pub fn record_forge_decision(seats: usize, bots: usize, elapsed: Duration)Expand description
The rules work between a seat answering and the next prompt appearing, split by room shape. Measured over two days of captures, a room with four seats and three bots runs this at a p99 of about 2.9s against 0.35s for every other shape, and puts 2-5% of decisions over two seconds where the others put none. A fleet-wide quantile averages the many clean rooms against the few bad ones and shows neither. Bot decision time is not the problem and is not recorded here: it sits at a p99 of 0.75s in exactly those rooms.