Sure, but my point is that most of Github's potential customers are not on Macs.

My point is that their UI could change, and it's easier to test out UX on something that you'll use everyday, than on something that you'll never use. They made Github because they wanted to use it themselves, not because they saw it as a money making scheme.

For productivity apps (and especially developer tools), it's not worth chasing customers on a platform you don't love -- you'll do a bad job and you'll be unhappy doing it.

And as Mac users, they may be consciously avoiding the reverse of an all too familiar situation, where non-Mac-using Windows developers shoddily port their app to OSX.

Right, but you build on the platform you like/know first, then you expand to others.

I get that dogfooding is important, but github employees aren't the target market.

Github employees are the target market, they use Git don't they? What use is developing a windows app that none of the developers will use initially?

Selling software where the target market is exactly the same as the people writing it seems like it's the exception not the rule. I assume most OpenTable devs do not own a restaurant.

sounds like a strawman argument to me...