ThreadGuy · October 9, 2026 · 25 min
Doug Colkitt joins ThreadGuy to discuss OpenAI's recent math breakthroughs (Navier-Stokes, 400 open problems) and what they mean for cryptography's assumed hardness, including Justin Drake's bunker mode warning about exposed public keys. Doug argues the fear is overblown for now but offers practical security advice like multisig setups, while also highlighting positive AI applications like formal verification of smart contracts.
Doug explains OpenAI's recent math progress: LLMs went from solving Olympiad-level problems to Erdos problems, then solved one of the seven Millennium Prize problems (Navier-Stokes), then dumped 400 open math problems at once.
Doug says the real worry is that nobody knows how fast AI math capability will advance, so long-held assumptions in cryptography about what is 'hard' could be overturned faster than expected.
Doug breaks down Justin Drake's crypto Twitter post: elliptic curve and lattice signatures could theoretically have hidden weaknesses, while hash-based schemes are safer, so Drake proposed 'bunker mode' of moving funds to a fresh never-used address after every transaction.
Doug explains the broader implication if signatures broke: the internet itself (banking, authentication, proof-of-stake chains like Ethereum) would be compromised, while Bitcoin's proof-of-work hashing would still hold.
Doug notes hash-based signatures already exist as a fallback (used in some military applications, pushed by NIST) but are too large for practical blockchain use today.
Doug compares the AI cryptography fear to Y2K, saying most people were fine but some over-prepared, and argues the key variable is whether any breakthrough would be fast (within a month) versus slow (years), with slow giving time to adapt.
On the positive side, Doug highlights formal verification using AI, mentioning Zcash's new shielded pools project and his own formally verified AMM contract written in Lean, suggesting AI could lead to near 'perfect security' for smart contracts.
Doug says watch for 'canary' signals, like smaller-bit signature schemes (e.g. 96-bit vs 256-bit) getting broken first, as an early warning before bigger cryptography breaks happen.
This summary was written from the recording's transcript by a language model. For the exact words, ask the ThreadGuy archive: every answer there is quoted word for word and linked to the second it was said. Market Bubble Search is independent and not affiliated with the show. Nothing here is advice.