The case
The nucleating-argument line proposes two measures.
- Inbound references per code. QUESTION-038: “A measure would also help. Inbound references per code tell an anchor from a structure that only persists.” A code that later entries cite is being built on; one that persists and is not cited is only kept.
- Turns per filed entry. CLAIM-186: “In a reconstructed exchange, count the turns in a stretch against the entries filed for it, and set that against the re-derivations and unannounced changes of thesis the workbench records for the same stretch.” Its defeat condition names the ratio “turns per anchor”.
Both are counted in the record’s grammar: codes, the relations between them, and statuses. The drift they are set against is also read from the record, as re-derivations and unannounced changes of thesis the workbench records.
What is held fixed: the exchange, and the record kept of it. What varies across stretches is pace and filing.
What it can show
If computed, whether the record’s own counts move with what the record calls drift: whether cited codes are the ones later structure rests on (CLAIM-185), and whether stretches with more turns per entry hold more re-derivations (CLAIM-186).
What it cannot show
It has not been computed. And both sides of each comparison are read in the grammar that shapes the argument (CLAIM-181). A rise in coding and linking registers as more anchoring whether or not the participants, by their own standards, drifted less. That is the threat CLAIM-190 named, and it is the failing critical question of ARG-028. A drift judgement made outside the record, blind, on stretches held out from it, is not part of this design.
Filed 2026-10-28 by the ADR-061 migration, as the premise CLAIM-190’s argument needed.