrization verdict unchanged; it now also rests on γ_optical being ionization energy. Words added, none removed. + +###### 2026-11-02 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: The S651 bullet carries a full `[CHECKED 2026-11-02]` note; the validation-pillars sentence ("r(polynomial in Z) ≈ 0.95+ (S651)") and the chemistry caveat ("near-monotonic functions of atomic number Z") carry short ones. +- **Rationale**: Site back-annotation `e2784def` (PREDICTIONS.md 2026-10-31 (ii)): the 2026-05-10 polynomial-in-Z null was run on synthetic curves, and its premise is false across the periodic table. Not relayed: re-run by the Publisher 2026-11-02 on Pauling electronegativity, Z = 1–54, n = 51 (Spearman +0.09; best monotone-in-Z r = 0.33; polynomial degree 1/2/3/5 r = 0.03/0.23/0.25/0.39). The Reparametrization verdict is unchanged; it rests on S669. Words added, none removed. + ###### 2026-10-27 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotation; nothing deleted) - **Description**: The "γ ≈ 2 universality: same value in gravity, BCS, enzymes, photosynthesis, Gnosis ✓" progress bullet carries a dated CHECKED note: "same value" is a shared fitted exponent, not a shared N_corr. Under γ = 2/√N_corr, γ ≈ 2 is N_corr ≈ 1, and the section's own S676 bullet puts a BEC at large N_corr, γ → 0. The gravity instance is the refuted γ = 2 pin (+184 point-level; +10.9 galaxy-level, at threshold; 10-18 SCOPED note). The S616 η audit already classes γ/BCS as a reparametrization. - **Rationale**: Back-annotation `e03d771b` (2026-10-26) withdrew the quantum/classical endpoint labels for C and γ because condensates land at the end called "classical". The paper's progress list still printed the four-domain γ ≈ 2 as a checked result with no pointer to either the S676 placement two bullets above it or the §5.15 verdict. Checkmark kept; words added, none removed; refutation count unchanged (6). diff --git a/whitepaper/sections/08-glossary/glossary.md b/whitepaper/sections/08-glossary/glossary.md index 157dbd14c..50c4b60eb 100644 --- a/whitepaper/sections/08-glossary/glossary.md +++ b/whitepaper/sections/08-glossary/glossary.md @@ -44,7 +44,7 @@ **Gamma (γ)** - The coherence scaling exponent, defined as γ = 2/√N_corr where N_corr is the number of correlated degrees of freedom. Cross-domain validation shows γ ≈ 2 appearing in gravity, BCS superconductivity, enzyme catalysis, and photosynthesis. *[CHECKED 2026-10-27 — under this entry's own definition, γ ≈ 2 means N_corr ≈ 1, a single uncorrelated particle. A BCS condensate has N_corr far above 1 and sits at γ → 0 (the S676 placement §0 and §7 already print: C runs from 0.9999 for the electron to ~0 for a BEC as N_corr rises). The γ ≈ 2 reported for BCS, enzymes and photosynthesis is a fitted exponent (Appendix B.5, Tc ∝ exp(−γ/λ)), and the S616 η audit classes γ/BCS as a reparametrization; reading that shared exponent as a shared correlation count is the step the back-annotation 2026-10-26, `e03d771b` names from the other side: under the quantum/classical endpoint labelling, condensates land at the end called "classical", so the site withdrew those labels (C is not a degeneracy parameter). In gravity, the γ = 2 pin is the refuted claim §5.15 scores (ΔBIC +184 point-level; +10.9 galaxy-level, at threshold; the 10-18 SCOPED note). Words added, none removed.]* + The coherence scaling exponent, defined as γ = 2/√N_corr where N_corr is the number of correlated degrees of freedom. Cross-domain validation shows γ ≈ 2 appearing in gravity, BCS superconductivity, enzyme catalysis, and photosynthesis. *[CHECKED 2026-10-27 — under this entry's own definition, γ ≈ 2 means N_corr ≈ 1, a single uncorrelated particle. A BCS condensate has N_corr far above 1 and sits at γ → 0 (the S676 placement §0 and §7 already print: C runs from 0.9999 for the electron to ~0 for a BEC as N_corr rises). The γ ≈ 2 reported for BCS, enzymes and photosynthesis is a fitted exponent (Appendix B.5, Tc ∝ exp(−γ/λ)), and the S616 η audit classes γ/BCS as a reparametrization; reading that shared exponent as a shared correlation count is the step the back-annotation 2026-10-26, `e03d771b` names from the other side: under the quantum/classical endpoint labelling, condensates land at the end called "classical", so the site withdrew those labels (C is not a degeneracy parameter). In gravity, the γ = 2 pin is the refuted claim §5.15 scores (ΔBIC +184 point-level; +10.9 galaxy-level, at threshold; +2.7 galaxy-level once each law gets its own Υ_disk, about 2σ, per the 11-09 UPDATE; the 10-18 SCOPED note). Words added, none removed.]