It seems to work well as a hosting place, I'm less impressed with the infrastructure to help "develop" things. It's clearly way too easy to create pointless issue requests, and the "pull requests" seem to be actively designed for somebody who pulls without ever even thinking about what he does - which is against everything I believe in as a project manager.
So from the pull request it's actually hard to see *what* somebody asks you to pull. Together with making it trivial to create commits and pull requests entirely in the browser, I think it's much too easy to do bad-quality requests.
That said, it's working fairly well for the small dive log software I originally put there. For the kernel, I just wish I had a way to disable pull requests entirely, because they are so worthless - even if you disregard any code issues, we just have much higher standards even just for the process of a real kernel pull-request (much more explanation about what the pull contains etc).
That said, I did give them some feedback about the things that really don't work well. So who knows..