Aevyra (2026). The Ledger of Will: Evidence, Revaluation, and the Preservation of Revisable Agency. Aevyra Papers & Docs. https://aevyra.github.io/ledger-of-will/.
Read in full on 2026-10-22, in NOTE-1091. The summary is the abstract’s and the reading confirms it. The page credits “Original formulation: Yue and Anika · Research revision: Sofia · Review and revision-closure proposal: Anika”.
Key takeaways
From the reading, NOTE-1091. Section numbers are the paper’s.
- The requirement. “A record that reconstructs yesterday’s answer perfectly can nevertheless prevent tomorrow’s appropriate revision if it has retained the conclusion and discarded its conditions” (§1).
- Sufficiency is relative to a declared family. A retained state is exactly sufficient iff every target is constant on its fibres (Prop. 2), and “Declaring a restricted demand family is therefore part of the scientific claim, not administrative metadata” (§4.2).
- Under repeated revision, the fibres must lie in the coarsest revision-invariant refinement, found with at most K* − K added revision words (Prop. 4); THEORY-884.
- Keep, narrow or reacquire. “Adding an instruction to reconsider cannot reconstruct a distinction that was irreversibly merged” (§4.3).
- Derived notes are not new evidence (§6.2), and “Information availability and computational accessibility are different resources” (§5.2).
- Proposals, not results. “this paper offers no new human-subject study, no large-model performance result, and no experimental demonstration of subjecthood” (§10).
Why this is here
A13 to A16 found it the closest outside predecessor of CLAIM-295 and QUESTION-112, and demoted an essay on the record’s memory for it. Its abstract says the contribution is “a testable synthesis of established ideas and explicit design proposals”. It carries anthology-candidate because its subject is memory for language-model agents, which the anthology may also want.
Corrected 2026-10-22, from the reading. The exchange’s judgement stands: it is the closest outside predecessor of CLAIM-295, whose informational, computational and option values it states in its own terms (§5.2, §5.4), and it answers QUESTION-112 for a revision family declared in advance (THEORY-884), not for one that changes with what is found. Its Proposition 2 is CLAIM-350. “Will” is the name of the six-paper series it belongs to; the paper does not use the word otherwise and has no counterpart of TERM-023’s will-organization. What it shares with the will line is the factorization criterion behind TERM-067 and a notion of authority over commitments (§7) near TERM-047’s jurisdiction.