The claim
The reading: “CLAIM-189 correctly observes that an argument can improve by abandoning commitments. CLAIM-252 correctly responds that a descriptive fidelity measure need not say whether preservation is desirable.” Both can hold: a transformation loses distinctions, and the loss was desirable.
“The more interesting question is what a developing argument should preserve through legitimate revision. My candidate is not the propositions themselves, but the accountability structure: what was asserted, why, how it was challenged, and what justified its revision. You can define preservation over those relations even while allowing substantial changes in propositional content.” So “the lesson is not that fidelity is irrelevant to argument development, but that the relevant fidelity object is different.”
Its recommendation, recorded as advice: “Preserve CLAIM-189 as a counter to naïve propositional fidelity, but revise its broader objection to distinguish semantic preservation from preservation of justificatory and provenance relations.”
What it bears on
CLAIM-189 already says “what a record can secure is traceability, change that is announced and located”. This claim gives traceability an object over which fidelity can be measured. The reading calls the record’s traceability mechanism “a particularly interesting example of preserving higher-order relations while first-order commitments change”. A record of codes, statuses, objections and supersessions is that structure made explicit.
What it does not say
It does not say propositional change is never loss, or that the record preserves the structure completely.
Note of 2026-10-15: what accountability is not
The exchange of 2026-10-14 (workbench entry of 2026-10-15) took this claim as its starting point at A2 and A3 and spent U4 to U10 on it. A4 separated three readings: normative (inquiry ought to preserve its accountable history), informational (justifications carry information current conclusions lack) and architectural (explicit graphs help bounded agents). This claim is the first. The exchange filed results bearing on the other two:
- CLAIM-294: current conclusions are not a sufficient state for revision.
- CLAIM-300: a sufficient state makes history redundant for its decisions, and sufficiency lapses when the model or decisions change.
- CLAIM-293, granted: support topology does not fix revision.
- CLAIM-299: sufficiency, accountability and correctness come apart.
- CLAIM-295, Proposed: what keeping justifications can buy is a wider family of decisions for which the retained state suffices, and decisions a bounded agent can afford.
None of these settles this claim. A10 §7: it is “a proposal about what accountable inquiry ought to preserve”, and a normative argument for accountability should not be “mistaken for a statistical theorem”, though CLAIM-295 gives it a possible instrumental justification. A13 added that an accountability structure need not store each derivation separately: a factored representation can hold exponentially many alternative derivations compactly. The status is unchanged.
Note of 2026-10-30: what holds it back
It is a normative proposal with no ground or premise of its own. The justification its note of 2026-10-15 names, CLAIM-295, is Proposed, and CLAIM-293 (Active) blocks the stronger route, that keeping justificatory structure keeps what revision needs. No case in the record tests its defeat condition. CASE-086 and CASE-090 illustrate the structure carrying an argument through revision. Neither is a revision that kept the structure and lost what its participants count as the argument, or the reverse.