Revise Pilot A per external pilot-design panel#91
Conversation
Apply a 16-point review from a panel of pilot specialists (impact-eval
methodologist, cash-transfer practitioner, sequencing strategist,
human-outcomes/dignity expert, ethics/credibility skeptic) to the
Minimal Pilot Proposal:
- Lead with mechanism integrity ("the walls hold"); state power/MDE
honestly; treat welfare as estimation, not pass/fail.
- Replace the deliberate real-goods scarcity cut with a SIMULATED
scarcity exercise (medicine excluded), with an audit-before-ration gate.
- Add a cash-control arm; household randomization; leakage-detection
validity; baseline spillover mapping.
- Validated welfare instruments (USDA HFSSM, CFPB, K6, Pearlin) plus a
mixed-methods qualitative arm, under explicit measurement-dignity rules.
- Funder-selected/paid independent reviewer; pre-registered, locked
thresholds; AI's design role disclosed up front.
- Honest "additive" promise backed by a hold-harmless fund; control-arm
ethics (post-trial access, compensation, attrition handling).
- Keystone-first plus a whole-system tabletop (folded into Pilot B);
resequencing (Voice parallel, Service Record gated on Voice);
asymmetric stop-gates.
- Honest budget (~$700k–1.4M, ~27-month program).
Adds the panel findings memo. Regenerates corpus (114 docs);
validate_corpus passes with 0 errors.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Do not claim the pilots prove real-scarcity behavior
After this revision switches Pilot A’s scarcity work to a paper simulation, the summary still says the four pilots would teach whether rationing works “under real scarcity” without black markets. That overstates the evidence this design can produce: the simulated exercise can test declaration and rule conformance, but it explicitly gives up the behavioral signal of a live shortage. Revise this summary so funders and reviewers do not treat the simulated shortage as proof of real-scarcity performance.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| If you are a **researcher**, the open questions sections are the research agenda. Each one is a question the pilot cannot answer on its own. | ||
|
|
||
| If you are a **funder**, the four pilot budgets total roughly **$790,000–$1,780,000** for an 18–24 month program (Pilot A $350k–860k; Pilot B $260k–570k; Pilot C $75k–145k; Pilot D $105k–205k). That is a small number for the question being asked. For comparison, the Finland UBI pilot (2017–2018) cost approximately €20 million. The single most important pilot, and the place to start, is **Pilot A** — it tests the survival floor and the non-convertibility wall that the entire design rests on. | ||
| If you are a **funder**, the four pilot budgets total roughly **$1.14M–$2.35M** (Pilot A ~$700k–1.4M; Pilot B $260k–570k; Pilot C $75k–145k; Pilot D $105k–205k). That is still a small number for the question being asked — for comparison, the Finland UBI pilot (2017–2018) cost roughly €20 million — and the figure is higher than an earlier draft's precisely because it now funds the staffing, benefits protection, a cash-control arm, and the pre- and post-pilot months a real, credible run requires. The single most important pilot, and the place to start, is **Pilot A** — it tests the survival floor and the non-convertibility wall the entire design rests on. Whichever pilots run, the whole-system question is answered first and cheaply by the tabletop (under Pilot B), not by an expensive bundled flagship. |
There was a problem hiding this comment.
This new funder budget makes the entry-point guidance in docs/public/00_start_here.md:98 stale: it still tells funders the prospectus is in progress and that there is “no detailed ask or budget to quote yet.” Because 00_start_here.md is part of the public corpus and explicitly routes funders/partners, readers can follow the main path and miss or distrust the newly published ask here; update that section/link so the public guidance matches this revised proposal.
Useful? React with 👍 / 👎.
1 participant
What
Applies a 16-point review of the Minimal Pilot Proposal (
docs/public/09_pilot_proposal.md) from a five-specialist panel — impact-evaluation methodologist, cash-transfer/in-kind practitioner, program-sequencing strategist, human-outcomes & dignity measurement expert, and an adversarial research-ethics/credibility reviewer. Each ran as a single-role reviewer; the full decision memo is added atdocs/superpowers/specs/2026-06-15-pilot-panel-findings.md.Key changes to Pilot A
Checks
npm run generate:corpus→ 114 documentsnpm run check:corpus→ in sync (exit 0)python3 scripts/validate_corpus.py→ 0 errorsThe companion funder-site changes (headline number, simulated-scarcity language, three arms, a "What changes for the people" section, rebuilt PDF) are in #90.
🤖 Generated with Claude Code