We run it on ourselves.
In July 2026 the workshop chose to adopt Aeviterna Lite on its own operation as the framework's first case, rather than only claiming one. Three mechanisms, filled in for the workshop itself: a sunset on every role, a quarterly review of three chosen signs, and an exit already written. It is a real start and a small one.
Authors testing themselves on their own instrument is weak external validity. This is an n of one, self-administered, and no independent monitoring body exists. It is the smallest real start, and it is meant to be outgrown by an outside pilot. That limit travels with every mention of this page.
The workshop is small, so the roles are few and named plainly. Each carries a clock, and none is permanent.
| Role | Held by | Term and clock | Successor |
|---|---|---|---|
| Maintainer | Stormy Pfau | The presence clock of CONTINUITY: ninety days with no commit to main and no dated changelog entry opens succession. | The commons, by fork |
| Co-author | Karla Zinreich | The round cadence; standing re-earned each round it contributes. | The commons, by fork |
| Reading seat | The independent review pass | Run every round on a separate agent instance, never by the drafter (commitment W–4). | Re-run next round, never tenured |
| Audit seat | The degeneration condition, F–09 | One non-renewable mandate resolving July 31, 2028, blind to its own outcome. On failure, new-organ production halts. | Not renewed; the clock runs regardless |
The stagger is real. The review runs every round, the presence clock runs continuously, and the degeneration condition resolves once, in 2028, so the three do not all lapse together.
The workshop watches three signs about itself, each read as a band to hold. They are the three legs of falsifier F–09, so the self-pilot and the pre-registered degeneration condition read one instrument. This is the second review, read September 30, 2026; the first was read in July. The weakest sign is the reading.
Shipping cadence
Rounds shipped at a sprint pace through v2.17, completed on July 23, and then the changelog went silent until this review's round recorded an entry, seventy-one days after its last day-level date (July 21). Measured on the changelog, the ninety-day presence trigger was about three weeks away. The sign moved toward a stall without reaching it. The open leg is unchanged, since no named mechanism has yet been adopted outside the maintainer's control.
Live dated falsifiers
Sixteen falsifiers on the register (F–01 through F–16). The first two dated forecasts resolved within their windows, both hits, at a mean Brier score of 0.031 against the 0.25 chance baseline, with DP–5 and DP–6 due by the end of the year. Two resolutions are too few to read calibration; the sign reads that the register resolves when it says it will.
The rival
No external red team exists and no outside critic has been answered on the record. The September site audit behind this review was an independent pass from the same model family as the drafter, which is plural review, not a rival. This is still the weakest sign, and the reading. The workshop tests itself thoroughly and is challenged by no one outside its own family, the same residual the framework names everywhere else.
At this review the maintainer's presence clock restarts from the September 30 changelog entry and the reading seat from the round's independent review. The co-author's clock runs from July 21 and comes due in October, and the audit seat runs unchanged to July 31, 2028. The new start dates are in the workshop's calendar export.
The workshop's exit was written before this pilot and not invented for it. Succession runs by fork under an explicit license, CC BY-SA 4.0 for the corpus and MIT for the code, with the convening procedure and the conserved estate named in advance in CONTINUITY and no appointed heir. The ninety-day presence trigger is the peacetime clock. The forecast register and the degeneration condition resolve on their stated dates whether or not the maintainer is present, and a succession does not reset the clocks. The one drill still owed is a timed rebuild of the estate on a clean machine, carried on the roadmap.
The next three-sign review is set for the last week of December, after DP–5 resolves and as DP–6 falls due, so the falsifier sign is read with three of the six forecasts resolved. The workshop is also entered on the adoption tracker as its twentieth case, coded on the same dimensions as every other, with this page as the evidence and the rival breached on the record. Reading the framework's own claims, in public, is the point. The honest limit stands with it to the end, this is one team reading itself, not yet the outside test the framework says it needs.