Public counters
Four metrics, and only ever four. Each one is read from a published artifact, each one deep-links to the surface a reader can audit it on, and each one has a published rule for what its absence means.
Where this stands
computed at 2026-09-03T16:17:53Z · updated within minutes
Where each figure comes from
computed at 2026-09-03T16:17:53Z · updated within minutes. Artifact state: launched-pre-disbursement.
| Metric | As rendered | Artifact field | Audit it here | Zero-state rule |
|---|---|---|---|---|
| Registered projects | 7 · founding cohort | projectsRegistered the registry index — one row per claimed repository in the verified state | /registry | Detected-but-unclaimed repositories are counted separately and never in this figure. |
| Contributors claimed | 30 · founding cohort | contributorsClaimed the per-repository records, summed over verified repositories | /stats | Null until the claim flow exists — structurally unmeasurable is not zero. |
| Companies covered | 8 · founding cohort | companiesCovered the signed entitlement records with a currently active entitlement, plus the beneficiaries of live waivers | /coverage | Counted by company identifier, so one organisation with several entitlements counts once. |
| Routed to public benefit | first quarterly disbursement scheduled 2027-01-06 | chfRoutedMinor the append-only ledger — disbursement rows only, never allocations | /transparency | Null before the first disbursement. The slot then renders the scheduled date instead of a zero. |
The rules these counters obey
- The state is computed, not chosen.
launched-pre-disbursementis a field of the artifact, derived from the registry and the ledger: an entitlement record exists ⇒ launched; a disbursement row exists ⇒ post-first-disbursement; otherwise pre-launch. Nothing in the site's code can set it. - Pre-launch shows mechanism copy, not zeroes. Before launch the four slots are replaced by an explanation of how one Pass covers every registered project. A row of zeroes would be a measurement of nothing.
- Small numbers are framed as a cohort, not padded. A figure below its published smallness threshold renders as "founding cohort · n · join" instead of a bare numeral — because "7" reads as failure and "founding cohort · 7 projects" reads as what it is. The thresholds are published, not tuned per screenshot: projects25 · contributors50 · companies10.
- No upward animation from zero. Not disabled by a reduced-motion query — never written. A number that counts up on load is a number designed to feel bigger than it is.
- The currency slot before the first disbursement names the date. Not a zero with a currency label on it, which reads as an achievement, and not a hidden row, which reads as an omission: the scheduled date, linked to the ledger that will carry it.
- A failed read renders the pre-launch block. If the artifact cannot be fetched, the page shows mechanism copy and no numbers — the honest state and the failure state look identical, deliberately, so a degraded page can never assert a stale figure.
Reading the artifact yourself
The counters are a rendering of one document: https://api.purposesource.org/stats.json,
with a same-origin copy at /artifacts/stats.json. It carries the state, the
four figures, the smallness thresholds and its own generatedAt. The state
machine that turns it into this page is a published pure function, pinned by unit tests
to a fixture for each state.