{"manifest_kind":"dsar_jurisdiction_coverage","total":109,"with_quantitative_deadline":99,"legal_review_needed":20,"high_confidence":89,"legal_reviewed":46,"engineering_encoded_unreviewed":63,"research_verified":18,"research_source_grades":{"mixed":4,"primary":12,"secondary":2},"source":"apps/api/src/dsar_jurisdictions.py","notes":"Counts computed live from the typed DSAR-deadline table. 'with_quantitative_deadline' is the subset whose deadline is a numeric statutory field; the remainder carry a qualitative standard ('without undue delay'). 'legal_review_needed' rows are encoded but flagged for legal confirmation before regulator-facing reliance. 'high_confidence' is simply total - legal_review_needed (rows not flagged for legal confirmation) — it is a review-status count and does NOT imply a numeric deadline: qualitative-standard rows count as high_confidence too. IMPORTANT: 'high_confidence' does NOT mean anyone confirmed the row — it means we did not flag it, which is a statement about our to-do list, not about the law. 'legal_reviewed' is the count a named human signed for (reviewed_by + reviewed_on + reviewed_scope all present on the row); it is the number to rely on. Rows outside that count are engineering-encoded from the cited statute and unconfirmed. 'research_verified' counts rows where sources were gathered and graded; it is a WEAKER claim than legal_reviewed and is never added to it. Research is evidence-gathering, review is a person standing behind the value, and a row can be thoroughly researched and still unsigned. 'research_source_grades' breaks that down, because 'researched' alone hides what matters: a row can be researched exhaustively and still rest on secondary sources where the official text is unobtainable."}