"If humans turn out to be near the Kondrashov limit —that is, if on average every gamete has one bad mutation created during the lifetime of its producer— it is obviously not going to be nearly enough to test a baby for the subset of the few hundreds or so of well-characterized genetic defects"