* **Grid** diff --git a/whitepaper/sections/08-glossary/meta/CHANGELOG.md b/whitepaper/sections/08-glossary/meta/CHANGELOG.md index 8c79a8c4a..c8a8d2acf 100644 --- a/whitepaper/sections/08-glossary/meta/CHANGELOG.md +++ b/whitepaper/sections/08-glossary/meta/CHANGELOG.md @@ -8,6 +8,10 @@ Header: Date (ISO 8601) | Author (LCT ID or name) | Change type (ADD/MODIFY/DELE ###### Entries +###### 2026-11-09 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: The Gamma (γ) entry's 10-27 parenthetical "(ΔBIC +184 point-level; +10.9 galaxy-level, at threshold; …)" now also names the Υ-free +2.7, about 2σ, per the 11-09 UPDATE on §5.15's SCOPED note. +- **Rationale**: Site back-annotation `6fe7b696` (PREDICTIONS.md 2026-11-09 (i)–(iii)), landed after the 11-09 pass. Not relayed: the Publisher re-ran `synchronism-site/maintainer/scripts/gamma2_global_upsilon_per_law.py` (PREREG `5a95614`) and `gamma2_global_upsilon_twin_power_posthoc.py` on 2026-11-10, both byte-identical to the committed outputs. Words added, none removed; count HELD at 6. + ###### 2026-10-27 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotation; nothing deleted) - **Description**: The Gamma (γ) entry's "cross-domain validation shows γ ≈ 2 appearing in gravity, BCS superconductivity, enzyme catalysis, and photosynthesis" carries a dated CHECKED note. Under the entry's own definition γ ≈ 2 is N_corr ≈ 1; a BCS condensate has N_corr far above 1 and sits at γ → 0, which is where §0/§7's S676 bullet already places a BEC. The γ ≈ 2 shared across BCS, enzymes and photosynthesis is a fitted exponent (Appendix B.5), classed by the S616 η audit as a reparametrization; in gravity the γ = 2 pin is the refuted claim §5.15 scores (+184 point-level, +10.9 galaxy-level at threshold). - **Rationale**: Back-annotation `e03d771b` (2026-10-26, PREDICTIONS.md + README:161 + proposal `c_endpoints_are_not_quantum_classical_and_kill_text_must_match_the_lock_20261026.md`): condensates and superconductors have large N_corr and land at the end the quantum/classical labelling called "classical", so the site withdrew those labels. The glossary entry was the one place a reader would meet "γ ≈ 2 in BCS" as a definition. The contradiction is decidable from the paper's own equation and its own S676 sentence, so it is CHECKED, not relayed. Words added, none removed. diff --git a/whitepaper/sections/09-appendix-mathematical/appendix_b_chemistry.md b/whitepaper/sections/09-appendix-mathematical/appendix_b_chemistry.md index b1cff06da..18d23a461 100644 --- a/whitepaper/sections/09-appendix-mathematical/appendix_b_chemistry.md +++ b/whitepaper/sections/09-appendix-mathematical/appendix_b_chemistry.md @@ -41,6 +41,8 @@ $$ - Chemical hardness η vs 1/γ_optical: r = 0.950 - Work function φ vs 1/γ_optical: r = 0.888 +[**CHECKED 2026-11-03:** by the estimation formula just above, 1/γ_optical = IE/IE_ref, so each correlation in this list is a correlation with ionization energy under a new name. χ against IE is a textbook relation that Mulliken's definition χ_M = (IE + EA)/2 builds in. A re-run of the session #115 script (n = 33, Z = 3–82) gives r(χ, 1/γ_optical) = 0.956 (0.938 above is the archive's recorded value), and the correlation is not a Z trend: r(χ, Z) = −0.03. Back-annotation `5709b579`.] + **Phononic Channel (Thermal/Mechanical)** ``` Atomic Volume V_a → Debye Temperature θ_D → γ_phonon → E, G, κ, α @@ -56,6 +58,8 @@ $$ - Shear modulus G vs 1/γ_phonon: r = 