There’s been a fair bit of discourse about the proof online – here are some bits that I found helpful.
Presentations
The original proof writeup is fairly dense, so a few people have been helping with clearer presentations.
Mark Sellke made some slides to present the openAI proof in a more readable way
https://msellke.com/Files/ETH-
Ryan O’Donnell is rewriting the openAI proof more directly as a matrix/permanent calculation
https://www.cs.cmu.edu/~
Shiva Kintali wrote up a simplified exposition as well
https://shivakintali.github.io/papers/ETH.pdf
Providing background context
Other folks have been working to provide some context for the result.
Lance Fortnow has a more technically oriented description of the result and its implications
https://blog.computationalcomplexity.org/2026/10/the-exponential-time-refutation.html
Levent Alpöge made an X post on the Anthropic proof that noted that that proof was generated using 14M output tokens from an internal model.
I’m sure there are other things I’ve missed, so feel free to add other pointers in the comments!
One reply on “Links To Other Work”
Still more things being posted – lecture from Ryan O’Donnell just went up on Youtube!