Verify a decision

{"actor":"judge","decided_at":"2026-08-25T03:08:00.200Z","decision":"review_revise","investigation_id":"28e9132f-6453-444b-94da-233ce895ed34","new_score":17,"page_slug":"term-labs","prev_score":22,"reason":"The reviewer confirmed 34 of 39 claims on this page, including every load-bearing detail of both loss events — the $8.5 million August 2026 governance exploit and the $1.6 million April 2025 oracle incident — across independent, tier-1 sources, and found no dead links, stale citations, or high-priority coverage gaps. Statistically, the disputed rate is very low (about 2.6%, one finding out of 39). However, that single disputed claim (claim_findings[27]) is not a trivial detail: it states that DeFi governance attacks caused $25.1 million across five 2026 incidents 'prior to' the Term Labs event and calls Term Labs' $8.5 million loss the largest in that cluster. Per the reviewer's sourcing, Term Labs' loss is actually one of the five incidents already counted in that $25.1 million (not additional to it), and a separate $20 million BonkDAO exploit from July 2026 is the largest in the cluster, not Term Labs. This double-counts the loss total and overstates the entity's relative severity within an industry comparison the page itself uses to editorialize about risk — exactly the kind of claim a risk-intelligence platform needs to get right, even when everything else checks out. It also sits in the same later sections of the page as a small cluster of lower-severity accuracy issues (claim_findings[19], [26], [28], [35]): a two-day date discrepancy between the page's own cited sources, an understated 16-vs-19-month interval between incidents, and a claim that is true but not actually supported by the sources cited for it. None of these alone would justify action, but together with the compound error in claim_findings[27] they indicate the page's comparative/analytical framing needs a light correction pass, even though its core factual record is solid.","score_delta":-5,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}