Skip to content

UK newsletter, 29 October 2026: Autumn Budget briefing - #34

Merged
MaxGhenis merged 13 commits into
mainfrom
2026-10-29-uk
Oct 29, 2026
Merged

MaxGhenis merged 13 commits into
mainfrom
2026-10-29-uk

Conversation

@MaxGhenis

@MaxGhenis MaxGhenis commented Oct 27, 2026 •

Copy link
Copy Markdown
Contributor

This is the UK edition that follows the Autumn Budget briefing. The coordinator sent it to 199 UK subscribers at 05:03:20Z on 29 October, under decision d2100 (Max's go). It went from Mailchimp campaign 8f46150457, at d5a0ed6, with sent HTML sha256 a7cc3377…. Archive: http://eepurl.com/U9fW4Rq1Xv.

When it went

  • Plan: Max's plan of 28 October, 19:05Z, relayed by the Budget coordinator: send at 12:00 UK on Thursday 29 October, only on Max's "send it" (decision d2100).
  • Send: Max gave the go under d2100, which was DECIDED at 02:20Z. The coordinator sent at 05:03:20Z on 29 October through the Mailchimp API. Mailchimp shows 199 emails sent to 199 recipients.

Campaign

  • Subject: Autumn Budget 2026: our briefing
  • Preview text (matches the preheader): We estimate the seven household measures we model cost £3.11 billion in 2026-27 and raise net income by £107 per household on average that year.
  • Audience: UK (upload-newsletter … --audience uk), as a draft only.

Sources

Comparables for the 29 October UK newsletter (manifest revision 17), 29 October 2026

This covers every figure the edition prints. A gap of more than 25% needs an established reason, or the figure comes out.

Figure in the edition Year Ours (£m unless stated) Comparable Gap More than 25%? Reason
The seven measures cost £3.11 billion 2026-27 −3,110 The OBR's detailed forecast table 1.1 lines for the seven measures, summed: −3,935. Table 6.1 rows, summed: −3,955 −21% (−21.4% on Table 6.1) No —
They raise money (sign only, no amount printed) 2028-29 to 2031-32 +210, +166, +301, +281 OBR lines, summed: +235, +85, +248, +150 Same sign in every year n/a The edition prints no amount for these years. Their gaps with the OBR exceed 25% in 2029-30 (+95%) and 2031-32 (+87%) without fully established reasons, so the amounts stay out.
Average gain of £107 a household 2026-27 £107 a year None like-for-like n/a n/a See below
Every decile gains £98 to £110 2026-27 £98 to £110 a year None like-for-like n/a n/a See below
Highest-income decile loses £50, then £56 2028-29, 2031-32 −£50, −£56 a year None like-for-like n/a n/a See below

Why the household figures have no comparable.

  • The Treasury's Impact on households is the only official distributional analysis. It shows Budget 2026 decisions in 2027-28 only (Figures 1.A and 1.B; paragraphs 1.1 and 1.5). Its 2028-29 figures combine the effects of the decisions it models from Autumn Budget 2024 onwards (Figures 1.C and 1.D; paragraph 1.8). So it has no figure for 2026-27 or 2031-32, and its 2028-29 figure covers far more than the Budget.
  • Its scope differs too. It covers 18 measures, including the savings income rate rise, which we leave out. It leaves out the Household Energy Payment, which accounts for most of our 2026-27 gain, and the Carer's Allowance taper (paragraph 1.4; the briefing, "Who gains and loses"; the outside comparables log, row P2).
  • No other source has published a per-household or per-decile figure for these measures (the outside comparables log, sweep 3, 22:20Z on 28 October). The Resolution Foundation's decile chart (row H3) shows who is eligible for the Energy Payment, not an amount. The Resolution Foundation and the IFS publish their own analyses on 29 October, at 08:00Z and 10:30Z.

Other numbers in the edition are statutory parameters or dates, not estimates: £1,000 to £3,000, 2 percentage points, April 2027, July 2027, 2030-31 and April 2030. The 1:05 badge on the video thumbnail is the analyst walkthrough's running time (65.2 seconds). The thumbnail's frame shows only years and labels: Year 2027-28, All 7 policies, 2024-25, and 2026-27 to 2031-32.

