I'm using "drift" and "genetic drift" interchangeably. The terms have gotten hard definitions from the population genetics literature that has had to isolate the variables clearly so that evolution can be expressed mathematically. *Mutation* pressure comes from the rate at which a particular mutation enters the population. Genetic drift is a purely statistical affair. It is like tossing a coin a number of times. You expect to get exactly 1/2 heads and 1/2 tails, but you will rarely get that in any particular series of tosses. Some sets of 10 tosses, for example, will give you 6 and 4; some 7 and 3 or even 2 and 8. This isn't a selection pressure, but in a small enough population, it can *effectively*act like one. If the population is very tiny, it can "fix" the "wild type" allele (that is, put it in 100% of the population) or it can "fix" any mutation to that allele, just because of its stochastic nature. In a larger population, it can cause fluctuations from generation to generation, resulting in smaller (percentage-wise) "random walks" of the population.