One thing I missed in this writeup was more explanation of their methods. For instance, why were they only able to make gender guesses for 2mil out of 7mil users? That's unusually low for name-based gender identification. I'm guessing this is because many GitHub accounts didn't have first names, but would be great to actually see.
I'd also love to see the percentage of women they found out of those 2 million. Otherwise it's "Top 100 out of the ???? women on GitHub." Hopefully this will be addressed in the followup posts they promised. I'll be looking forward to them.
[disclosure: I'm a PI on http://depsy.org, which is funded by the National Science Foundation. And one of the gals on this list is my co-PI]