I've also had a number of conversations recently on the state of academic publishing and how it can be improved, so I put together a reddit-based interface to the arxiv pre-print server, where hopefully people can now discuss up and coming papers and let the cream naturally rise to the top: http://arxaliv.org/
I'm also a developer on http://lmfdb.org/ but that's a rather technical site for research mathematicians.