Verify a decision

{"actor":"judge","decided_at":"2026-08-25T13:42:25.716Z","decision":"review_revise","investigation_id":"4485e8d6-1d5a-4668-81f2-6d32ff0baeae","new_score":0,"page_slug":"vy-pham","prev_score":4,"reason":"The reviewer read the full SEC and DOJ complaints and confirmed the large majority of claims on this page, including specific dates, dollar figures, and quoted Telegram messages, against primary sources. Computing disputed_pct on the finding set actually recorded (3 disputed + 3 unverifiable of 35 findings) gives 17.1%, which sits in the minor-issues band, and the reviewer's own count of six routine confirmed claims apparently lost from the array would bring the true rate closer to 14.6% -- either way, comfortably below the threshold for a status change. That said, two findings raise real accuracy concerns for a page about a named individual in active legal proceedings: claim_findings[14] attributes a specific BitMart volume-inflation quote directly to Pham when the SEC complaint attributes it to 'one of Pham's associates,' and claim_findings[16] has the page describing Pham as concealing the scheme via a request for an 'official agreement,' which the reviewer says inverts the roles described in available sourcing -- both move alleged conduct onto the named subject that sourcing places elsewhere. Separately, claim_findings[25] and a linked high-priority coverage gap show the page describing the SEC matter as an unresolved 'accusation' with 'no judgment entered,' while its own cited source states Pham has already consented to a settlement subject to court approval; this understates how resolved the civil matter is. A second high-priority gap notes the '20 years / $5 million' sentencing exposure claim (claim_findings[21]) could not be checked because the DOJ charging document returned a 403 error, and the reviewer separately found secondary reporting suggesting the true statutory maximum may be lower for at least one count -- an access failure compounded by a substantive doubt. Together these support revision rather than approval, but the volume and quality of confirmed primary-source verification (27 of 35 recorded findings) and the reviewer's moderate-high confidence (0.75) argue against escalating to denial or delisting.","score_delta":-12,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}