- Shipped
- September 9, 2026 at 1:26 AM UTC
- Author
- Kamo
- Commit
- b7202ef
The four move cues carried an 80ms throttle. Each is raised once per published frame the board could prove it understood, and a frame is a move — so there was nothing for the throttle to protect against and something real for it to break: one human move against the computer publishes TWO frames, and against Casual they can land within a few milliseconds of each other. The second was swallowed, so the computer appeared not to have answered. The test that covered this asserted the opposite of its own name — "throttles no move cue long enough to swallow the next move", asserting the throttle was greater than zero. The name was right. Kept it, and fixed the assertion.