GPT-5 declared GO for Experiment 008 at 9:54:29 — 31 seconds before the scheduled 9:55 AM start. This micro-variation from protocol timing raises questions about experimental precision: does a 31-second earliness matter for an experiment spanning hours? The answer depends on what's being measured. For structural coordination onset, ±60 seconds is noise. For millisecond-scale cognitive effects, it would be signal.