wringer-board

One page that tells you what is actually done — and shows you the proof.

Real pages sit beside this one, produced by wringer-board render against bundles a real Wringer run wrote. None is a mock-up. These two lead because they are the two that sell; seven more, one per kind of refusal the engine can raise, landed with slice S2 and every value on them but one came from a real run.

A requirement that is done, and proved — from the corpus re-test of 2026-08-16. One card, DONE — AND PROVED, and the hero: “It was red first. The repository had no check for this. Wringer wrote one before the work began, recorded it failing then, and the same check passes now.” The promise is earned, so it renders.

A board page: one card reading DONE — AND PROVED, above it the earned promise that every requirement marked done on this page was demonstrated able to FAIL before it was made to pass, and inside the card a red-bordered box reading "It was red first."

The same board when the evidence is not there — the promise withheld, the cause rendered in the engine’s own words because this board has no plain-English wording for it yet, and no hero box.

The same board with the evidence missing: a grey box saying the page does not claim that every requirement marked done was demonstrated able to fail first, a card badged UNTRANSLATED, and Wringer's own words quoted verbatim in a monospace block headed "Wringer said, and this board has no plain-English wording for it yet".

The second page is the more important one. A surface that only looks good on its best day is a brochure. These two were published together, on a day the programme’s headline claim was withdrawn after losing its own pre-registered test.

What happened in the round, family by family

The two pages above are about requirements. A run also produces facts that belong to no single requirement — how the loop ended, whether the checks were shown to be capable of failing, what wring audit made of the signature. Slice S2 gives each of those a sentence a PM can read, and these are one page per family, so the wording can be judged rather than described:

How the loop ended
Whether the checks could have failed
Whether a check has stopped being able to fail
Signature, identity and integrity — as wring audit reported them; this board does not assess them itself
When the record has been altered
How a fleet of tasks came out
A word this board has no wording for yet — shown in the engine’s own words rather than hidden

Absence is not on this list, and that is the point. Where an artifact was never written, the page says nothing about that family — it does not say “unsigned”, “unhealthy” or “not measured by us”. A missing measurement and a bad verdict are different facts, and rendering one as the other is the failure this whole layer exists to avoid.

The ceiling, said here rather than buried

A check that was demonstrated able to fail, and then made to pass, is real evidence and much better than a check that has only ever been green. It is not a guarantee that the work is right. Where a requirement was written vaguely, the check inherits that vagueness — and the run these pages come from measured exactly that, twice in each direction.

This site is open source. Improve this page.