Actually, that might be because of their antispam filters. They were always fighting websites with countless subdirectories or subdomains, having completely wiped them off their SERPs in the past.

GitHub does fit the "huge website with lots of duplicate content" description very well.

Then again, Twitter seems to be doing fine. But I have a suspicion that they have some sort of agreement (and GitHub doesn't), because I have to regularly "report as spam" (note how that implies Google decides what's spam and what isn't) Twitter emails and they still pop up in the inbox after a while.