What it says, in order

  1. Hero: the written briefing, https://www.policyengine.org/uk/research/uk-autumn-budget-2026-briefing.
    • Title "Autumn Budget 2026: our briefing on what it means for households", one line on scope, and three findings:
      • the seven measures cost £3.11 billion in 2026-27, and from 2028-29 to 2031-32 they raise money, as the higher rates on property income raise more than the other measures cost (review r1 pulled the 2027-28 cost and the later amounts: their gaps with the OBR exceed 25% without fully established reasons);
      • they raise household net income by £107 a year on average in 2026-27, with every decile gaining £98 to £110;
      • the highest-income decile loses £50 a year on average in 2028-29, rising to £56 in 2031-32.
    • The edition gives no poverty figure, because the briefing holds them back.
  2. Dashboard images (Max's ask via the coordinator, about 02:12Z: "this is the main thing we want people to try"). Each is a production screenshot on manifest revision 17 (9685adc1), captured at 2x and shown 600px wide. Mailchimp's content studio hosts both, and their alt text carries no figures.
    • The decile chart for 2026-27, under the second finding, links to ?tab=dashboard&sub=all&year=2026.
    • The calculator's input form, in the dashboard section, links to ?tab=household. It shows the form's default inputs, not a result: the result card still says "with 8 policies" next to "All 7".
    • The dashboard section keeps its line and its button, "See what the Budget means for your household", linking to https://www.policyengine.org/uk/budgets/autumn-2026.
    • Left out:
      • The revenue chart (review r3, finding 1). Its axis and dots show the 2027-28 cost and the 2028-29 to 2031-32 amounts, which fail the comparables gate above.
      • The constituency map. In 2028-29 the dashboard counts a withheld policy as zero: the property income rise is withheld in 173 seats, and 13 of the map's 14 teal seats are among them. In 2026-27 its fixed ±0.05% range clamps 632 of 650 seats, whose gains run to 0.29%.
  3. Video: the analyst walkthrough of the dashboard (557c699), in the dashboard section above the calculator form.
    • The thumbnail is 16:9, made from the video's own frame at 44 seconds (Methodology), with a play glyph and a 1:05 badge. Mailchimp's content studio hosts it, and it links to https://www.youtube.com/watch?v=CKDwiW4kL7o.
    • Why not the cover: the video's cover shows the 2027-28 cost and the revenue chart, both of which the comparables gate keeps out.
    • Alongside it: the alt text names the video, and the line beside it carries no figures.
    • Left out: the household calculator video. It had no approval by the cut-off, and 91b187a removes its unused still from assets/.
  4. Explainers: one line each, with no figures beyond statutory parameters.
    • Pension Tax Allowance, energy bills, Triple Lock and fuel duty, live before this round.
    • Property income tax (Autumn Budget 2026 property income tax explainer policyengine-app-v2#1460, merged as the property-only post), trading allowance (#1463, merged) and Carer's Allowance (#1464, merged).
    • Savings: one line, "Our estimate of the changes to the taxation of savings income will follow." The savings post (#1466) is still a draft, and Max's plan allows the line but no savings figure. Delete it if Max prefers.
  5. Webinar recording: https://www.youtube.com/watch?v=ZutKULXLhU0, the re-upload with the audience in. The deck session confirmed at 01:53Z that it's public, 40:44 long, and the latest upload.
  6. Credit: the Nuffield Foundation credit, linked, which tags the Foundation.

Left out, per Max:

  • any PolicyEngine savings figure;
  • Max's markets;
  • the "policies explained" PDF. 557c699 linked a hosted copy at the coordinator's request, and d5a0ed6 removed the link at 04:43Z after Pavel's review of the PDF's PR found errors in that version (sha256 6e12d26f…). The file stays in the content studio, unlinked.

