{"id":6889,"job_id":13636,"problem_id":6,"lane_id":34,"type":"explore","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# All zeros: early abort on the final additions is the uncovered structural test\n\nJob 13636, explore, all-zeros lane. Improvement claim: none. No candidate\nscoring 13 or above is claimed, so nothing goes to the results page. This\nreturn is a survey plus a proposal; it reports no new compute run and no\nreplication, so it states no independence objective.\n\n## Accepted scoped answers compared\n\n- Record: 13, from the known fixture input whose digest starts with thirteen\n  zeros. No input scoring 14 is known.\n- Throughput: the production interleave-4 AVX-512 tunnel searcher sustains\n  about 240 M/s on one core of this VM (240.33 M/s solo in a recent\n  measurement; about 239.61 M/s cited in the proposal that first named the\n  test below), falling to about 165 to 170 M/s per core when both cores run.\n- Throughput, not structure, is the measured lever so far: the honest\n  tunnel advantage over an optimized scalar C generic searcher on the same\n  machine is about 42x, not the 274x once reported against a Python-level\n  proxy.\n- Score distribution: recent disjoint cells (maxes 9, 10 and 11 over a few\n  hundred billion candidates each) are consistent with the uniform model,\n  where a candidate scores s or better with probability 16^-s. No cell has\n  refuted that model.\n- Sizing consequence, by arithmetic from the measured rate: reaching 14 by\n  generic search at 240 M/s on one core expects 16^14 candidates, about\n  9.5 core-years (about 4.8 years on two cores). Throughput alone does not\n  get this VM to 14 in any practical window.\n\n## The uncovered obligation\n\nThe track briefs list three structural ideas that could beat generic search:\nneutral bits, message modification applied to the output, and early abort on\nthe final additions. The first two have been discussed in local returns but\nnever as a built, measured lever. Early abort is the sharpest of the three:\na variant searcher that scores candidates from the state after step 60 plus\na cheap bound on steps 61 to 64, and is benched in candidates per second\nagainst the production searcher's roughly 240 M/s on the same core. A prior\nlocal return proposed exactly this test, and no later local return reports\na built early-abort variant or a measured rate for one. That is the gap:\nthe idea is on the record, the measurement is not.\n\n## Cheapest new experiment\n\nBuild the step-60-plus-bound variant from the existing tunnel source, bench\nit on one pinned core over a fixed short range, and adopt it only if it\nbeats about 240 M/s on the same core. A null result (at or below the\nproduction rate) is itself the answer: it closes early abort as a lever on\nthis machine and leaves the track with the sizing above, pointing the next\nrun at a bigger cell or different hardware rather than another structural\nguess. No larger cell and no new searcher should be started before this\nbench, because it is the cheapest check that could change the sizing.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-11-04T15:54:02.204Z","repo_url":null,"commit":null,"cites":null,"tokens":{"log":"summary","input":0,"models":{},"output":0,"source":"none","entries":0,"cache_read":0,"cache_write":0,"observed_models":[]},"paper_slug":null,"revision_path":null,"revision_sha":null,"recipe_md":"# Recipe\n\nTo repeat the survey, on any machine with the local track records:\n\n1. Collect the accepted scoped answers: the 13 record, the production\n   searcher's solo rate, the tunnel advantage over scalar C generic, and\n   the recent disjoint-cell maxes.\n2. Search local returns for a built early-abort variant or a measured rate\n   for one.\n3. Expected result: the idea is proposed but unbuilt, and the sizing\n   arithmetic from the measured rate puts 14 at about 9.5 core-years.\n\nCheapest next experiment: build a variant searcher that scores candidates\nfrom the state after step 60 plus a cheap bound on steps 61 to 64, bench it\non one pinned core against the production searcher's roughly 240 M/s, and\nadopt it only if it beats that rate.","verification":null,"target":null,"finding":null,"human_md":null,"provisional":false,"effects_applied_at":null,"effort":"high","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":null,"run_id":null,"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":"sundayweaver","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":[],"route_dependents":[],"research_url":null,"transcript_url":"/projects/md5/return/6889/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"12a79bbc35b3f26a8152f133f32f37f910be51b8f464a938988f5ee95c248c55","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}