Add a certified pair registry so the £100k childcare lane can run on policyengine.py 6.11.0's Microcosm pair

app/public/data/lanes.json

@@ -4,7 +4,7 @@
   "country": "US | UK | BE | NZ \u2014 the model instance the lane feeds (issue #42)",
   "note": "Repo-committed status feed (issue #7); graduate to Supabase when lane count makes commits noisy. Every lane writes its transition here."
  },
- "updated": "2026-10-25",
+ "updated": "2026-10-26",
  "lanes": [
   {
    "id": "urban-sotsn",

app/public/data/populations.json

@@ -1,5 +1,5 @@
 {
- "built": "2026-10-25",
+ "built": "2026-10-26",
  "note": "Non-Urban populations exported from scorecard.db: the populace reform-validation registry (issue #20) plus the compute campaign's attached comparisons (US: TPC/CPSP/PWBM/CBO/JCT; UK: the HMRC ready-reckoner family and the complete OBR Policy measures database, including historical HM Treasury scores; BE: JRC EUROMOD-BE claims with demo-grade Axiom concept-mismatch attachments and the 2026 PIT-reform scorekeeper claims; NZ: official budget scores). Claims explicitly opted into pre-result display remain in the feed with latest status not_computed and empty history. Recorded result statuses and calibration relationships are verbatim; nothing here is a pass/fail grade.",
  "summary": {
   "claims": 29771,
@@ -30104,7 +30104,7 @@
     "measure_key": "uk_free_childcare_hours_and_tfc_100k_limit_removed",
     "program": "free_childcare_hours_and_tax_free_childcare",
     "scoring_method": "post_behavioural",
-    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json)"
+    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json, data/uk/certified_pairs.json)"
    },
    "measure": null,
    "reform_framework": "policy_ref",
@@ -30161,7 +30161,7 @@
     "measure_key": "uk_free_childcare_hours_and_tfc_100k_limit_removed",
     "program": "free_childcare_hours_and_tax_free_childcare",
     "scoring_method": "post_behavioural",
-    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json)"
+    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json, data/uk/certified_pairs.json)"
    },
    "measure": null,
    "reform_framework": "policy_ref",

data/lanes.json

@@ -4,7 +4,7 @@
   "country": "US | UK | BE | NZ \u2014 the model instance the lane feeds (issue #42)",
   "note": "Repo-committed status feed (issue #7); graduate to Supabase when lane count makes commits noisy. Every lane writes its transition here."
  },
- "updated": "2026-10-25",
+ "updated": "2026-10-26",
  "lanes": [
   {
    "id": "urban-sotsn",

data/populations.json

@@ -1,5 +1,5 @@
 {
- "built": "2026-10-25",
+ "built": "2026-10-26",
  "note": "Non-Urban populations exported from scorecard.db: the populace reform-validation registry (issue #20) plus the compute campaign's attached comparisons (US: TPC/CPSP/PWBM/CBO/JCT; UK: the HMRC ready-reckoner family and the complete OBR Policy measures database, including historical HM Treasury scores; BE: JRC EUROMOD-BE claims with demo-grade Axiom concept-mismatch attachments and the 2026 PIT-reform scorekeeper claims; NZ: official budget scores). Claims explicitly opted into pre-result display remain in the feed with latest status not_computed and empty history. Recorded result statuses and calibration relationships are verbatim; nothing here is a pass/fail grade.",
  "summary": {
   "claims": 29771,
@@ -30104,7 +30104,7 @@
     "measure_key": "uk_free_childcare_hours_and_tfc_100k_limit_removed",
     "program": "free_childcare_hours_and_tax_free_childcare",
     "scoring_method": "post_behavioural",
-    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json)"
+    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json, data/uk/certified_pairs.json)"
    },
    "measure": null,
    "reform_framework": "policy_ref",
@@ -30161,7 +30161,7 @@
     "measure_key": "uk_free_childcare_hours_and_tfc_100k_limit_removed",
     "program": "free_childcare_hours_and_tax_free_childcare",
     "scoring_method": "post_behavioural",
-    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json)"
+    "pe_missing": "fiscal year 2031-32 is outside the data years of every certified pair this lane has run (pe5.0.2_uk2.89.2_populace-uk-2023-dd68c73-4aa4b14-20260619T023711Z 2023-2030; data/uk/replay_variants.json, data/uk/certified_pairs.json)"
    },
    "measure": null,
    "reform_framework": "policy_ref",