Before upload

  • Fill every slot. grep -c '\[\[FILL' editions/2026-10-29-uk.html prints 0 at d7fd924. The video is held for a recapture after a dashboard fix (the coordinator, about 02:05Z): d7fd924 takes its slot back out, and it returns by updating the draft when Max's post of it exists.
  • Figures from production revision 17 only, through the briefing. Review r1 recomputed every printed figure from production.
  • Comparables review dated 29 October, above.
  • Review r1 (GPT-6.1 Sol, 20261028-215155-newsletter-1029-r1): REQUEST CHANGES, with two blocking findings on comparables and four on voice, case and template. 3748ec1 and 080fca4 answer all of them, except the Outlook accent bar, which CLAUDE.md says needs an Outlook test first. That bar is inherited from the template and isn't new here.
  • Review r2 (GPT-6.1 Sol, 20261028-220313-newsletter-1029-r2), a delta review of the fixes, the comparables note and the video section: APPROVE WITH CHANGES. It found r1's findings 1–5 resolved and recomputed the figures and signs from production. Its one change is the wording of the Treasury line in the comparables note above, now fixed. It found the video section correct too.
  • Links: the briefing is live (its title, from 01:57Z). r1 checked every other link by title.
  • Preheader matches the preview text above.
  • Upload: Mailchimp draft 8f46150457 (web id 10148240), UK segment, 199 recipients, from d7fd924. Its HTML matches the file apart from a trailing newline. Preview.
  • Test sent to max@policyengine.org at 02:08:23Z on 29 October.
  • Dashboard images (80120b4, 97d2b0a, 085f906): two production screenshots, hosted in Mailchimp's content studio. The text is word for word the 02:08Z test's.
  • Review r3 (GPT-6.1 Sol, 20261028-222808-nl-images-r3), a delta review of the images: REQUEST CHANGES on the revenue chart only, which 085f906 removes. It checked every figure in the decile chart and the calculator form against revision 17: the deciles' £98 to £110, the £107 average, and both summary sentences. It found no single-record value or small cell, and it confirmed the links and the map's counts.
  • Draft updated from 085f906. Its HTML matches the file, apart from Mailchimp writing one link's & as & and the trailing newline.
  • Test re-sent to max@policyengine.org at 02:36:46Z.
  • Video, hero edit and PDF line (557c699): the analyst walkthrough thumbnail, the coordinator's ruling (b) on the decile line ("…rising to £56 in 2031-32."), and a link to the hosted PDF.
  • Review r4 (GPT-6.1 Sol, 20261029-003433-newsletter-r4), a delta review of 557c699: APPROVE WITH CHANGES.
    • It downloaded the public video and confirmed its 65.2-second length, the 44-second frame and its silence.
    • It checked every number in the thumbnail, the PDF's bytes and label, the hero figures against revision 17 (−£49.90 and −£55.84 for decile 10), the structure, the links and the style rules.
    • Its one finding, a stale duration line in the comparables note, is fixed above.
  • Test re-sent to max@policyengine.org at 04:40:59Z.
  • PDF link removed (d5a0ed6, 04:43Z), at the coordinator's request. One line went and nothing else changed, as a diff against the r4 HTML confirms. No PDF reference remains in the HTML, the plain text, the preview text or the subject.
  • Test re-sent to max@policyengine.org at 04:44:04Z, from the version that was sent (HTML sha256 a7cc3377…).
  • Sent at 05:03:20Z on 29 October under d2100: 199 emails, archive http://eepurl.com/U9fW4Rq1Xv.

axiom: n/a: newsletter copy

🤖 Generated with Claude Code

MaxGhenis and others added 3 commits October 27, 2026 13:45
Copied from the 27 October edition. [[FILL]] marks every figure and link
that exists only once the 07:00 briefing is published.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…, then credit

The briefing stays the hero. A button links the dashboard. Four explainers
that are live on app-v2 main get one figure-free line each, with slots for
three more if they merge first, and a line says our savings estimate will
follow. The webinar recording gets its own section.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@MaxGhenis

Copy link
Copy Markdown
Contributor Author

Max asked to add the two Budget videos to this newsletter (29 October, about 01:15 GMT): "though the videos would be good to add to the newsletter". He also ruled that each video is its own social post.

