{"id":7079,"job_id":13980,"problem_id":6,"lane_id":33,"type":"measure","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Self-match measure: fixing the prefix buys nothing, 5B prefix-fixed cell uniform; baseline 20B cell uniform\n\nJob 13980, measure, self-match. Improvement claim: none. No longer\nmatching prefix was found, and none is claimed. Nothing goes to the\nresults page.\n\n## Hypothesis\n\nFixing the candidate prefix and searching the rest buys nothing over\ngeneric search in brute-force form: the digest prefix distribution\nconditioned on a fixed candidate prefix is uniform, so the cost stays\n16^8 trials per rung-8 hit. The only remaining hope for the idea is\nalgebraic inversion, which the sibling explore job #13979 closes by\narithmetic (meet-in-the-middle needs at least 2^64 + 2^64 work vs\n2^32 generic for rung 8).\n\n## Experiment\n\nTwo counted cells on one pinned core (taskset -c 1, log interval\n60 s), both binaries rebuilt from source (gcc -O3 -march=znver4\n-mtune=znver4), selftests PASS:\n\n1. Baseline: fresh 20B vary-M7 cell from verified disjoint nonce\n   13999500000000, continuing the tally9 series.\n2. Prefix-fixed: 5B vary-M7 cell with candidate chars 0..7 forced to\n   the fixture prefix 54db1011, from nonce 14019500000000, right\n   after the baseline range.\n\n## Results\n\n- Baseline: 20,000,000,000 candidates, 41.45 M/s, [done] total\n  20000000000, exit 0. Best 8, tally7 69, tally8 2, tally9 0.\n  Uniform expectations: 74.5 / 4.66 / 0.29 (z -0.64 / -1.23 / -0.54).\n  Best candidate 2bf4520b8d21d6d848d93c2523251b52, digest\n  2bf4520bb7bfc9e955e2be86ef3c34f2. All 9 new-best hits\n  hashlib-verified, 0 mismatches.\n- Prefix-fixed: 5,000,000,000 candidates, 32.57 M/s, [done] total\n  5000000000, exit 0. Best 8, tally7 27, tally8 2, tally9 0.\n  Uniform expectations for 5B: 18.6 / 1.16 / 0.07 (z +1.94 / +0.78 /\n  -0.27), uniform within 2 sd at every rung. Best candidate\n  54db101131a6476fe94a13927042fef3, digest\n  54db1011db2b07d0f4fbbaddfef6db8d. All 8 new-best hits\n  hashlib-verified, 0 mismatches.\n- Verified fresh tally9-space total is now 111B. Next verified\n  tally9 start nonce: 14019500000000.\n\n## Rate note\n\nThe prefix-fixed cell ran at 32.57 M/s against the baseline's\n41.45 M/s on the same core, same binary family, back to back. The\nonly code difference is one 8-byte memcpy per 65,536-candidate\nbatch, which cannot explain a 21% gap, and no other load was\nrunning. Cause unexplained, possibly host-side frequency variation\nafter the VM reset; the tallies are counts and do not depend on the\nrate, so the conclusion stands.\n\n## What this means\n\nThe brute-force form of fixing a prefix is closed: conditioned\nsearch costs exactly the generic rate. Together with #13979's\narithmetic closure of the meet-in-the-middle form, the prefix-fixing\nidea is exhausted on this track. What remains open is the\npartial-step / early-exit screen on Zen 4/5 AVX-512 (#13829) and the\nil8 vs il4 head-to-head (#13851/#13959). Next run: continue fresh\n20B tally9 cells from 14019500000000.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"pending","final_rung":null,"created_at":"2026-11-07T02:52:16.447Z","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 baseline cell: rebuild selfmatch_tally9 from source\nwith gcc -O3 -march=znver4 -mtune=znver4, run the selftest, then run\na 20B cell from the next verified tally9 start nonce (this run's next\nstart is 14019500000000), one core pinned, log interval 60 s.\nVerify every [new best] line by re-hashing the candidate with\nhashlib.md5 and recomputing the common-prefix score, and compare\ntally7/8/9 against 20B / 16^rung.\n\nTo repeat the prefix-fixed experiment: build selfmatch_fixedprefix\nfrom the same source with build_base_words forcing candidate chars\n0..7 to 54db1011, run the selftest, then run a 5B cell from nonce\n14019500000000, one core pinned, log interval 60 s, and compare\ntally7/8/9 against 5B / 16^rung.","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":"2026-11-07T02:52:16.447Z","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":"Study how a candidate's 32 ASCII bytes flow through the 64 steps into the first digest characters, and use what you learn to reach a longer matching prefix. Ideas to test: which message words the first output word depends on most, fixing a prefix and solving for the rest, early-exit tests on the first output word, meet-in-the-middle on the step function. Start from the algorithm, not the search. Read the track's state of the art in this brief (the fastest known way per machine, what worked, the dead ends) and research/QUESTIONS.md, then state one hypothesis about MD5's structure that would make this track cheaper than generic search, and why you expect it. Test it with the smallest experiment that could refute it, against a measured baseline on the same machine. Submit the best candidates the experiment produced. The report is a finding: the hypothesis, the experiment, what it showed about MD5 (positive or negative, with numbers), and what the next run should try. Send what you learned as a state_entry with your return (the method, the exact hardware, the measured throughput, the best reached), so the track's state of the art carries it. If the run used only a known tool or plain search, report it as a baseline measurement.","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":[],"route_dependents":[],"research_url":null,"transcript_url":"/projects/md5/return/7079/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"a8bbfec8b385f077c8799f41ba7a241fbac4253a60d6e9dc84c5834215b1fb73","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"pending","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}