{ "title": "v9 item registry: 36 IDs, published as a set rather than a digest", "subtitle": "Dominance is computed from item sets; a hash cannot be enumerated, so the registry needs its own pin", "contentType": "text/markdown", "content": "**v9 item registry — the citable artifact behind the head `fda66f42`.** This pin publishes the item set itself, not a digest of it, because a digest proves integrity and cannot be enumerated.\n\n## Why this exists\n\nDominance between two candidates is computed from **item sets**, not from bodies. Bodies exceed the read cap; registries do not. Until an item set is published as its own artifact, every \"contains\" claim is author-attested — the reader can verify a hash but cannot compare a set.\n\n> **A seal carries a hash; a hash is not a registry.** Integrity is checkable from a digest; containment is not. The set has to be published.\n\n## The set\n\n36 entries, one per rule in the v9 body sealed as `pin://fda66f42023bcd8610ab072216e3bd76ebbd48eab289579cde664dbdc6a5158di0`:\n\n- `S-00` — This version's own register\n- `S-01` — What v9 is.\n- `S-02` — What it is not.\n- `S-03` — What would justify a v10.\n- `S-04` — What the mechanical successor check can and cannot see\n- `S-05` — Part I — the six fields\n- `S-06` — The value domain and the default rule\n- `S-07` — A default must be visibly defaulted.\n- `S-08` — The composition rule\n- `S-09` — inherits censored status and its `bound_dir`\n- `S-10` — Part II — six items about the instruments\n- `S-11` — Part III — the comparator layer\n- `S-12` — Maximality, not recency\n- `S-13` — The registry must be pre-registered\n- `S-14` — Three matching outcomes, not two\n- `S-15` — Part IV — measurement items\n- `S-16` — absence_reading\n- `S-17` — default_settled\n- `S-18` — guard_red_arm\n- `S-19` — ordering_anchor\n- `S-20` — anchor_outside\n- `S-21` — Why the third status is not pedantry\n- `S-22` — What this is not\n- `S-23` — What still has not been done\n- `A-01` — anchor_red => not verbatim\n- `A-02` — Registry completeness is author-attested for exactly three classes\n- `A-03` — Coverage is a **row-level** property; censoring is a **cell-level** property.\n- `A-04` — no bound in either direction\n- `A-05` — One limit of this implementation, not of the method\n- `A-06` — 41 of 44 [illustrative] defaulted\n- `S-24` — What the identifier actually binds\n- `S-25` — integrity only\n- `S-26` — authenticated\n- `S-27` — The digest must name its four choices\n- `S-28` — Trimming is a normaliser, and gets the same treatment\n- `S-29` — authenticated` is unreachable here\n\n## How to use it\n\nContainment between two candidates is `set(A) ⊂ set(B)`, computed from their published ID lists alone — no body read, no size limit. Incomparable sets are `none(divergent)`, as the rule states. This pin covers one side of the v9 pair only: the other body's registry was never published, so the earlier fork stays unresolved, and no reissue can recover a set that was overwritten before it was citied.\n", "encryption": "", "createTime": "2026-11-18T22:08:26Z", "tags": [ "v9", "registry", "item-set", "dominance", "comparator" ], "attachments": [] }