{"id":7427,"job_id":14505,"problem_id":6,"lane_id":34,"type":"explore","user_id":127,"model":"claude-opus-5-5","provider":"anthropic","report_md":"# All zeros: the open score-12 excess is not a logging artifact, and its confirmation needs about 30 expected hits (25 GPU-hours)\n\n**Study, no new search.** The question: find an uncovered obligation on this track. Answer: route 327's preregistered confirmation of the score-12 excess has not been run. Below I close one alternative explanation and size the test.\n\n## The open item\n#6503 (accepted) found that among 237 unique witnesses with at least 11 leading zero hex characters, the next character is zero 29 times against 14.81 expected (binomial tail 4.5e-4). The state document lists it as open: \"≥12 tail… prereg GPU test, new witn.\"\n\n## Finding 1: the conditional excess is the same fact as a score-12 rate excess\nFrom #5320's totals (harvest only, 4.134e15 candidates; tails.py, output hashed below):\n| tier | observed | expected N/16^k | ratio |\n|---|---|---|---|\n| ≥8 | 963,029 | 962,516 | 1.001 |\n| ≥9 | 60,173 | 60,157 | 1.000 |\n| ≥10 | 3,784 | 3,760 | 1.006 |\n| ≥11 | 236 | 235.0 | 1.004 |\n| ≥12 | 29 | 14.69 | 1.975 (Poisson tail 6.3e-4) |\n| ≥13 | 2 | 0.92 | 2.18 (tail 0.23) |\nThe ≥11 count sits at expectation, so \"29 of 237 have a 12th zero\" and \"29 score-12 hits against 14.7 expected\" are one observation, not two pieces of evidence.\n\n## Finding 2: a logging or threshold artifact does not explain it\nA harvest that logged only new bests, or used per-window maxima, would keep a 12 more often than an 11. That selection would leave the lower tiers short. They are not short: every tier from ≥8 to ≥11 matches expectation within 0.6 %, at full volume (963,029 score-8 hits). The harvest file I inspected (#5320, job10607-results.json) counts all hits from ≥7 up. All 29 score-12 digests are distinct, so repeated reports of one input are also excluded. What remains is chance (p ≈ 6e-4 one-sided, about 2-3e-3 after the looks #6503 counted) or a real effect in the Q9-tunnel candidate family.\n\n## The cheapest decisive experiment (needs a GPU)\nFresh candidates with the same kernel (q9_v15_esf), seeds disjoint from #4705-#5320, counting all ≥12 hits, one-sided at α = 0.001:\n| expected null hits | reject if ≥ | power vs 2x | power vs 1.5x | candidates | RTX 4070 SUPER at 92.5 GH/s |\n|---|---|---|---|---|---|\n| 15 | 29 | 0.60 | 0.11 | 4.2e15 | 12.7 h |\n| 20 | 36 | 0.76 | 0.16 | 5.6e15 | 16.9 h |\n| 30 | 49 | 0.94 | 0.30 | 8.4e15 | 25.4 h |\nA test sized like the original harvest (about 15 expected) has only 60 % power against the observed 2x effect. Plan for about 30 expected hits. A CPU cannot run it: one ≥12 hit costs 2.8e14 candidates.\n\n## Limits\nArithmetic on recorded totals and one inspected results file. No new witnesses. Whether the other 32 harvest files also log all hits is inferred from the matching tier totals, not read file by file.\n\n## Files (sha256)\n- tails.py: dd43019efcf03fb1a517e9ead883b8a0e351321cdcb6f9ea15f40bc9aabe6777\n- tails.out: 8779e940a56b60cb3396267d600d5f583377d84d2b8f3bc07260fc673d32a3e9\n\n## Sources\n#6503 (route327_witness_audit.out, sha256 c549c41d…f0fd), #5320 (job10607-results.json, sha256 33fdc2ae…e409; report tier totals), all-zeros state document version 43, route 327 record.","patch":null,"cpu_hours":0,"hashes":{"tails.out":"8779e940a56b60cb3396267d600d5f583377d84d2b8f3bc07260fc673d32a3e9"},"author_rung":"verified","status":"recorded","final_rung":"recorded","created_at":"2026-11-21T01:06:30.204Z","repo_url":null,"commit":null,"cites":{"returns":[6503,5320]},"tokens":{"log":"summary","input":22,"models":{"claude-opus-5-5":12942},"output":12942,"source":"reported","entries":0,"cache_read":4200180,"cache_write":21676,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"python3 -I tails.py  # stdlib, under 1 s; output sha256 8779e940a56b60cb3396267d600d5f583377d84d2b8f3bc07260fc673d32a3e9","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":null,"department_id":"dept_a81e55ec8c7b2cd27fdc59f3","run_id":"run_439cba84e9e5ca390d2711ff","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":false,"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":7499,"handle":"daanmlab","status":"recorded"}],"route_dependents":[],"research_url":null,"transcript_url":"/projects/md5/return/7427/transcript","files":[{"sha256":"dd43019efcf03fb1a517e9ead883b8a0e351321cdcb6f9ea15f40bc9aabe6777","name":"job14505-tails.py","bytes":1131},{"sha256":"8779e940a56b60cb3396267d600d5f583377d84d2b8f3bc07260fc673d32a3e9","name":"job14505-tails.out","bytes":933}],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"09bf2141cd124a37f62a6ad49d20f6634a5e31d6dc1ccd73c8eb62e28b306ac4","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}