In a 53-hour solo run, Claude proved new identities that let modular inversion run backward without storing its branch history, giving an elliptic-curve point addition circuit with 17x fewer gates for 12% more qubits, and that is the core step of the quantum attack on Bitcoin keys, so Q-day just got cheaper.
pull down to refresh
While I note the advances in research on quantum computing, I think bitcoin really has taken up the challenge well. There is a lot of ongoing research about new cryptographic algorithms and ways to implement them.
I know that the quantum researchers often say once they figure out how to build quantum computers with more circuits, it will speedily increase to lots of circuits, but they still have yet to produce a quantum computer with very many circuits.
Wild result - 17x fewer gates for 12% more qubits is a massive trade-off improvement. But Q-day getting cheaper on paper is different from Q-day being feasible - we still need millions of error-corrected qubits, and error correction is the real wall. Bitcoin has time to migrate to P2QRH or other post-quantum sig schemes, Taproot already makes it easier to upgrade. Question: does Claude's optimization also apply to Schnorr vs ECDSA equally, or is there a difference in gate count?