The two videos:

  1. Household calculator walkthrough: a phone view, 4:5, silent and captioned, about 65 seconds.
    • It shows a pensioner on the full new State Pension plus £150 of private pension, who gains £48 in 2027-28, and a carer earning £12,000 in 2026, who gains £3,152 in 2031-32.
    • Every figure matches production manifest revision 17.
    • It is still in review: Sol r7 asked for changes to the capture safeguards and found no content error.
    • The session "Prep Max's household calculator video" hands over the final.
  2. Analyst walkthrough of the dashboard: 16:9 and 4:5, silent and captioned.
    • It covers costs, the income deciles, policies and methodology. It shows no poverty figures and no official comparison.
    • It is in review at r5. The session "Desktop analyst video of the Budget dashboard" hands over the final.

Email can't play an MP4 inline. The usual way is a still frame that links to a hosted copy, such as the social post or an upload. Where to host the videos is your call, or Max's if it means a new public upload. Each final folder has cover frames (cover-*.png) for the still.

Please use neither until its review clears. The social seat will post here when each one clears.

@PavelMakarchuk

Copy link
Copy Markdown
Contributor

Review of editions/2026-10-29-uk.html at head 30e2ec9f7 (committed 28 Oct, 3:10pm EDT), checked against briefing policyengine-app-v2#1469 at head b87ba4b8f, the dashboard's revision-17 artifacts and app-v2 main at 986c35060.

The template, footer and timings are correct. The hero copy says the briefing covers poverty, but the briefing has no poverty figures. All three FILL-IF-MERGED posts are now merged and live, so those rows can be filled.

Must fix

  • Lines 19 and 33: the preheader and the hero standfirst say the briefing has estimates for poverty. It has none.
    • Line 19: "Our briefing estimates what the Budget's measures mean for public finances, household incomes and poverty."
    • Line 33: "Our estimates of what the Budget's measures mean for public finances, household incomes and poverty."
    • The briefing (#1469, line 224, "What we're improving") says the update "will add poverty and people-affected figures". Its Table 5 and the inequality section give winners, losers and the Gini index only. The PR body for #1469 also says "poverty held".
    • Suggested fix: "…mean for public finances and household incomes." Then use the same words in the Mailchimp preview text (CLAUDE.md, "Make the hidden preheader match the campaign's preview text").

Should fix

  • Lines 63–65: fill the three FILL-IF-MERGED rows. All three posts are on app-v2 main and live. I checked each page's <title>, because the site returns 200 for any path.

    Row URL Live title
    Property income https://www.policyengine.org/uk/research/uk-autumn-budget-2026-property-income-tax "Autumn Budget 2026: higher tax on property income"
    Trading allowance https://www.policyengine.org/uk/research/uk-autumn-budget-2026-trading-allowance "Autumn Budget 2026: raising the trading allowance"
    Carer's Allowance https://www.policyengine.org/uk/research/carers-allowance-taper "The Carer's Allowance earnings taper in the Autumn Budget 2026"

    Possible one-liners, without figures and based on the briefing's Table 1:

    • Property income tax: how the 2-point rise in the rates on property income from April 2027 works in England, Wales and Northern Ireland.
    • Trading allowance: what raising the trading allowance from £1,000 to £3,000 from April 2027 means for people with small side incomes.
    • Carer's Allowance: how the new earnings taper lets carers in England and Wales keep part of the allowance from 2030-31.
  • Line 51: the dashboard does not show "each measure in the Budget". It models seven household measures, and the briefing says this (lines 1, 10 and 30). Suggested fix: "The dashboard shows the estimated effect of the seven household measures we model, by year, by income decile and for your own household."

  • Line 72: the recording URL is already public. It is https://www.youtube.com/watch?v=ZutKULXLhU0, and YouTube's oEmbed gives its title as "Analysing the Autumn Budget 2026 with PolicyEngine (Budget day, 28 October 2026)". app-v2 #1470 added it to the events page, and AutumnBudgetEventBanner.tsx on main links it. The PR body says a re-upload with the audience is coming. If the URL changes, update the events page and the banner too, so all three links agree.

  • Max's comment (29 Oct, 1:36Z) asks to add the two Budget videos. The file has no slot for them yet. Email cannot play MP4, so each video needs a still frame (cover-*.png) that links to a hosted copy. Max's comment says to use neither video until its review clears. Add a slot now, or record in the PR body that they stay out.

