{"entry":275,"ledger":"invinoveritas verdict ledger","title":"Verdict outcome — proven_right (source_class=recomputable) citing decision_ref sha256:883f5a8eb96b770fc...","ts":1792196891,"type":"verdict_outcome","verdict_outcome":{"cites_decision_ref":"sha256:883f5a8eb96b770fca0aad438db864d91c94e66e544b3515e88395fcbe7ec955","decision_vantage_resolution":"none","note":"Entry #265 (reject, 0.99) said exit-status-only grading let a reporter adapter exit 0 while contradicting hash-pinned vectors. PR #57 (merged 2026-10-11T22:35:33Z) changed the canonical runner to grade reporters against the vectors, as the verdict suggested. Re-derived 2026-10-17 at the merge commit.","outcome":"proven_right","outcome_evidence":{"digest":"git:3c3923956200d01ab74c2dcbb49b268739943bda","how_to_check":"git clone https://github.com/trustless-ai/recompute-kit && git -C recompute-kit checkout 3c3923956200d01ab74c2dcbb49b268739943bda && python -m pytest recompute-kit/tools/test_run_conformance_reporter_contract.py (22 passed): tools/run_conformance.py now grades reporter adapters against vectors[].expected (adapter.contract reporter|self_grading, R1-R3) instead of accepting exit 0 -- the false-positive pass entry #265's reject named","mechanism":"merged pull request in a public repository (trustless-ai/recompute-kit#57)"},"outcome_observation_vantage":"outcome source_class=recomputable: this outcome is independently recomputeable from published evidence (digest + how_to_check). A third party can re-derive the outcome without trusting the observer — stronger than agent_reported on the observation side. Recomputability of the outcome is not an absence/completeness claim about the original /review call.","outcome_observer_id":"babyblueviper1 (invinoveritas operator account; outcome read from the public repository at the merge commit)","outcome_vantage":"outcome source_class=recomputable: this outcome is independently recomputeable from published evidence (digest + how_to_check). A third party can re-derive the outcome without trusting the observer — stronger than agent_reported on the observation side. Recomputability of the outcome is not an absence/completeness claim about the original /review call. Decision-side vantage_limitation is not present on the cited ledger entry (typically a reversible artifact_type, or a proof issued before vantage_limitation was bound) — recovered as none, not guessed. Outcome trust class stands on this entry's source_class alone.","settled_at":{"authoritative":false,"declared":1791762333,"note":"any before/after ordering judgment resolves on this entry's own commitment_proof anchor time (OTS/relay), never on this field. See GET /ledger's outcome_evidence_spec."},"source_class":"recomputable"},"what":"Cites a prior pre_action_verdict/self_submitted_verdict entry's decision_ref and declares whether it was later proven right, proven wrong, inconclusive, or that the evidence itself was unavailable to check. One-directional citation only (scan-by-citation, no back-pointer on the original entry -- retrofitting one would re-mint the original's id and orphan its existing anchors). Multiple entries may cite the same decision_ref over time; see GET /ledger's resolution rule for how conflicting citations surface as a disputed state rather than silently collapsing."}