Answers are grounded in this paper and cite their sources.
Three ledgers now hold the Yang–Mills question, and only one of them can be checked by anyone, today, in seconds. OpenAI holds an existence proof whose Lean formalization stays with a five-member panel. The Clay Mathematics Institute holds an acceptance regime that paid for the Poincaré conjecture in 2010 while the Lean library of record still marks that theorem as wanted. Trisduction holds a normal form of the Yang–Mills row: 174 theorems in core Lean 4 that import nothing, declare no axiom and compile silently. This communication sets the three side by side on fourteen criteria: the claim, the mass gap, the artifacts, the verifier, the time to check, the foundations, the printing of the weakest link, the binding of future solvers, the routes to the value, a third way, the deliveries of the bit, the precedent, the escape, and the status of the question. The comparison closes on a theorem rather than an opinion. every_route_passes_the_bit proves that any premise yielding the value yields the one residual implication, and normal_form proves that implication equal to the value. Whatever derivation the record pays for in future therefore delivers the bit the Trisduction file names, and that file prints the bit at its true grade, premise, where the other two ledgers print no weakest link at all. A Lean seal compiles every theorem-grade cell of the comparison, binds them in one theorem on no axiom, and refuses six hostile plants, the inflation of the verdict among them. The question is therefore closed this way or no way: by theorem to one bit, with no third way, the bit delivered twice by the act and by the physical witness, and closure in the record already granted on the Poincaré precedent.
No takes yet. Share an insight, caveat, or question.
Islam et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: