Redux: (∃ Stochastic Natural Latent) Implies (∃ Deterministic Natural Latent)

David Lorell reported on LessWrong that using large language models and the Lean4 proof assistant, he's proven a theorem about hidden variables: if a system can be explained by a random internal model, then it can also be explained by a deterministic one—though with a quantifiable tradeoff. The result formalized a proof that had failed a year earlier due to a critical flaw. The proof has a formal constant of one thousand seven hundred seventy-one, though empirical testing suggests the true value is around two. Lorell plans to reduce the constant further and simplify the proof.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton