[joestringer]

I'd be curious to see the same analysis done but with alternate keyboard layouts, such as dvorak or colemak.

[lmm]

It'd come up exactly the same, unless you're going to pick out the dvorak-users on github. Even then I wouldn't expect much difference. We're measuring characters in code, not actual keypresses.