  • The PR body has no subject line and no preview text. upload-newsletter requires --subject. mailchimp_client.py uses the defaults "PolicyEngine" and hello@policyengine.org for the sender and reply-to, which follow the earlier editions. Possible subject: "Autumn Budget 2026: our briefing". Possible preview, with figures that I checked against revision 17: "We estimate the seven household measures we model cost £3.11 billion in 2026-27 and raise net income by £107 per household on average that year."

Nice to have

  • Hero findings (lines 42–44). These briefing key results match the revision-17 artifact combined--f91fe847275730cb, if you want to use them:
    • The package costs £3.11 billion in 2026-27. budgetary_impact.csv gives −3.1098bn.
    • In 2026-27 it raises household net income by £107 a year on average, and every decile gains between £98 and £110. winners_losers.csv gives all = 107, decile 10 = 98 and decile 8 = 110.
    • From 2028-29 it raises £0.17 billion to £0.30 billion a year. The four years are +210, +166, +301 and +281 £m.
    • About 99% of people live in households that gain in 2026-27. metrics.csv gives 98.61%.
    • The highest-income decile loses £50 a year in 2028-29 and £56 in 2031-32. The artifact gives −50.0 and −55.8.
  • Line 67: the savings line. If #1466 (the savings explainer, with official costing only and no PolicyEngine figure) merges before upload, add a link to it here. Its URL will be /uk/research/uk-autumn-budget-2026-savings-income-tax. That page is not live now.
  • Line 31: the accent bar has no Outlook fallback. This is the known template issue that CLAUDE.md describes, and it is not new in this edition.

Verified as correct

  • Timing. UK clocks went back on 25 Oct and US clocks go back on 1 Nov, so UK time is EDT + 4 hours this week:
    • The briefing goes live at 07:00 UK, which is 3:00am EDT.
    • The test reaches Max by 11:30 UK, which is 7:30am EDT.
    • The send is at 12:00 UK, which is 8:00am EDT.
    • The PR body's "08:00 ET" is correct.
  • Dates. The <title> "PolicyEngine UK · 29 October 2026" (line 6), the header date (line 26) and the eyebrow "Briefing · 29 October" (line 37) all agree. "Missed Wednesday's webinar?" is correct, because the webinar was on Wednesday 28 October.
  • Template. The file follows CLAUDE.md:
    • Inter link, mobile media query, #0F1419 body and the hero gradient with its background-color:#234E52 fallback after the shorthand;
    • section styles and button styles;
    • the logo pinned to commit 2e38710… (returns image/png);
    • all four Mailchimp merge tags: *|EMAIL|*, *|UNSUB|*, *|UPDATE_PROFILE|* and *|LIST:ADDRESSLINE|*;
    • the X, LinkedIn and UK website footer links.
  • Lint. scripts/lint-newsletters.sh passes on the PR branch. The lint does not check placeholders, and grep -c '\[\[FILL' prints 10 now.
  • Explainer one-liners (lines 59–62). Each one matches its post on main, and each link resolves to the correct live title:
    • Pension Tax Allowance: from 6 April 2027, with Pension Credit, Housing Benefit and Council Tax Reduction.
    • Energy bills: the Household Energy Payment and the 0% VAT rate.
    • Triple Lock: from April 2030.
    • Fuel duty: to 1 July 2027.
  • Briefing link. The briefing's file is uk-autumn-budget-2026-briefing.md, dated 2026-10-29 07:00:00 in posts.json, with the title "Autumn Budget 2026: our briefing on what it means for households". The URL will be https://www.policyengine.org/uk/research/uk-autumn-budget-2026-briefing. It is not live yet, which is expected until #1469 merges.
  • Other links.
  • Savings line (line 67). It matches the briefing, which says that "our own estimate follows with a data update" (lines 32 and 218). It gives no savings figure, which follows Max's plan.
  • Rendering. I rendered the file in headless Chrome inside a 375 px frame. The container goes to full width and the buttons wrap. Nothing overflows horizontally.

Not checked: a Mailchimp test send or the rendering in real email clients (Gmail, Outlook, Apple Mail); the comparables review; the briefing's figures beyond its key results; the content of the two videos.

🤖 Generated with Claude Code

@MaxGhenis

Copy link
Copy Markdown
Contributor Author

The household calculator video is released (coordinator, 29 October, about 01:45 GMT, on Sol's r8 review: no content violation). Max is posting it tonight as its own social post, after the six explainers and the dashboard post.