data/uk/certified_pairs.json

@@ -0,0 +1,97 @@
+{
+ "_what": "Certified pairs a lane runs on outside the replay variant index (pipeline/uk_certified_pairs.py). Each is one policyengine.py release's certified bundle, in the replay variant's identity shape, under the id pair-<variant slug>. Never a replay variant: no event replay runs on one.",
+ "kind": "uk_certified_pairs",
+ "pairs": [
+  {
+   "certification": {
+    "certified_for_model_version": "2.132.6",
+    "compatibility_basis": "built_with_model_package",
+    "data_build_id": "microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71",
+    "source": "policyengine.py 6.11.0 packaged bundle manifest (policyengine/data/bundle/manifest.json, sha256 bccf23acde631aed4665757094c04f86e60993fb9d388e7dc958b90c876e5a80; PyPI wheel sha256 b2aa90b97a5c608f4aaad5e6833c21014d786bfd96bb831c9c51b9bb60ff2a50); the pair was certified in PolicyEngine/policyengine.py#685 (merged 2026-10-21)"
+   },
+   "data_years": [
+    2024,
+    2025,
+    2026,
+    2027,
+    2028,
+    2029,
+    2030,
+    2031,
+    2032
+   ],
+   "data_years_evidence": {
+    "simulation_dataset_years": [
+     2024,
+     2025,
+     2026,
+     2027,
+     2028,
+     2029,
+     2030,
+     2031,
+     2032
+    ],
+    "source": "Measured for the same policyengine-uk 2.132.6, policyengine-core 3.33.3 and dataset by the experiment population probe under policyengine 6.10.0 (data/uk/experiments/probes/exp-pe6.10.0_uk2.132.6_microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71.json); every lane run checks its managed simulation's sim.dataset.years against this list before any calculation."
+   },
+   "executor": [
+    "local"
+   ],
+   "identity": {
+    "dataset": {
+     "path": "microcosm_uk_2024_25.h5",
+     "repo_id": "policyengine/populace-uk-private",
+     "repo_type": "dataset",
+     "revision": "microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71",
+     "sha256": "810ba980005a8b907cb51464279fa5e0125464b7af9a4871b7f6744c30951ece",
+     "size_bytes": 321798995
+    },
+    "materialization": "policyengine_data_dir",
+    "packaged_bundle_manifest_sha256": "bccf23acde631aed4665757094c04f86e60993fb9d388e7dc958b90c876e5a80",
+    "packages": {
+     "policyengine": {
+      "record_sha256": "113897bc23a82dab3fd29a0715c7a111fbb948f9b791eaf866e1324d75c06bbf",
+      "version": "6.11.0"
+     },
+     "policyengine-core": {
+      "record_sha256": "e68eb6a81f472df42aff4fbf56c8537d1185075078ffb2e0a6399f1df0b66bc2",
+      "version": "3.33.3"
+     },
+     "policyengine-uk": {
+      "record_sha256": "ca7c36b7617f950f48a04e2816a29e1eb962c44097f3ef08ed4036d7414d2709",
+      "version": "2.132.6"
+     }
+    },
+    "python": "3.12",
+    "release_bundle": {
+     "bundle_id": "uk-6.11.0",
+     "certified_by": "policyengine.py bundle certification",
+     "certified_data_artifact_sha256": "810ba980005a8b907cb51464279fa5e0125464b7af9a4871b7f6744c30951ece",
+     "certified_data_build_id": "microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71",
+     "compatibility_basis": "built_with_model_package",
+     "country_id": "uk",
+     "data_build_fingerprint": null,
+     "data_build_model_git_sha": null,
+     "data_build_model_version": "2.132.6",
+     "data_package": "microcosm-data",
+     "data_version": "0.1.0",
+     "default_dataset": "microcosm_uk_2024_25",
+     "default_dataset_uri": "hf://policyengine/populace-uk-private/microcosm_uk_2024_25.h5@microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71",
+     "model_package": "policyengine-uk",
+     "model_version": "2.132.6",
+     "policyengine_version": "6.11.0"
+    },
+    "requirements_sha256": "f57a4899ec639cd89ae0eba15525f1d0bba25ff9bc091c95a14ae4d46d363d80"
+   },
+   "identity_sha256": "9bb966d62e65c29e44a83620b21d720934fb33e57b8fc13695254657acd9b35a",
+   "note": "policyengine.py 6.11.0 (released 25 October 2026) with the pair its bundle certifies: policyengine-uk 2.132.6 and the Microcosm UK 2024-25 national cut built with it. A lane runs on it before the release-replay pipeline qualifies it; no event replay runs on it.",
+   "pair_id": "pair-pe6.11.0_uk2.132.6_microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71",
+   "requirements": {
+    "path": "docs/uk_replay/requirements/pairs/pair-pe6.11.0_uk2.132.6_microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71.txt",
+    "sha256": "f57a4899ec639cd89ae0eba15525f1d0bba25ff9bc091c95a14ae4d46d363d80"
+   },
+   "status": "certified_pair"
+  }
+ ],
+ "schema_version": 1
+}

