The Moran process: the simplest possible selection process in a population of two traits A and B. In each generation, a random part of the population is chosen to reproduce at the expense of the opposite trait.
The take-home message is this: random genetic drift always leads to one species supplanting another, but only if the drift is sufficiently particulate. As the drift becomes more and more continuous, so the duration of the fixation process approaches infinity.
To test this, start with mu=1 and notice that the simulation quickly runs to fixation (either A or B equal to 10). mu is the quantity of population that mutate between A and B in a single generation. For mu=1, this means all individuals in either the A or B population transfer to the other side in the first generation, so clearly fixation occurs in the first generation.
But now reduce mu to 0.5, and notice that fixation takes a little longer. Then try mu = 0.25, 0.2, 0.1, 0.05 and 0.01 and notice how long fixation takes.
This was Gregor Mendel's great insight: genes are particulate. If they were not, traits would not fixate, but would instead smear over time into averaged trait values.