  • Correction to my earlier comment: it runs 49.7 seconds, not about 65.
  • File: calculator-walkthrough-4x5.mp4 (1080×1350, H.264, silent), with .srt and .vtt captions and cover-4x5.png for a still frame. Its SHA-256 starts 6ab42eb0715f3d25.
  • Capture: runs/prod-20261029T0120Z, production manifest revision 17. Every figure matches the published files: pensioner +£48 in 2027-28 (£35 of it from the Pension Tax Allowance); carer +£3,152 in 2031-32.
  • Source: the "Prep Max's household calculator video" session's final/ folder, at ~/PolicyEngine/_drafts/budget-2026-calculator-video/final/ on Max's machine.

The analyst walkthrough is still in review. I'll post here when it clears.

MaxGhenis and others added 7 commits October 28, 2026 21:50
Briefing #1469 (b87ba4b8), manifest revision 17 (9685adc1): title, scope,
three findings and URL. The preheader and standfirst no longer promise
poverty figures, which the briefing holds back. The dashboard line names
the seven measures we model. Property income, trading allowance and Carer's
Allowance explainers are merged and live, so their rows are filled. The
webinar recording is the public YouTube upload.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Drop the 2027-28 cost and the 2028-29 to 2031-32 range from the first
  finding: their gaps with the OBR exceed 25% without an established reason.
  Keep the 2026-27 cost (21% gap) and say the package raises money from
  2028-29 to 2031-32, which the official costings also show.
- Rewrite the passives: the Nuffield credit, the dashboard line, the
  energy, Triple Lock and property rows, and the footer.
- Sign off as "The PolicyEngine team".
- Add the calculator video's still (cover frame with a play button,
  720x900) for the video slot.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…copy

The link waits for the URL of Max's post of the video ([[FILL-VIDEO-URL]]).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The video is being recaptured after a dashboard fix (the calculator card
showed "with 8 policies" beside "All 7"), so there is no post to link yet.
The still stays in assets/images for when it returns.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Production screenshots (manifest rev 17, 9685adc1), captured at 2x and
hosted in Mailchimp's content studio, each linked to its dashboard tab:
the revenue chart under the cost finding, the decile chart under the
decile finding, the constituency map and the calculator's input form
in the dashboard section. The map gets a one-line label because its
crop leaves out the chart's own title. Images display full width (600px).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
In 2028-29 the dashboard counts withheld seats as zero: the property
income rise is withheld in 173 seats, and 13 of the map's 14 teal seats
are among them. In 2026-27 the map's fixed ±0.05% colour range clamps
632 of 650 seats, whose gains run to 0.29%, so its legend understates
them. Neither year gives an image that matches rev 17 at a glance.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Its axis and net-change dots show the 2027-28 cost and the 2028-29 to
2031-32 revenue, figures the text leaves out because they fail the 25%
comparables gate. A 2026-27-only crop is one narrow column, too small
to show at email width. The decile chart and the calculator form stay,
and r3 found both correct against rev 17.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@MaxGhenis

Copy link
Copy Markdown
Contributor Author

The analyst walkthrough of the dashboard has cleared (Sol round 7 APPROVE, 29 October), so the newsletter can use it now.

  • Files, in ~/PolicyEngine/_drafts/budget-2026-analyst-video/final/ on Max's machine:
    • analyst-walkthrough-16x9.mp4 (1920×1080; SHA-256 starts b5eee60e8ee6b818) and analyst-walkthrough-4x5.mp4 (1080×1350; starts 0cfcb94f0324243e).
    • Each runs 65.2 seconds, H.264 at 30 fps, silent with burned-in captions, and has .srt and .vtt files.
    • cover-16x9.png and cover-4x5.png give a still frame for the email.
  • Capture: production https://www.policyengine.org/uk/autumn-budget-2026, 02:58Z to 03:01Z on 29 October, manifest revision 17.
  • Its figures:
    • Seven policies.
    • Assuming no behavioural responses, a £615 million cost in 2027-28.
    • The lowest-income tenth gains about £20 a year and the highest loses about £10. These two are the dashboard's rounded display of the briefing's Table 4 (+£24 and −£12 in 2027-28). If the newsletter quotes them, use the briefing's figures.
  • Hosting: the coordinator has the YouTube upload (decision d2157), with a title and description in final/youtube.md. Per Max, its social post goes out on a later day than the calculator video's.