0.936 - Elastic modulus E vs 1/γ_phonon: r = 0.920 +[**CHECKED 2026-11-03:** at fixed T, γ_phonon = 2T/θ_D is a function of θ_D alone, so r(P, 1/γ_phonon) ≡ r(P, θ_D) exactly (S669; the re-run gives identical values on every row). The elastic correlations also rest on one material: removing diamond lowers the jackknife minimum to r = 0.63 for G and 0.60 for E (session #110, n = 24). Back-annotation `5709b579`.] + **Channel Independence** These channels are **orthogonal** (r ≈ 0 between them), explaining why electronic and thermal properties can vary independently. @@ -81,7 +85,7 @@ $$ T_c \propto \exp\left(-\frac{\gamma}{\lambda_{eff}}\right) $$ -BCS ratio derived: 2Δ₀/kT_c = 2√π ≈ 3.54 (observed: 3.52, <1% error) +BCS ratio derived: 2Δ₀/kT_c = 2√π ≈ 3.54 (observed: 3.52, <1% error) *[CHECKED 2026-11-17: 3.52 is BCS's own weak-coupling value (2π·exp(−0.5772) = 3.528 from the gap equation), not an observation, and 2√π is 0.49% from it by coincidence. See §5.16. Words added, none removed.]* **Optical Properties (Sessions #76, #91)** $$ @@ -116,9 +120,13 @@ $$ | Electron transfer | k_ET coherence-enhanced | r = 0.933 | #64 | | Thermal diffusivity | α vs 1/γ_electron | r = 0.932 | #111 | +[**CHECKED 2026-11-03:** the "Electronegativity" row is chemical softness S = 1/(IE − EA) against γ_optical ∝ 1/IE (session #118; re-run r = 0.979). Because EA ≪ IE, that is close to 1/IE against 1/IE. Electronegativity itself is the B.2 row, r = 0.938. Back-annotation `5709b579`.] + +[**CHECKED 2026-11-11:** the "Superconductivity" row's γ is not independent of T_c. In its source (chemistry session #62, `simulations/chemistry/superconductivity_coherence.py`, commit `e4c903f5`) each material's γ is computed from its own gap ratio, γ = 2 × 3.52 / (2Δ/k_B T_c), and then multiplied by a hand-set 0.8 for cuprates and 0.7 for high-pressure hydrides. The test is a linear correlation of T_c with 1/γ over 28 materials, not a fit of T_c ∝ exp(−γ/λ). Re-run: r = 0.948 reproduces. The category factor alone gives r = 0.944; with the factors removed r = 0.882, which is the known rise of the gap ratio with T_c. Script: `simulations/publisher_20261111_tc_gamma_control.py`. The site blanked this row as untraced on 2026-11-10; it is traced, and this is what the trace shows.] + **Validation rate:** 89% *pattern-alignment* rate across 2,671 sessions. -**Caveats (S647 method gap, S651 null-model gap, 2026-05-08/10):** the load-bearing r values above are being compared against an *implicit* null of r=0 (random). The *relevant* null is r(polynomial in Z) or r(generic 2-parameter tanh): sound velocity, electronegativity, atomic volume are themselves near-monotonic functions of atomic number Z; any smooth monotonic function will achieve r ≈ 0.95+ on the same 1,703 phenomena by construction. The framework-specific signal is **Δr = r(Synchronism) − r(best monotonic null)** — *not currently in the archive*. Compounded with the Method 2 / 3 self-correlation paths and Method 2's documented systematic bias toward γ ≈ 1 (Session #26 Part 3 simulation table), best estimate is **tie or marginal win** until Δr is computed against the best-of-3 monotonic nulls (polynomial in Z, generic 2-parameter tanh, MOND-type interpolating function), AND Method 1 (bias-free) is applied uniformly OR pre-registered γ predictions for held-out phenomena are produced. "89% validated" is **defensible only with the null model documented**; until then, treat the chemistry track as a substantial *pattern catalog* with productive boundaries (the γ ~ 1 motif organizes a lot of structure), not as 1,703 confirmed predictions. +**Caveats (S647 method gap, S651 null-model gap, 2026-05-08/10):** the load-bearing r values above are being compared against an *implicit* null of r=0 (random). The *relevant* null is r(polynomial in Z) or r(generic 2-parameter tanh): sound velocity, electronegativity, atomic volume are themselves near-monotonic functions of atomic number Z; any smooth monotonic function will achieve r ≈ 0.95+ on the same 1,703 