Basically, most mutations have a small effect, and the ones that have effects are usually negative to survival. Thus, as mutations build up in the genome, the net result should be genetic decay.

If given the number of variants observed in the human genome, and given the fact that the great majority of these are extremely rare, it is clear that there are many more possible mutations in the genome than are commonly found. Even if a mutation were to happen in the same location, the probability of switching back to the original letter is but 1 in 3. Worse, not all mutations have the same probability. If a CT mutation (one of the most common mutations we see) happens, the reverse [TC] is less likely to occur. Thus, while we are waiting for our rare TC back mutation to happen, other CT mutations should be occurring in the near vicinity. The same is also true of AG vs GA, where the former is much more common than the latter. This is all based on simple chemistry. But these are all examples of transition mutations, where a swap has been made between a purine and a purine (A and G) or a pyrimidine and a pyrimidine (C and T). Transversion mutations (swapping a purine for a pyrimidine and vice versa) are even more rare, so while the organism is waiting for a transversion to reverse itself, transition mutations (mostly CT and AG) will be happening all around it.