Verify renormalisation-census-2026
Every number in this study was computed from the frozen package described below. The digests pin the exact bytes; the command reproduces them. Nothing here is a summary of the artefact — it is the artefact's fingerprint.
Package stamp
| Study | renormalisation-census-2026 |
| As of | 2026-07-28T19:30:27.531Z |
| Environment | production |
| Methodology | 5.1.0 |
| Population | published |
| Repository commit | d94ad2c2ee4d5b294590ea0cdd55cc6346377d6d |
| Generator sha256 | e84165ed460cd4acf8cc63ef734df065f30bb802a214d79b9337a3b4042c13ce |
Files
| File | sha256 | Bytes | Rows |
|---|---|---|---|
| census.json | f1ff035b7b1b0e9d7ca01b87efde095ebc321b65127fabbb9f0ec8501cb67683 | 21490 | — |
| census-cells.csv | f91551324460551a18850b331e626e7f37072247f6c4b0e1e178a454d5ad3e2e | 1608 | 28 |
| query.sql | cf9687101f170b5d5bd4a473b0110b00d98864079746951cee2f2fbdfde14fd6 | 1281 | — |
| README.md | 359c7ae9b9460cd46735f5c62a6feceae9651fdfa6e251736b048c54ff390031 | 298 | — |
4 file(s), 24677 bytes in total. The package lives at docs/research/renormalisation-census-2026/ in the IndexFair repository, which is not public. The digests above therefore pin the bytes for anyone we hand the package to — they are not, on their own, something a stranger can independently reproduce today.
Check it yourself
Integrity verification is offline: given the package, the check below needs no database, no network and no credentials, and it is the same check our CI runs on every commit. Because the repository is not public, ask us for the package and we will send it, with these digests, to anyone checking a number before a deadline — contact.
pnpm --filter @bdr/workers exec tsx \ scripts/research/build-study-package.ts renormalisation-census-2026
If you hold the package but not a repository checkout, you do not need our script at all — digest the files yourself and compare against the table above:
openssl dgst -sha256 *
The script fails closed: a missing file, a changed byte, a truncated row or a manifest that disagrees with the bytes beside it all stop the run rather than degrade it. If your copy verifies and your recomputation still disagrees with the published number, that is a finding about us and we want it — tell us. Factual correction is free, for everyone, always.
The study
This receipt describes the data behind /research/renormalisation-census-2026, which carries the method, the denominators and the limitations. A digest proves which bytes were used; it does not make the inference from them correct. Read both.