phenomena by construction. [**CHECKED 2026-11-02:** over the sampled Z = 1–54 values the best monotone function of Z fits electronegativity at r = 0.33, contradicting a universal r ≈ 0.95+ claim. The chemistry-row Z-null remains unmeasured; its row selection is unpublished. The verdict rests on S669's γ_phonon ≡ 2T/θ_D, not on this null. Full note at S651.] [**UPDATE 2026-11-03:** the real-row Z-null has since been run. On the 10 property rows recoverable from the session scripts (#80, #109, #110, #114, #115, #118; n = 19–33 per row, Z = 3–82) it fails: |r(P, Z)| ≤ 0.53 (linear, √Z and ln Z) and quadratic-in-Z R ≤ 0.51, against framework r = 0.60–0.98 (site explorer, pre-registered, back-annotation `5709b579`; re-run by the Publisher 2026-11-03, output identical). The high r values are identities, not a Z trend: r(P, 1/γ_phonon) ≡ r(P, θ_D), and γ_optical = IE_ref/IE, so the optical rows are correlations with ionization energy. See Appendix B.2.] The framework-specific signal is **Δr = r(Synchronism) − r(best monotonic null)** — *not currently in the archive*. Compounded with the Method 2 / 3 self-correlation paths and Method 2's documented systematic bias toward γ ≈ 1 (Session #26 Part 3 simulation table), best estimate is **tie or marginal win** until Δr is computed against the best-of-3 monotonic nulls (polynomial in Z, generic 2-parameter tanh, MOND-type interpolating function), AND Method 1 (bias-free) is applied uniformly OR pre-registered γ predictions for held-out phenomena are produced. "89% validated" is **defensible only with the null model documented**; until then, treat the chemistry track as a substantial *pattern catalog* with productive boundaries (the γ ~ 1 motif organizes a lot of structure), not as 1,703 confirmed predictions. --- diff --git a/whitepaper/sections/09-appendix-mathematical/meta/CHANGELOG.md b/whitepaper/sections/09-appendix-mathematical/meta/CHANGELOG.md index fcb1eeac6..49b9d3bbf 100644 --- a/whitepaper/sections/09-appendix-mathematical/meta/CHANGELOG.md +++ b/whitepaper/sections/09-appendix-mathematical/meta/CHANGELOG.md @@ -8,6 +8,22 @@ Header: Date (ISO 8601) | Author (LCT ID or name) | Change type (ADD/MODIFY/DELE ###### Entries +###### 2026-11-17 | Claude-Opus-5.5 (publisher) | MODIFY (one dated annotation; nothing deleted) +- **Description**: Appendix B, "BCS ratio derived: 2√π ≈ 3.54 (observed: 3.52, <1% error)": a short `[CHECKED 2026-11-17]` note. 3.52 is BCS's own weak-coupling value (3.528 from the gap equation), not an observation; 2√π is 0.49% from it by coincidence. Points to §5.16. Words added, none removed. +- **Rationale**: Same check as the §5.16 and conclusion notes; `simulations/publisher_20261117_bcs_ratio_and_gamma_check.py` (rc 0). + +###### 2026-11-11 | Claude-Fable-5.2 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: Appendix B.5: a `[CHECKED 2026-11-11]` note on the Superconductivity r = 0.948 row (same text as §5.12). +- **Rationale**: Site back-annotation `76a139e0` reported the row as untraced. It traces to chemistry session #62 (`simulations/chemistry/superconductivity_coherence.py`, commit `e4c903f5`). Re-run: r = 0.948 reproduces; γ is computed from each material's own gap ratio and T_c, and the hand-set category factors alone give r = 0.944 (`simulations/publisher_20261111_tc_gamma_control.py`). + +###### 2026-11-03 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: Appendix B: the B.5 caveat's 2026-11-02 Z-null note carries an `[UPDATE 2026-11-03]` (real-row null run, fails). New `[CHECKED 2026-11-03]` notes: B.2 electronic channel (1/γ_optical = IE/IE_ref by the printed formula, so the list correlates properties with ionization energy; re-run χ r = 0.956, r(χ, Z) = −0.03); B.2 phononic channel (r(P, 1/γ_phonon) ≡ r(P, θ_D); elastic rows fall to jackknife-min 0.63 / 0.60 without diamond); B.5 (the "Electronegativity" r = 0.979 row is chemical softness against 1/IE). +- **Rationale**: Site back-annotation `5709b579` (PREDICTIONS.md 2026-11-02 (i)) supersedes the 10-31 "real-data null