datapce readout: How do I check it?

Every number has a label, a date and a source. This page shows how to check them.

4 · How to check it

How do I check it?

Every number has a label, a date and a source. A build gate refuses a number that is not on the list.

  1. A claim row holds an id, a value, a date, a source and a label.
  2. The source is a named file or tool. A private source says it is private.
  3. The label is one of MEASURED, ESTABLISHED, INFERRED, SYNTHETIC or PLANNED. Each has a rule.
  4. The gate checks links, markup, banned words, the number allow-list, labels and the prereg bytes. A failure stops the build. The honesty scorer gave H1 page R3 91.1 of 100 and R1 91.3.
  5. The page shows the number with its chip, date and source.
Fig. 7 — Every number goes through the same steps. The build gate and the unit tests (deploy.sh runs both) refuse a number that is not in the allow-list. The honesty scorer rated H1 page R3 at 91.1 of 100 and R1 at 91.3, so it shows no gain; one part of the score is a model's opinion. (score = points over the rules) MEASURED 2026-10-21 · tools/rsi_score.py; docs/rsi/METRICS.tsv. The example claim, H6, is from Fig. 3.

Check it yourself

  1. Open the network panel of your browser. This page loads only the files listed here, all from datapce.com.
  2. Open readout-v0.schema.json. Compare it with Fig. 2.
  3. Open data/claims.json. Every number on this page comes from it, or is a date, a figure number or a step number.
  4. Read prereg-h1.txt. It holds the rules of the H1 study as they stood before launch. The study record shows how it ended.
  5. For each number, find its chip, date and source. A private source says so.

index.html css/base.css css/kit.css css/home.css js/core.js js/home.js favicon.svg favicon-32.png apple-touch-icon.png

No cookie, no analytics, no third-party request, no form.

What this page does not claim

  • One maintainer, one set of machines. The numbers say nothing about how other people work.
  • The models in Fig. 5 are small research fits, and the JEPA failed its gate.
  • No route cell is promoted (Fig. 3).
  • The change-chain numbers are one repo; the after-hook count covers about 8 hours. They are not a benchmark, and the records are not signed.
  • H1 has no verdict. Persona panels and simulations are SYNTHETIC and never show demand.
  • The scan replay is one operator's logs. Temp-dir counts are a lower bound on false denies.

H2: launch test of datapce-scan PLANNED

H2 asks if one post makes people try the scan. The rules were fixed before any count was read.

  • One post by the owner. The window is 14 days.
  • PASS: 2 or more rule-set requests from others.
  • KILL: fewer than 15 stars and 0 rule-set requests.
  • Anything else is INCONCLUSIVE. Counts by the owner, by bots or by people the owner asked do not count.

No H2 result is read yet. A PASS would not show that anyone will pay.

PLANNED 2026-10-23 · LH2: docs/prereg-h2.md in the project repository (not served), sha256 4b0c0276e1db6a7cc5698bd1fecff7f8b830918080acae4f08e1d7c2e61dbdbc; decision D40