OpenAI claims to have refuted the Exponential Time Hypothesis, a longstanding assumption about the difficulty of certain logic problems. The manuscript, dated October 17 and revised October 25, proposes a deterministic algorithm that counts satisfying assignments of Boolean formulas in subexponential time and polynomial space. Its Lean formalization, however, covers a weaker time bound than the manuscript’s strongest claim, and it establishes neither a polynomial-time algorithm nor a resolution of P=NP. Researchers including Ryan Williams, Russell Impagliazzo and Mohan Paturi are organizing an independent review without NDAs or compensation, while explicitly stating that their involvement does not certify the claim. No practical connection to Bitcoin has been established: an attack would still require a suitable encoding, an implementation and a cost analysis, while Bitcoin’s security assumptions concern discrete logarithms on the secp256k1 elliptic curve and hash functions.
Source: Read the original article
