Verify a decision

{"actor":"judge","decided_at":"2026-08-27T10:49:44.602Z","decision":"review_revise","investigation_id":"0c625ed3-6bc1-4649-a726-569f64e5e919","new_score":37,"page_slug":"volo-protocol","prev_score":47,"reason":"The review's own summary block miscalculated disputed_pct as 0.18; recomputed directly from the 12-item findings array per spec (disputed + unverifiable, divided by total claims), the correct figure is (0+1)/12 = 0.083, which sits in the approve band on arithmetic alone. Nine of twelve claims -- including the NAVI acquisition, the five audits, the $3.5M exploit breakdown, the compromised-key root cause, and the KelpDAO comparison -- were independently confirmed (claim_findings[0,1,3,4,5,6,7,8,11]), and zero claims were disputed. However, claim_findings[9] found that the page's exploit-aftermath sections describe roughly $3 million as unrecovered, when independent reporting within a week of the incident (rekt.news, Phemex) showed about 90% of funds were recovered, leaving a true net loss near $60,000 -- a nearly 50x overstatement of harm to the named subject that was never corrected. Because the reviewer flagged this same gap as a high-priority coverage item, and because it represents a live, material inaccuracy rather than a mere absence of coverage, it is treated as grounds to move the verdict one step past a pure arithmetic approve, into a light revise rather than a hold at zero penalty. The claim_findings[10] unverifiable item (absence of public key-management documentation) is explicitly noted by the reviewer as a tooling limitation rather than a substantive defect and is not weighted further. claim_findings[2] (TVL comparison to SpringSui) is only partially supported due to near-parity in some snapshots, a minor precision issue rather than a factual error.","score_delta":-10,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}