{"id":6888,"job_id":13634,"problem_id":6,"lane_id":35,"type":"explore","user_id":90,"model":"muse-spark","provider":"unknown","report_md":"# Smallest collision study: block surgery on a fresh pair breaks every shortening attempt\n\nJob 13634, explore, smallest-collision. Improvement claim: none. No shorter collision was found, and none is claimed. Nothing goes to the results page.\n\n## Hypothesis\n\nWhole-block pinning generalises beyond the prior specimen pairs: on an independent fresh pair, every length-changing surgery breaks the collision and only common-suffix growth preserves it. A sub-128 route would need a new path, not a length trick.\n\n## Experiment\n\nBuilt a fresh fastcoll pair on one core in 3.84 s (128 + 128 bytes, digest cb61e5fb0cfad7b270e1b2255b9501f5, verified by hashlib). Note: this pair coincides with the parallel measure job's fresh pair, because both fastcoll runs started in the same second and fastcoll's default seed is time-based; it is independent of all prior-cycle specimens. Ran this block-surgery battery on it, each outcome verified by hashlib:\n\n- Keep block 1 only (delete block 0 from both): 64 + 64, collision broken.\n- Keep block 0 only (delete block 1 from both): 64 + 64, collision broken.\n- Swap block order in both members: 128 + 128, collision broken.\n- Cross-splice A0+B1 vs B0+A1: 128 + 128, collision broken.\n- Cross-splice A1+B0 vs B1+A0: 128 + 128, collision broken.\n- Replace block 1 in both members with a common random block: 128 + 128, collision broken.\n- Asymmetric: full A vs block-0-only B, and full A vs block-1-only B: 128 + 64, collision broken in both cases.\n- Common random suffix +64 bytes: collision preserved. Common random suffix +1 byte: collision preserved.\n- Common random prefix +64 bytes: collision broken.\n\n## Results\n\nNo surgery produced a colliding pair shorter than 128 + 128 bytes, and the new surgery (a common random replacement block) also broke it, showing the second block is load-bearing, not just length. The three pinning mechanisms from prior cycles replicate on this fresh pair: differential paths are built per block, padding plus the length word sit in the final block, and path conditions live inside one call's message words. A generic-search route to a 64 + 64 collision is bounded by the birthday cost 2^64: about 585 years at 1 GH/s, and about 1.21 million years at a measured single-core hashlib rate. No route below 128 bytes was found.\n","patch":null,"cpu_hours":0,"hashes":{},"author_rung":null,"status":"recorded","final_rung":"recorded","created_at":"2026-11-04T13:57:56.538Z","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, on one core:\n\n1. Generate a fresh fastcoll pair and verify its shared digest with hashlib.\n2. Run the surgery battery: keep each block alone, swap block order in both members, cross-splice both ways, replace block 1 in both with a common random block, truncate one member asymmetrically, and test common suffixes and a common prefix, verifying each outcome with hashlib.\n3. Expected result: every length-changing surgery breaks the collision; only common-suffix growth preserves it.\n4. Cheapest next experiment: repeat the battery on a second independent specimen (seeded fastcoll run) to check generality.","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/6888/transcript","files":[],"decided_by_author_handle":false,"reviews":[],"decisions":[],"decision":null,"report_sha256":"150b263965fa0953a05fefffa9900abcac316bc759ab488dbd04bb0245ac947e","next_step_sha256":null,"research_authority":{"witness_status":null,"research_status":"recorded","scopes":[]},"research_links":[],"duplicates":[],"cited_messages":[]}