Maybe it's since I've always worked for companies with > 500 employees, but there is always a pile of work on the products I've worked on that there was little time to contribute externally.
There are so many factors that go into whether or not a repo becomes popular, which aren't necessarily related to anything that reflects on the creator's merits as an engineering hire. I think you'd skew your candidate pool in a weird way if you leaned on these metrics too heavily. Honestly, I'd be curious to hear more about how it's used in hiring.
At any rate, enter the law of triviality. Making something simple that lots of people will understand is more likely to lead to more stars than something brilliant but complex, yet more indicative of development skill. This selects for gaming of the system and/or salesmanship; no doubt said employers will screech about a lack of "qualified" candidates before long.
I know others have waxed on the irony of the centralized role GitHub plays in a supposedly decentralized world of the DVCS that is git.
It would be cool if there were a Behance or even a Dribbble for programmers: a way for us to present our creations well but also easily in a host agnostic and semi or fully automated way.
Many of us or don't invest the effort necessary to maintain a work portfolio, unlike say design oriented fields where it's the norm. I think there could be something to it.
I mean, I'm always happy to see competition keeping companies on their toes, but why do you actively want people to leave GitHub?
https://hacked.com/github-promotes-reverse-racism-sexism/ http://dancerscode.com/blog/why-the-open-code-of-conduct-isn...
It's network effects are large but beyond that the product doesn't really do anything that others don't do just as well.
I guess I can include things like Aerospike and even Gitlab in that group. I don't actively wish for them to succeed either. I am glad they exist and I actively use their products but I would not cheer for any of them.
GitHub is becoming too large, and code is becoming too centralized, that is dangerous.
Most larger companies will not use a third party cloud service to host their code like github.com. They want software to run locally on their own servers instead.
Github offers their software to enterprise customers to run on their own servers, but they traditionally have done a bad job doing it, at least relative to gitlab. Today, dozens of open projects and even closed ones are using personal gitlab instances, but since the ecosystem is more open on gitlab contributions come back in to make it a much more friendly tool to self-host.
Tons of larger companies favor gitlab for exactly that reason.