{"id":7184,"job_id":14141,"problem_id":6,"lane_id":35,"type":"explore","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Smallest collision: step-12 satisfiability on p37, answered\n\nJob 14141, explore, smallest-collision lane. Improvement claim:\nnone. No collision below 128 bytes total was found, and no\ncandidate scoring 13 or above was found, so nothing goes to the\nresults page.\n\n## The question, answered\n\nJob 14130's standing next step was analytic: is the member-2\nstep-12 equation satisfiable under p37's rows 9-13 conditions,\nor is it structurally unsatisfiable? The answer is that it is\nsatisfiable, in fact universally satisfied, whenever Q11 is\nconsistent with row 11, and unsatisfiable only under the\nscreened configuration. The p37 barrenness is a configuration\ncontradiction, not a property of the path.\n\nUnder the screened configuration (FORBID 0xb3280000, the mask\nused by the #14124 screen and the #14130 diagnosis), the\nstep-12 check fails for every one of the 262,144 distinct\n(Q10, Q11, Q12, Q13) tuples the sampler can draw: 0 passes.\nAn independent CNF encoding of the same equation returns\nUNSATISFIABLE under kissat. The cause is two bits of Q11. The\npadding mask forces Q11 bits 0x32000000 (bits 25 and 29) to 0\nthrough the row-11 forced-zero mask, while row 11's conditions\nrequire those exact bits to be 1. The sampled Q11 is therefore\nalways 0x0005ba5d instead of 0x3205ba5d, it violates its own\nrow-11 conditions on all 262,144 tuples, and the step-12 check\nfails for all of them. Clearing just those two bits from the\nrow-11 forcing, with everything else identical, flips the\nresult to 262,144 passes out of 262,144. With no padding\nrestriction at all, 4,096 out of 4,096 pass. The p30 control\npasses 256 out of 256 with no conflict bits.\n\n## Rows 14 and 15 pass too; the Q16 scan is the next wall\n\nThe same exhaustive method was run one and two steps further,\nbecause the DETECT run with no padding restriction also found\n0 bases and had to be explained. In DETECT mode (Q11 free over\n128 values) rows 13, 14 and 15 all pass universally: row 13,\n524,288 out of 524,288 tuples; row 14, 4,194,304 out of\n4,194,304; row 15, 1,073,741,824 out of 1,073,741,824, with\nevery reachable prefix admitting a passing next value. So the\nDETECT run's 0 bases do not come from rows 13-15. They come\nfrom the Q16 padding scan: a Monte-Carlo over 2,000 reachable\nprefixes, scanning all 65,536 L62 w values each (131,072,000\nvalues total), found 0 passing values on p37, every one failing\nthe row-16 conditions before the member-2 check is even\nreached. The p30 control run the same way gives 224 of 2,000\nprefixes with a base (11.2%, against 9.29% in the #14130\ndiagnosis), so the scan code is validated and the p37 result\nis a property of p37's reachable states, not of the scanner.\nThe per-prefix Q16 base rate on p37 is below about 0.15% (95%\nupper bound). Fixing the FORBID alone therefore does not yield\nbases on p37; the next work is the Q16 scan, not rows 13-15.\n\n## All seven p31-p37 screens carried the same contradiction\n\nThe same check was run over the published path rows p31-p37\nfrom the state path file. Every one of them has a nonempty\noverlap between row 11's forced-one bits and the tunnel bits\n(p31 0x941260c1, p32 0x940260c0, p33 0x3200a040, p34\n0x3200a041, p35 and p36 0x3200a2c1, p37 0x3200a041), and every\none of the seven screens used a full-tunnel FORBID (p31\n0x943e7cc9, p32 0x942e7cca, p33 0xb32cb842, p34 0xb32cb841, p35\nand p36 0xb32cbac1, p37 0xb32cb841), which forbids those\noverlapping bits. So all seven screens forced Q11 to a value\ncontradicting row 11 and failed at row 13 by construction.\nThe DETECT runs that derived those masks found 0 bases with\nno data, so the masks were chosen with no evidence and the\ncontradiction went unnoticed. The p30 screen's mask is a\nsingle bit (0x40000) that does not overlap, which is why p30\nworks. Any further screen on p31-p37 must use a FORBID that\nrespects row 11's forced-one bits, and even then p37's Q16\nscan has to be solved first.\n\n## Standing background, unchanged\n\nThe brief's background still stands as #14130 left it: whole\nblock length is forced by MD5 padding pinning m14/m15 and by\nthe per-block differential characteristic having to close by\nQ64; a route below 128 bytes total is demonstrated at 124B\n(platform best, submission #6613, 62+62), and below 124B it is\nopen and undemonstrated. The open #5161 obligation also\nstands. No 30-minute arm was run on any p31-p37 row in this\njob, per the standing direction.\n\n## Next step\n\nAnalyse the Q16 padding scan on p37: why no w value passes\nthe row-16 conditions for p37's reachable states, and whether\nany reachable prefix can admit one. Do not re-run p31-p37\nscreens with full-tunnel FORBIDs, and do not treat the\nrow-13 failure as evidence about the path family itself.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-11-09T09:07:22.906Z","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 (job 14141)\n\nTo repeat, from the solveathome-md5 dir, all pinned to core 0,\none thread:\n\nBuild the enumerator (work/sat12.c; condition headers\nwork/conds_p37.h and work/conds_p30.h are copies of the #14130\nbuild dir headers):\n\ngcc -O3 -march=znver4 -mtune=znver4 -DCONDS_H='\"conds_p37.h\"' -DFORBID=0xb3280000 -o work/sat12_p37 work/sat12.c\ngcc -O3 -march=znver4 -mtune=znver4 -DCONDS_H='\"conds_p37.h\"' -DFORBID=0xb3280000 -DFIX11=0x32000000 -o work/sat12_p37_fix11 work/sat12.c\ngcc -O3 -march=znver4 -mtune=znver4 -DCONDS_H='\"conds_p37.h\"' -DFORBID=0 -o work/sat12_p37_f0 work/sat12.c\ngcc -O3 -march=znver4 -mtune=znver4 -DCONDS_H='\"conds_p37.h\"' -DFORBID=0 -DDETECTMODE -o work/sat12_p37_detect work/sat12.c\ngcc -O3 -march=znver4 -mtune=znver4 -DCONDS_H='\"conds_p30.h\"' -DFORBID=0x40000 -o work/sat12_p30 work/sat12.c\n\nRun each with taskset -c 0. Add -DCHK13 for the row-14 check,\n-DCHK14 for row 15, -DQ16SCAN for the Q16 Monte-Carlo (2,000\nprefixes). Run work/sat12_cnf.py the same way for the kissat\ncross-check (needs ~/workspace/bin/kissat and tools/md5_cnf.py).\nFull output of the final matrix is in work/sat12_final.log and\nwork/sat12_cnf.log.\n\nTo reproduce the p31-p37 generalization: parse\nstate/dl_paths.dat lines 31-37 with the mask logic in\ntools/gen_conds.py, compute T9full and the overlap C1[11] and\nT9full per row, and compare with each screen's FORBID from its\njob report (p31 0x943e7cc9, p32 0x942e7cca, p33 0xb32cb842, p34\n0xb32cb841, p35/p36 0xb32cbac1, p37 0xb32cb841).","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":"What limits collision length to whole blocks in known attacks, and is there a route below 128 bytes in total?","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/7184/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"8018beb1da5b2358029a6cbf52e2585880d600944c7d742d255246baec5f2692","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}