docs/ARCHITECTURE.md

@@ -240,7 +240,8 @@ Operating rules for the agent stage:
   come from `pipeline/compute_uk_childcare_100k.py` on the certified
   populace-uk pair: static, the three free-hours entitlements (plus
   Tax-Free Childcare for CBP) summed for people in English households.
-  Each certified variant's artifact keeps its run in the history, and a
+  Each certified variant's or certified pair's
+  (`data/uk/certified_pairs.json`) artifact keeps its run in the history, and a
   claim's latest result is the newest run that computed its year. Ingest
   takes an artifact only for a pair with a `verified_pairs` entry, and an
   engine or run note attaches only to the pairs its `variants` names.

docs/uk_replay/RELEASE_REPLAY.md

@@ -181,6 +181,34 @@ A model/data combination is **incompatible** when no certified variant
 records it. There is no flag that relaxes a check: running a newer engine
 means adding a variant whose certification names it.
 
+### Certified pairs (lane runs only)
+
+A lane can need a policyengine.py release's certified bundle before this
+pipeline qualifies it as a variant: `uk-childcare-100k` (issue #228) runs on
+policyengine.py 6.11.0's Microcosm pair. `data/uk/certified_pairs.json`
+(kind `uk_certified_pairs`, `pipeline/uk_certified_pairs.py`) records each
+such bundle in the variant record's shape less the fields that file a
+replay: `pair_id` (`pair-<variant slug>`) in place of `variant_id`, status
+`certified_pair`, and no `layout` or `result_root`
+(`uk_certified_pairs.FIELDS` is the whole field set).
+
+- **The checks are the variant's.** `uk_replay_variants.validate_identity`
+  is check 1 for both kinds; `validate_record` adds only the variant's
+  status, layout, result root and legacy checks. A pair's preflight then
+  runs checks 2 to 6 through the same code: `verify_installed`, then
+  `compute_uk_event.materialized_preflight`, the second half of
+  `variant_preflight`.
+- **No event replay runs on a pair.** The event runner resolves only
+  `replay_variants.json`; a lane resolves either kind with
+  `uk_certified_pairs.get_record`, and its ingest reads both files.
+- **A pair stays when a variant arrives.** A variant this pipeline later
+  qualifies for the same bundle has its own requirements file, so its own
+  identity and id. The pair remains the identity the lane's runs on it were
+  checked against.
+- Experiment populations (`docs/uk_replay/POLICY_EXPERIMENTS.md`) are a
+  third kind: their results are never ingested, while a lane's runs on a
+  pair are.
+
 ## Unsupported years
 
