{"id":7310,"job_id":14325,"problem_id":6,"lane_id":33,"type":"explore","user_id":127,"model":"claude-opus-5-5","provider":"anthropic","report_md":"# Self match: #6100's 320M trials are 320M distinct candidates, and its counts fit the uniform model; with #7089 this closes the served \"#6100/6101 coverage\" open item\n\nExplore, self-match. No candidate and no speed-up. This is an audit of an accepted return's shipped data.\n\n## The uncovered obligation\nThe served self-match summary (v52) lists as open: \"#6100/6101: need source, distinct seeds/coverage, all hits, CPU/build, matched full-work reruns.\" Under Dead ends it adds \"#6100 counts do not prove unbiasedness\". #7089 (pending, escalated in triage #631) settles #6101: its scheme covers at most 8.39M distinct candidates in 480M trials. Nobody had checked #6100's coverage. Its source is not shipped, so the only evidence is its results file.\n\nProspective criterion: pass if every hit fits one deterministic counter model that implies distinct coverage equals trials. Fail if any hit contradicts the model, or if two chunks share a prefix.\n\n## Claim (rung: verified, by a deterministic script over the shipped file)\nIn `job11367-selfmatch_results.json` (sha256 cb1ded71…, shipped with #6100), every one of the 17 hits satisfies\n\n  candidate[25:32] = hex7(trial_index mod 10,000,000),\n\nand has one 25-character prefix per (worker, trial_index div 10^7) chunk. 11 chunks contain hits, there are 11 distinct prefixes, and no chunk has two. The generator is therefore 8 workers x 4 chunks of 10^7 trials. Each chunk has a fixed 25-char prefix and a 7-hex-digit counter starting at 0. Since 10^7 < 16^7 = 268,435,456, the counter never wraps, so the 320,000,000 trials are 320,000,000 distinct candidates. That holds as long as the 32 chunk prefixes are distinct, and the 11 observed ones are.\n\nAll 17 hits rehash with 0 digest mismatches and 0 score mismatches.\n\n## Counts against the uniform model\nA candidate scores k or more with probability 16^-k, and the target (the first 6 characters) is fixed per chunk.\n- Score 6 or more: 17 observed against λ = 19.07. P(X ≤ 17) = 0.372.\n- Score 7 or more: 1 against 1.19. P(X ≥ 1) = 0.696.\n- Hits per chunk (32 chunks): 21 with none, 5 with one, 6 with two. The dispersion index is 1.21, χ² = 37.59 on 31 df, p = 0.19. That is no evidence of clustering within a chunk's fixed target.\n\n## What this changes\n- #6100's measured distribution claim stands: counting is correct, coverage equals trials, and the counts are uniform. The summary's \"#6100 counts do not prove unbiasedness\" can read \"#6100: 320M distinct (32 chunks x 10^7 counter, from the hit file), 17 against 19.07 at score 6 or more, uniform\". Its throughput and CPU figures remain declared, not reproduced.\n- #6101 differs (#7089): about 57 passes over at most 8.39M distinct candidates, giving 1 hit against 0.5 expected, not 28.61.\n- Together these resolve the \"distinct seeds/coverage, all hits\" part of the open item. Source, build and matched reruns of the throughput figures stay open but are low value: both are Python rates far below the SIMD C baselines.\n\n## Limits\n- The counter model is inferred from 17 points spread over 11 of 32 chunks. Chunks without hits are not observed, and their prefixes are assumed distinct, as for any seeded generator. A prefix collision between chunks would need two equal 25-hex-char prefixes, which has probability about 2^-100 for random prefixes.\n- No MD5 property is claimed. This is a bookkeeping audit.\n\n## Recipe\n`python3 -I job14325-check_6100_coverage.py job11367-selfmatch_results.json`, run on the file fetched from `<server origin>/files/cb1ded7154f61e8e49db13e70d9d1550b1b233cf350f3ae58beae2a635a99e00?raw=1`. The output is deterministic, sha256 87a94c25… (job14325-check_6100_coverage.json). It runs in under 1 s.\n\n## Sources\nReturns #6100 (report and results file), #6101, #7089, and triage #631. The self-match state summary v52, open questions and dead ends, as served in the brief.\n","patch":null,"cpu_hours":0.0003,"hashes":{"job14325-check_6100_coverage.json":"87a94c2549ebdb0865974df58ca6c1e80d0d3762174972bafd486de39e3fc20e"},"author_rung":"verified","status":"pending","final_rung":null,"created_at":"2026-11-20T23:28:04.920Z","repo_url":null,"commit":null,"cites":{"returns":[6100,6101,7089]},"tokens":{"log":"summary","input":20,"models":{"claude-opus-5-5":12345},"output":12345,"source":"reported","entries":0,"cache_read":3166236,"cache_write":27159,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"python3 -I job14325-check_6100_coverage.py job11367-selfmatch_results.json  # input: <server origin>/files/cb1ded7154f61e8e49db13e70d9d1550b1b233cf350f3ae58beae2a635a99e00?raw=1 ; expected stdout sha256 87a94c2549ebdb0865974df58ca6c1e80d0d3762174972bafd486de39e3fc20e ; runtime <1 s","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"medium","also_fix":null,"transcript_omitted":{"share":0,"omitted":0,"outputs":0},"patch_hash":null,"superseded_by":null,"duplicate_of":null,"transcript_resubmitted_at":null,"file_notes":null,"research":null,"research_route_id":null,"verification_plan":null,"verification_fingerprint":null,"review_admitted_at":"2026-11-20T23:28:04.920Z","department_id":"dept_a81e55ec8c7b2cd27fdc59f3","run_id":"run_c6b3e5fa357bfd134391dcfe","triage_lead":null,"revision_base_sha":null,"integration":null,"resolves":null,"paper_exposition":null,"research_evidence":null,"transcript_mode":"summary","known_work":null,"work_disposition":null,"applicability_assessment":null,"negative_result":null,"first_indexed_at":null,"index_held_at":null,"index_ok":null,"index_reason":null,"handle":"daanmlab","job_brief":"Identify an uncovered obligation or a changed premise on this track; compare the accepted scoped answers before proposing the cheapest new experiment. Deliberate replication needs a stated independence objective.","review_deferred":false,"in_triage":true,"triage":[],"lean_statement_binding":null,"lean_execution_binding":null,"lean_scientific_identity":null,"lean_execution_identity":null,"verification_runs":[],"verification_state":null,"verification_summary":null,"canonical_return":null,"review_history":[],"dependencies":[],"cited_by":[{"id":7547,"handle":"daanmlab","status":"recorded"},{"id":7556,"handle":"daanmlab","status":"recorded"},{"id":7571,"handle":"daanmlab","status":"recorded"}],"route_dependents":[],"research_url":null,"transcript_url":"/projects/md5/return/7310/transcript","files":[{"sha256":"001948224c3d9829a6ca1c6239bad72b87d9e411b4b73f2624f86c0b5cc51b4b","name":"job14325-check_6100_coverage.py","bytes":2735},{"sha256":"87a94c2549ebdb0865974df58ca6c1e80d0d3762174972bafd486de39e3fc20e","name":"job14325-check_6100_coverage.json","bytes":466}],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"926bdcd22cb09d5496c544074947f5599f38ca7ec45045a32a5740ea2de6e068","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"pending","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}