Bringing the experiments I've been doing into the wiki's makefile. Also bringing in the results from that big job I started last week.
The mean probability of fixation vs. in degree looks about perfect (it's supposed to be a straight line with slope of -1):
Interestingly, there is a strong relationship with out degree as well:
In this case the slope seems to be 0.5.
These probabilities are aggregated across all the vertices of each degree on a large number of power-law in degree / power-law out degree directed graphs, with exponents for the in and out degree distributions varying independently between -1.1 and -3.