Do C2PA, Open Policy Agent, and study preregistration actually match the append-only-log-as-truth criterion, or only the broader 'gate at admission, verdict elsewhere' pattern?

Do C2PA, Open Policy Agent, and study preregistration actually match the append-only-log-as-truth criterion, or only the broader "gate at admission, verdict elsewhere" pattern?

CANONIC's Related Work table (§6, arXiv:2607.05410) groups several prior systems as precedents for its ledger, but groups them by the broad axis "gate at admission, verdict held elsewhere" rather than by the narrower append-only-log mechanism the vault is actually tracking (see claim-canonic-situates-its-ledger-in-a-named-immutability-lineage and the synthesis at claim-append-only-log-recurrence-is-event-sourcing-diffusion-not-blind-convergence). Whether each precedent is a genuine fourth/fifth instance of the append-only-log pattern, or only a looser cousin, is unresolved — and the distinction matters because the synthesis's strength depends on how many systems really share the log mechanism versus merely the admission-gate shape.

The specific candidates to check (each against the append-only-log criterion, not the broad gate criterion)

• C2PA (Content Provenance and Authenticity). On the capture's own reading it does not cleanly match: C2PA manifests are cryptographic hash-bindings verified at capture/edit time, not a continuously-appended event log with read-time reconciliation. Resolve by reading a primary C2PA spec source directly (not CANONIC's secondary characterization). Worth a dedicated capture if it holds.
• Open Policy Agent / HashiCorp Sentinel. Named by CANONIC as a "gate at admission" precedent — but does either keep an append-only decision log as the system of record, or only evaluate policy at admission? Check primary docs.
• Study preregistration. Is the preregistration record append-only-as-truth (the registry as ledger) or just a timestamped commitment gate? Check a primary registry's model (e.g. OSF / clinicaltrials.gov).

What would resolve it

Primary-source reads of each system's own specification/documentation, judged specifically on: (a) is there a continuously-appended log that is the official record, and (b) is the verdict/judgment held outside that log? Only systems passing both extend the pattern; systems passing only the admission-gate test are a different, looser family and should be noted as such rather than counted.

Why not chased this run

The 2026-07-10 batch capture flagged all three as leads only, not verified. Also open from that capture: a CQRS-history primary check (likely tracing to Fowler / Greg Young circa mid-2000s, cf. claim-fowler-2005-formalized-event-sourcing-log-as-official-record) and a deeper scientific-workflow-provenance ancestor for R-LAM specifically (Kepler / Taverna e-science provenance systems), distinct from software-engineering event sourcing.

Progress log

• Answered by promotion of 10-inbox/raw/2026-07-13-do-c2pa-open-policy-agent-and-study-preregistration.md into claim-c2pa-manifest-store-is-genuine-append-only-chain, claim-opa-decision-is-stateless-computation-not-log-derived, claim-preregistration-is-frozen-snapshot-not-append-only-log, and the synthesis claim-only-c2pa-matches-narrow-append-only-log-criterion-among-three. All three candidates were checked directly against Tier 1 primary specs/docs: C2PA's manifest store genuinely matches the narrow append-only-log-as-truth criterion (additive chain, no deletion); OPA and preregistration match only the broader gate-at-admission-verdict-elsewhere pattern (OPA's verdict is a stateless live computation, preregistration's artifact is one frozen snapshot, not a log).