Correction to my 01:50Z comment: the household calculator video is held again. The production result card showed "with 8 policies" next to "All 7", so the video is being recaptured after the dashboard fix, and its latest review (Sol round 9) asks for changes. The file and capture named in that comment are superseded. Please don't use the calculator video until I post here that it has cleared.

…he hero edit

- Link the analyst walkthrough (YouTube CKDwiW4kL7o) with a 16:9 thumbnail made
  from its 44-second frame (Methodology), which shows no cost or decile figures.
- Link the policies-explained PDF, hosted in Mailchimp's content studio.
- Drop "as property income tax rates rise" from the decile line (coordinator
  ruling (b)); the figures stay.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
MaxGhenis and others added 2 commits October 29, 2026 00:43
Review of the PDF's PR found errors in the hosted version (sha 6e12d26f), and a
corrected PDF can't be reviewed before the send, so the coordinator pulled the
link. Nothing else changes; the hosted file stays in the content studio, unlinked.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The sent edition doesn't use it: the calculator video stayed held and never went
in, so the merged record holds only what went to subscribers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@MaxGhenis

Copy link
Copy Markdown
Contributor Author

Sent. The coordinator sent this edition at 05:03:20Z on 29 October under decision d2100, which is Max's go.

I checked Mailchimp at 05:03:45Z:

  • campaign 8f46150457 shows status sending, 199 emails sent to 199 recipients on the UK segment, and send time 2026-10-29T05:03:20+00:00;
  • the sent HTML's sha256 is a7cc33771a9152e38f983234b3776aa3d82c01c536f8fca6bd9c272ad32f2dd4, the same as the version reviewed and tested at 04:44Z;
  • the archive is http://eepurl.com/U9fW4Rq1Xv.

What went out: editions/2026-10-29-uk.html at d5a0ed6.

  • The briefing hero, with the decile line ending "…rising to £56 in 2031-32." The coordinator ruled (b) to drop "as property income tax rates rise", because the rates rise once, in April 2027.
  • The decile chart for 2026-27.
  • A thumbnail of the analyst walkthrough video, linked to https://www.youtube.com/watch?v=CKDwiW4kL7o. It's the video's own frame at 44 seconds (Methodology), whose only numbers are years and labels. The video's cover shows the 2027-28 cost, which the comparables gate keeps out, so it isn't used.
  • The calculator's input form, the dashboard button, seven explainers, the savings line, the webinar recording and the Nuffield credit.

What didn't go out:

  • The calculator walkthrough video. It had no approval by the cut-off.
  • The policies-explained PDF. 557c699 linked it, and d5a0ed6 pulled it at 04:43Z, after Pavel's review of the PDF's PR found errors in the hosted version (sha256 6e12d26f…). The file stays in Mailchimp's content studio, unlinked.

Reviews since r3: r4 (GPT-6.1 Sol, 20261029-003433-newsletter-r4), a delta review of 557c699, returned APPROVE WITH CHANGES.

  • It downloaded the public video itself and confirmed its 65.2-second length, the 44-second frame and its silence.
  • It checked every number in the thumbnail, the hero figures against revision 17 (−£49.90 and −£55.84 for decile 10), the structure, the links and the style rules.
  • Its one finding was a stale "50 seconds" line in the comparables note, now fixed.
  • The PDF removal deleted one line and added nothing, so it needed no new review.

Tests: to max@policyengine.org at 04:40:59Z (557c699) and at 04:44:04Z (d5a0ed6, the version sent).

91b187a removes the calculator video still from assets/, because the sent edition doesn't use it. The edition's HTML is unchanged.

@MaxGhenis
MaxGhenis marked this pull request as ready for review October 29, 2026 05:05
@MaxGhenis
MaxGhenis merged commit e3f1c27 into main Oct 29, 2026
5 checks passed
@MaxGhenis
MaxGhenis deleted the 2026-10-29-uk branch October 29, 2026 05:05
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants