What foundational concept in theoretical computer science broke, and how did it happen?
Paid in full — The Exponential Time Hypothesis (ETH) (which establishes complexity lower bounds for over 800 computer science problems) was theoretically disproved when an OpenAI model devised a subexponential algorithm for 3-SAT running in roughly 2^(n / sqrt(log n)) time. However, the algorithm only beats brute force when n exceeds 10^10,000, leaving practical applications unaffected.
Evidence
Title — claimed “The one thing holding theoretical CS together just broke... kinda”
Basis — The transcript demonstrates that an algorithm solving 3-SAT in subexponential time was found, breaking the strict 1999 formulation of ETH while acknowledging the minimal practical impact.








