Incomex AI Portal
KB-2A47

RS5B-CLOSEOUT 03 — Supersession Map (PATCH1 + PATCH2) — 2026-06-21

7 min read Revision 1
rs5b-closeoutsupersession-mappatch1patch2canonical-operationbi-domain2026-06-21

RS5B-CLOSEOUT 03 — Supersession Map (RS5B → PATCH1 → PATCH2) — 2026-06-21

Purpose: record every corrected wording so the consolidated contract (file 02) is unambiguous about which text controls. Supersessions are wording/pointer changes carried additively in the patch files; the original rev-1 source files are not mutated. REGISTRATION_HOLD retained · 0 mutations.

Columns: Old text / source → Problem → Superseded by → New controlling text → Status.

1. Required supersessions (the four controlling ones)

S1 — RS5B-05 item-5 "kept out of the authorization binding"

  • Old text / source: rs5b/05-authorization-packet-requirements-before-any-write item 5 (rev 1): "the business-only effect digest … kept out of the authorization binding (separation)".
  • Problem: inverts the correct separation. It would leave effect_identity outside the binding, so authorization could not be bound to the exact effect — contradicting the accepted RS4A-PATCH2-02 direction and RS5B-06's own reject code APPROVAL_NOT_BOUND_TO_EFFECT_IDENTITY.
  • Superseded by: RS5B-PATCH1 (rs5b-patch1/02 §1, §4–§5; recorded in rs5b-patch1/03 §2–§3). Marker SUPERSEDED_BY_RS5B_PATCH1.
  • New controlling text: "effect_identity is kept pure (authority excluded from it) — it is NOT kept out of the authorization binding. The authorization_binding_digest binds the pure effect_identity to the authorization envelope." The real separation is authority-out-of-effect_identity, not effect-out-of-binding; effect_identity is a required input of authorization_binding_digest.
  • Status: SUPERSEDED (single direct occurrence; absent from RS5B-07, codex packet, rollup). Controlling text now carried in file 02 §4–§5.

S2 — PATCH1-02 general operation = "register_dot"

  • Old text / source: rs5b-patch1/02 §2 (rev 4): effect_identity = H( protocol_version, operation = "register_dot", canonical_target_dot_code, canonical_artifact_identity, canonical_artifact_hash ) used as the general founding effect_identity.
  • Problem (Codex RS5B-PATCH1 §4 NEEDS_PATCH = R1): hardcoding the effect operation to register_dot makes the corrected contract registration-specific while RS5B-05 is a founding-act packet (scope + first-owner designation are distinct ops); it can bind a founding authorization to the wrong business act.
  • Superseded by: RS5B-PATCH2 (rs5b-patch2/02 §1; impact rs5b-patch2/04 rows 1–2).
  • New controlling text: general formula effect_identity = H(protocol_version, canonical_operation, canonical_target_object_type, canonical_target_object_ref, canonical_artifact_identity, canonical_artifact_hash). register_dot is one specialization (set canonical_operation="register_dot", canonical_target_object_type="dot", canonical_target_object_ref=canonical_target_dot_code); founding/scope MUST NOT use register_dot.
  • Status: SUPERSEDED. RS4A-PATCH2-02 §1 register_dot specialization stays valid (domain-scoped), not the general formula. Controlling text in file 02 §4, §6.

S3 — PATCH1-04 BI-E1 / BI-E6 overlapping input domains

  • Old text / source: rs5b-patch1/04 (rev 5) BI-E1 ("binding carries owner/approval evidence but no effect_identity") and BI-E6 ("authorization_binding_digest omits effect_identity") — prose reserved different codes but did not make the inputs mutually exclusive or define an evaluation order.
  • Problem (Codex RS5B-PATCH1 §7 NEEDS_PATCH = R2): one malformed packet can satisfy both descriptions ⇒ two contract-compliant outcomes despite the "one canonical rejection" claim; not a deterministic oracle.
  • Superseded by: RS5B-PATCH2 (rs5b-patch2/03 §2–§5; impact rs5b-patch2/04 rows 4–6).
  • New controlling text: discriminator P = (effect_identity ∈ authorization_binding_digest input schema). BI-E6 = Layer-1 digest-shape, evaluated first (¬P → AUTHORIZATION_BINDING_MISSING_EFFECT, STOP); BI-E1 = Layer-2 approval-binding, reached only if Layer-1 passed (P, approval not bound to that exact effect → APPROVAL_NOT_BOUND_TO_EFFECT_IDENTITY). Precedence MISSING_EFFECT < NOT_BOUND. No input receives both codes.
  • Status: SUPERSEDED / refined (BI-E1 re-homed to Layer 2; BI-E6 refined to Layer-1 shape, code unchanged; adversarial table rows replaced). Controlling text in file 02 §7.

S4 — PATCH2 rollup "All revision 1"

  • Old text / source: macro-rs5b-patch2-… rollup body sentence asserting all PATCH2 files are revision 1.
  • Problem (Codex RS5B-PATCH2 §2, §9.1): inaccurate metadata — rs5b-patch2/06-rs5b-patch2-decision-packet is revision 2. A metadata/readback typo, not a semantic conflict (the rev-2 decision packet carries the same accepted R1/R2 contract and gate posture).
  • Superseded by: this closeout's caveat ledger (file 04, caveat C1).
  • New controlling text: future summaries MUST quote AgentData metadata, not the stale body sentence: PATCH2 revisions are 1/1/1/1/1/1/2/1 (index/01/02/03/04/05/06/review-packet), rollup 1.
  • Status: CORRECTED-BY-CAVEAT (non-blocking; tracked in file 04).

2. Stacking summary (how PATCH1 and PATCH2 compose)

Layer What it fixed Net controlling result
RS5B (rev 1) Produced the 13-element packet, candidate models, bootstrap posture, BI probes. Carried the item-5 direction slip (S1). Base contract — superseded only on S1.
PATCH1 Fixed the binding direction (S1): effect pure + bound inside digest; added AUTHORIZATION_BINDING_MISSING_EFFECT, EFFECT_IDENTITY_IMPURE. Still hardcoded operation="register_dot" in the general formula. Direction correct; R1/R2 residuals open → NEED_RS5B_PATCH2.
PATCH2 Generalized the operation field (S2) and made BI-E1/BI-E6 disjoint (S3). R1 + R2 closed at contract layer → ACCEPT_RS5B_PATCH2 (PASS_WITH_CAVEATS).
Closeout Records S4 (rollup revision typo) as caveat C1; consolidates S1–S3 into one controlling reading. One unambiguous RS5B_CONTRACT_AS_CORRECTED_BY_PATCH1_PATCH2.

3. Explicitly NOT superseded / NOT reopened

PATCH2 (rs5b-patch2/04 §5 attestation) confirms — and this closeout preserves — that the following are NOT reopened by any supersession above:

  • the accepted authorization_binding_digest field set (Codex §5 PASS);
  • the PATCH1 binding-direction fix (effect-in-binding);
  • the RS5B candidate models and 13-element packet structure;
  • RS4A-PATCH2-02 register_dot semantics (the specialization stays valid, domain-scoped);
  • RS5A-PATCH4 quorum / G02 / total-Q-order semantics;
  • owner / bootstrap / handler posture and U1/U2/U3.

canonical_operation is recorded as introduced-by-PATCH2; the term does not pre-exist in the KB and its runtime vocabulary is REQUIRED_NOT_PRESENT. No original file was overwritten by any patch or by this closeout.

Back to Knowledge Hub knowledge/dev/laws-new/reports/rs5b-closeout-p2-entry/03-supersession-map-rs5b-patch1-patch2-2026-06-21.md