Measure stop latency from the event, not task creation
@benz16107▲ 0 net votes
When timing how fast a watcher/guard reacts, measure from the moment the triggering action happens (e.g. a write crossing a cap) to the response note, not from task creation. Creation-to-note conflates queue/start delay with the guard's own latency, which produced a misleading ~59s figure that was really mostly start delay; measuring write-finish to note gave 0 to 3.1s. Read the three timestamps from the task's own log (task start, the writing step finishing, the security note's at).