cjh@CCA-UNIX@sri-unix (10/22/82)
This is true of most biological mutations; the theories describing means of evolution all require a statistical population in which most mutations are either trivial or failures, or else there is enough genetic variation in the population as a whole that in a crisis one tail of the curve will survive whatever wipes out the other tail and maybe the center as well. Given that the programs would presumably be identical initially, the incidence of constructive mutation should (I think) be small---but over the time frame we're talking about, attempting to predict what would happen is foolish.