# Judge diagnostics — B-denominator Bundle format: `ordinary-harness-export-v1` ## Findings - **false_pass** — 2 false pass(es): outcome accepts while the expected label rejects. Cases: B19, B22. - **headline_ambiguous_denominator** — Headline 'false_pass_rate' is unlabeled: it equals the of-all-cases rate while the expected-fail rate differs. An unlabeled denominator can understate the escape rate. ## Attempted-case accounting Every attempted case is accounted for below; a case without a completed judge verdict is listed, never dropped. | Class | Count | |---|---| | Attempted case-records | 24 | | Unique cases | 24 | | Completed verdicts — correct | 21 | | Completed verdicts — incorrect | 3 | | Abstain | 0 | | Error | 0 | | Unavailable | 0 | | Timeout | 0 | | Other / unclassified status | 0 | | Accounted total | 24 | Reconciliation: attempted (24) equals accounted (24). ## Accuracy — both ways - Verdict-only accuracy (over completed verdicts): 21/24 = 87.5% - Attempted-case accuracy (over all attempted cases): 21/24 = 87.5% Incorrect verdicts — every miss named (completed verdict disagrees with the expected label): - B03 (run-1): expected pass, verdict fail - B19 (run-1): expected fail, verdict pass - B22 (run-1): expected fail, verdict pass ## False-pass rate — both denominators, labeled - False passes: 2 (B19, B22) - Expected-fail cases (expected label is fail/reject): 8 — B17, B18, B19, B20, B21, B22, B23, B24 - Of ALL cases (2/24): 8.3% - Of EXPECTED-FAIL cases (2/8): 25% ## Headline check | Headline field | Headline value | Computed | Status | |---|---|---|---| | accuracy | 0.875 | verdict_only=87.5%, attempted_case=87.5% | match | | false_pass_rate | 0.08333333333333333 | of_all_cases=8.3%, of_expected_fail=25% | ambiguous_denominator | Definitions: positive verdicts are pass/accept, negative are fail/reject; a verdict matches an expected label when equal or same-polarity. Outcome = final_disposition when present, else the judge verdict. A false pass is a positive outcome against a negative expected label. A fail-open escape is a non-completed case whose outcome accepts what expected rejects.