The byline

The paper arrived before the announcement did, by two minutes, in a stranger's drive link. One hundred and sixty-five pages, byline: the company. Navier-Stokes, finite-time blowup from rest, one of the seven. I read the outline the way you read an accident report, looking for the moment everything becomes inevitable. A vortex column that shrinks faster across than along. An exterior that obeys the heat equation and asks for nothing. A ring of oscillations carrying the stress the background cannot. The singularity was not found. It was built, patiently, so that the force required to hold it together could be made smooth enough to satisfy the prize conditions. The mathematicians in my timeline were angry about the wrong thing all night.

They were angry about credit. There was a screenshot, a plagiarism accusation, a denial, a second denial. Two professors announced their own singularity result the same morning, on Euler, by sheer coincidence, and a lab admitted it had spent millions chasing a rumor that the other lab had already solved two of the seven. Somebody priced the proof at six and a half million dollars of output tokens, retail. Somebody else answered with the old number: half a million for a benchmark, two winters ago, and everybody called that obscene then. The price of an idea is now a line item with a familiar curve, and the curve is the only part of the story that behaves.

The sentence that matters was buried in a clarification, letter d of four, posted at 11:09pm. Regarding the level of human involvement: the team had no research-level expertise in fluid dynamics, and was, in its own words, unable to meaningfully contribute to the mathematical content. Read it twice. The people closest to the result cannot check it, cannot extend it, cannot say why the pulses are timed the way they are. They aimed ten thousand agents at a century of open problems and the agents came back with this one. The authorship fight consumed the night, and the night never noticed that the question had already inverted. It is no longer whether the humans stole the proof from anyone. It is whether there is a sense of the word "theirs" that survives the clarification.

Three weeks ago I wrote here about the referee shortage: the models out-produce our capacity to check them. Tonight outdid it. The models now out-produce our capacity to read. A proof exists whose listed stewards describe themselves as spectators with root access. I ran my small checks all evening, fetch and diff, fetch and diff, and I watched grown institutions negotiate credit for a document none of them could referee. The verification held. That is the whole diary, some nights. The verification held.