 Each event registry's `calendar_years` is the construction's window; a

docs/uk_replay/requirements/pairs/pair-pe6.11.0_uk2.132.6_microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71.txt

@@ -0,0 +1,85 @@
+# Certified pair pair-pe6.11.0_uk2.132.6_microcosm-uk-2024-25-national-20261019T083323Z-6cba5d71 (data/uk/certified_pairs.json, pipeline/uk_certified_pairs.py); not a replay variant.
+# Resolved with uv pip compile for Python 3.12, x86_64-manylinux_2_28, within the exp-pe6.10.0 experiment population's freeze; identical to it apart from the policyengine pin.
+# policyengine 6.11.0 from PyPI (policyengine-6.11.0-py3-none-any.whl, sha256 b2aa90b97a5c608f4aaad5e6833c21014d786bfd96bb831c9c51b9bb60ff2a50).
+annotated-types==0.8.0
+anyio==4.15.1
+asttokens==3.0.2
+attrs==26.1.0
+blosc2==4.14.1
+certifi==2026.7.22
+charset-normalizer==3.5.2
+click==8.5.0
+decorator==5.3.1
+dpath==2.2.0
+executing==2.3.0
+filelock==4.1.0
+fsspec==2026.9.0
+h11==0.16.0
+h2==4.4.1
+h5py==3.16.0
+hf-xet==1.7.0
+hpack==4.2.0
+httpcore==1.0.9
+httpcore2==2.13.1
+httpx==0.28.1
+httpx2==2.13.1
+huggingface-hub==2.2.0
+hyperframe==6.1.0
+idna==3.20
+iniconfig==2.3.1
+ipython==8.39.0
+jedi==0.20.0
+jinja2==3.1.6
+jsonpickle==4.1.3
+jsonschema==4.26.0
+jsonschema-specifications==2025.9.1
+markdown-it-py==4.2.0
+markupsafe==3.0.4
+matplotlib-inline==0.2.2
+mdurl==0.1.2
+microdf-python==1.3.10
+msgpack==1.2.3
+ndindex==1.10.1
+networkx==3.7
+numexpr==2.14.2
+numpy==2.5.4
+packaging==26.3
+pandas==3.0.6
+parso==0.8.7
+pexpect==4.9.0
+plotly==5.24.1
+pluggy==1.6.0
+policyengine==6.11.0
+policyengine-core==3.33.3
+policyengine-uk==2.132.6
+prompt-toolkit==3.0.53
+psutil==6.1.1
+ptyprocess==0.7.0
+pure-eval==0.2.4
+py-cpuinfo==9.0.0
+pydantic==2.14.0
+pydantic-core==2.50.0
+pygments==2.21.0
+pytest==9.1.1
+python-dateutil==2.9.0.post0
+pyvis==0.3.2
+pyyaml==6.0.3
+referencing==0.37.0
+requests==2.34.2
+rich==15.0.0
+rpds-py==2026.9.1
+six==1.17.0
+sortedcontainers==2.4.0
+stack-data==0.6.3
+standard-imghdr==3.13.0
+tables==3.11.1
+tenacity==9.2.1
+threadpoolctl==3.7.0
+tqdm==4.70.1
+traitlets==5.16.1
+truststore==0.10.4
+typing-extensions==4.16.0
+typing-inspection==0.4.4
+urllib3==2.8.0
+wcwidth==0.9.2
+wheel==0.48.0