OpenAI faster FFT practicality debate

Ben Recht’s October 12 critique challenges the practical meaning of an OpenAI preprint claiming a small theoretical improvement over the usual fast Fourier transform bound. Recht argues that enormous hidden constants and a model allowing exact complex-number arithmetic make the method impractical on real computers. OpenAI’s September 25 preprint remains a theoretical claim; the criticism does not by itself establish that its proof is false. Scientific American had earlier described the claimed bound as a break with a longstanding barrier while noting that its improvement is tiny. An October 13 Hacker News discussion shows a live disagreement: some readers share the concern about practical speed, while others argue that impractical theoretical results can still guide later research. No usable faster implementation has been verified here.