06 — Findings & Interpretation

What the F-1 read-only dry-run actually proved, plus discrepancies surfaced against the readiness package. Honest reporting per the governing law (no overclaim).

6.1 F-DR-1 — gap grain is (object × scope), not object

Finding. The live v_object_owner_gap definition CROSS JOINs governance_responsibility_scope, so its grain is (object × active-scope). Over 3 collections × 6 active scopes it returns 18 rows, not the "≤3 gap rows" headline in readiness doc 03 §3.3.

Impact. Bounded and benign for the dry-run (18 ≤ 38 ceiling), but the readiness ceiling table conflated grains: it correctly fanned v_object_effective_owner by 6 scopes (≤18) yet left v_object_owner_gap at object grain (≤3). The true gap ceilings are:

• 3-collection subset: 18 (3 × 6).
• Full 35 governed collections: 210 (35 × 6), not 38.

Action. Future dry-run/activation gates (doc 05 §5.2 in-flight ceilings) should use the per-scope ceiling. Not a blocker; a correction to fold into the next macro and any seam-wiring (F-4) authorization.

6.2 F-DR-2 — the coverage path is sound and honest

The end-to-end path — inventory seam → effective-owner → gap — produces the correct verdict: every governed+born object with no accountable owner in an active scope is flagged GAP. With ownership empty, it returns the maximal honest gap set and zero false PASS. The decaying-verdict / candidacy-not-mandate design (readiness doc 06) is fully exercisable as read-only logic; persistence and emit are the only parts gated behind ratification, and they were correctly not performed.

6.3 F-DR-3 — inventory seam wireable at collection grain today (decision-gated)

The Seam-A expression over collection_registry yields exactly the expected 3 rows with correct requires_owner/born flags. The seam is wireable now at collection grain — the only thing missing is the ratified decision to CREATE OR REPLACE it onto the real source (F-4), which crosses inert→active for coverage and requires L2+L4 (readiness doc 02 §2.0). Engineering is ready; authorization is not.

6.4 F-DR-4 — axis layer remains fake/test-only (unchanged)

No axis coverage was attempted. axis_registry / axis_assignment / coverage_rule remain ABSENT, so all axis/topic/assignment/UI-exposed detectors stay fake/test-only (readiness doc 02 §2.4). Any axis dry-run is NO-GO until that substrate is ratified and built. This dry-run deliberately stayed on the live, buildable inventory/ownership path only.

6.5 F-DR-5 — zero-emit / zero-issue boundary held perfectly

event_outbox (governance) and system_issues both unchanged. The 18 findings exist only as computed log lines in this package (≤50-line cap respected). No event-type activation, no T7 routing, no notification, no audit row. The boundary in readiness doc 06 is empirically confirmed.

6.6 F-DR-6 — missing GPT-review artifact (documentation discrepancy)

The mission named gpt-review-operational-readiness-package-verified-go-f1-readonly-2026-06-02.md as a state-recovery input. It does not exist on disk or in the KB architecture path (verified via list_documents). The GO decision is nonetheless solidly grounded in the published readiness package (doc 08 §8 = GO read-only/rollback-only F-1/F-2) and the KB closing doc operational-readiness-package-publication-verification-and-f1-decision-2026-06-02.md (rev1, decision GO_F1_READONLY_DRY_RUN). No authorization gap — only a missing/misnamed review artifact. Logged for the operator to reconcile (it may live under a different name, or the GPT review may be pending).

6.7 What the dry-run did NOT and could NOT prove

• It did not prove the containment path (Seam B) — that seam stays WHERE false; inherited ownership was not exercised (ownership is empty anyway). Out of F-1 scope.
• It did not prove object-grain behavior (1.08M birth_registry) — collection grain only.
• It did not prove persistence/emit behavior — those are F-5/F-6/F-7, ratification-gated.
• It cannot speak to performance at scale — 3 collections, ≤5s, trivially fast.

6.8 Findings verdict

BOUNDED, EXPLAINABLE, HONEST. One real correction (F-DR-1, gap grain → per-scope ceilings), one confirmation the path is sound and fails safe (F-DR-2), one buildability confirmation (F-DR-3), the axis NO-GO reaffirmed (F-DR-4), the zero-emit boundary held (F-DR-5), and one documentation discrepancy to reconcile (F-DR-6).