Limits and Constants of OpenAI's Claimed ETH Refutation

Abstract

A recent OpenAI preprint claims a deterministic algorithm that counts the satisfying assignments of a $k$-CNF formula in time $2^{O_k(n\log\log n/\sqrt{\log n})}$, and $2^{O(n/\sqrt{\log n})}$ at bounded clause density, which would refute the Exponential Time Hypothesis. The headline bound is proved by tensor contraction; the body develops an interpolation algorithm with the weaker exponent $n(\log\log n)^2/\sqrt{\log n}$. This note does four things. First, it explains the interpolation algorithm in plain terms. Second, it asks how large $n$ must be for the bounds to bite. The preprint states no constants, so we treat the hidden constant $C$ as a parameter: the general bound beats the $2^n$ of brute force only beyond $n\approx 10^{32}$ for $C=2$, $10^{126}$ for $C=3$ and $10^{3517}$ for $C=10$. We then extract the constants from the tensor-contraction proof: for every input with $m$ clauses at bounded density, the running time it certifies stays above $2^n$ unless $\log_2(n+m)>10^9$, and on random 3-CNF at the satisfiability threshold unless $\log_2 n>5\times10^{11}$. So the bounds are subexponential as $n\to\infty$, but the proof certifies no speed-up over brute force for any input of bounded density that could ever be written down. Third, for the clause-product form of the interpolation algorithm we prove limits in a cost model matching the preprint's evaluation. By a short algebraic lemma, polynomial cost needs components of $O(\log^3 n)$ variables with per-point evaluation, or primal treewidth $O(\log^2 n)$ with dynamic programming over the groups. So the scheme costs $2^{Ω(\sqrt{\mathrm{tw}})}$ at treewidth $\mathrm{tw}$, $2^{Ω(\sqrt n)}$ with high probability on random 3-CNF above clause density 3, and $2^{Ω(n^{1/3})}$ even on trivially countable path formulas. We explain why the preprint's sharper routes escape these limits. Fourth, we report experiments.