As I have said before, sometimes you actually have to play the game. An interesting thing that surprised me years ago was that the time until fixation of a highly deleterious allele was lower than a neutral allele. In other words, if you had an allele with negative fitness implications the time it takes to traverse 0% to 100% on the frequency chart for the population will be shorter than for one with no fitness implication (neutral fixation in generations is 4 * effective population size). Why? Because for a highly deleterious allele to fix it has to do it quickly, selection works against it over the long run. The expectation of fixation is low for these alleles, but when they do fix they are characterized by speedy action. Low effective populations resultant from inbreeding is a case where deleterious alleles fix, the power of selection in this case is swamped by the stochastic effects of drift.