Verify a decision

{"actor":"judge","decided_at":"2026-08-27T08:07:12.511Z","decision":"review_revise","investigation_id":"59338f15-dca7-49d1-b4b9-6b4ed44edb9f","new_score":56,"page_slug":"global-dollar","prev_score":68,"reason":"Of 26 claims checked, 6 are disputed or unverifiable (23.08%), placing this page in the 10-30% band. The page's central claim -- that USDG is issued by Paxos Digital Singapore under MAS supervision -- is confirmed precisely, and its treatment of the NYDFS consent order, the yield-sharing business model, and the attestation-vs-audit distinction all hold up (claim_findings[0], [5], [9], [12], [18]-[20]). However, the page has three separate year-level date errors, each contradicted by its own cited sources: the EU/MiCA launch is dated 2026 instead of 1 July 2025 (claim_findings[2]), the SEC's closure of its BUSD investigation is dated 2023 instead of 9 July 2024 (claim_findings[17]), and an OKX listing is dated June 2026 instead of June 2025 (claim_findings[25]). It also names the wrong accounting firm for USDG's current attestations -- Marcum instead of KPMG, per the page's own cited transparency source (claim_findings[6]) -- and compresses two distinct February 2023 regulatory events into a single incorrect date (claim_findings[16]). None of these disputed items overturn a core allegation; they are date and named-entity errors layered on top of an otherwise well-sourced page, but their number and the fact each is directly contradicted by the page's own citations point to a systematic sourcing/dating problem rather than isolated typos. Two additional findings are classified stale rather than disputed and sit outside the numerator, but weigh toward the higher end of this band: a market-cap figure presented without an as-of date understates USDG's size by 15-25% as of publication, and the page's description of Paxos Trust Company as NYDFS-supervised is superseded by its December 2025 conversion to OCC national trust status. A high-priority coverage gap -- the EU/MiCA claim was never checked against a regulator's own register -- reinforces the revise verdict already indicated by disputed_pct, rather than changing the verdict category. Reviewer confidence (0.80) is solid, supporting a standard rather than softened penalty within the band.","score_delta":-12,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}