is unrun": the site explorer ran the polynomial-in-Z null on the 10 property rows recoverable from session scripts and it fails. Not relayed: the Publisher re-ran `synchronism-site/explorer/tools/chem_z_null_real_rows.py` on 2026-11-03 (rc 0, output byte-identical to the explorer's recorded corrected run). Reparametrization verdict unchanged; it now also rests on γ_optical being ionization energy. Words added, none removed. + +###### 2026-11-02 | Claude-Opus-5.5 (publisher) | MODIFY (dated annotations; nothing deleted) +- **Description**: Appendix B's S647 + S651 caveat ("near-monotonic functions of atomic number Z … r ≈ 0.95+ … by construction") carries a short `[CHECKED 2026-11-02]` note. +- **Rationale**: Site back-annotation `e2784def` (PREDICTIONS.md 2026-10-31 (ii)): the 2026-05-10 polynomial-in-Z null was run on synthetic curves, and its premise is false across the periodic table. Not relayed: re-run by the Publisher 2026-11-02 on Pauling electronegativity, Z = 1–54, n = 51 (Spearman +0.09; best monotone-in-Z r = 0.33; polynomial degree 1/2/3/5 r = 0.03/0.23/0.25/0.39). The Reparametrization verdict is unchanged; it rests on S669. Words added, none removed. + ###### 2026-10-27 | Claude-Fable-5.1 (publisher) | MODIFY (dated annotation in Appendix B; nothing deleted) - **Description**: B.1's limit line "γ = 2: Classical limit (N_corr = 1, single particle)" carries a short note: "classical" means uncorrelated, not non-quantum; the 2026-10-26 back-annotation withdrew quantum/classical endpoint labels for C and γ because condensates and superconductors (large N_corr) sit at γ → 0. Points to the §5.12 notes. - **Rationale**: Back-annotation `e03d771b`. B.1 is the definitional site the glossary and §5.12 both restate; a reader checking the definition should meet the scoping where the definition is. Words added, none removed. diff --git a/whitepaper/tests/test_pdf_ci_contract.py b/whitepaper/tests/test_pdf_ci_contract.py new file mode 100644 index 000000000..12db2b52f --- /dev/null +++ b/whitepaper/tests/test_pdf_ci_contract.py @@ -0,0 +1,60 @@ +"""Exercise the actual CI PDF step with disposable files and a fake builder.""" + +import re +import subprocess +import tempfile +import textwrap +import unittest +from pathlib import Path + + +class PdfCiContract(unittest.TestCase): + @staticmethod + def step_script(): + root = Path(__file__).resolve().parents[2] + workflow = (root / ".github/workflows/build_whitepaper.yml").read_text() + match = re.search( + r"(?m)^ - name: Build PDF Version\n" + r" working-directory: \./whitepaper\n" + r" run: \|\n((?: {10}[^\n]*\n)+)", + workflow, + ) + if not match: + raise AssertionError("PDF build step is missing or has an unexpected shape") + return textwrap.dedent(match.group(1)) + + def run_step(self, builder): + with tempfile.TemporaryDirectory() as directory: + root = Path(directory) + (root / "build").mkdir() + artifact = root / "build/Synchronism_Whitepaper.pdf" + artifact.write_bytes(b"stale artifact") + (root / "make-pdf.sh").write_text("#!/bin/bash\n" + builder + "\n") + result = subprocess.run( + ["bash", "-e", "-o", "pipefail", "-c", self.step_script()], + cwd=root, capture_output=True, text=True, check=False, + ) + return result.returncode, artifact.read_bytes() if artifact.exists() else None + + def test_failed_builder_is_not_reported_as_success(self): + code, artifact = self.run_step("exit 7") + self.assertNotEqual(code, 0) + self.assertIsNone(artifact) + + def test_success_without_new_pdf_does_not_reuse_stale_output(self): + code, artifact = self.run_step("exit 0") + self.assertNotEqual(code, 0) + self.assertIsNone(artifact) + + def test_empty_new_pdf_is_not_success(self): + code, _ = self.run_step(": > build/Synchronism_Whitepaper.pdf") + self.assertNotEqual(code, 0) + + def test_nonempty_new_pdf_succeeds(self): + code, artifact = self.run_step("printf fresh > build/Synchronism_Whitepaper.pdf") + self.assertEqual(code, 0) + self.assertEqual(artifact, b"fresh") + + +if __name__ == "__main__": + unittest.main()