Christine Hine is a good place to go as said,
dana boyd is also good, and Ray Kozinets is really excellent. David Miller at UCL has started doing a bit of stuff too, and he's a more traditional anthro person.
If you're more interested in the quant side of things, Mike Thelwall is a really great place to start and Richard Rogers, from a slightly more theoretical vein though still doing actual research, is also excellent. It's much more difficult though, lacking any semblance of a traditional sample causes a lot of problems, as well as a lack of any real demographic information, which makes analysis pretty tricky.
In terms of institutes, the Oxford internet institute has lots of good staff, Microsoft and Google's research arms are always worth a look, UCL and Goldsmiths are both developing pretty strong programs in digital research and Sciences Po is starting to do a lot of work under Bruno Latour (some theory, lots of good empirical stuff too).
There's a hell of a lot of material to draw on, as you've probably found out, and it is still new, which is both good and bad because it allows you to experiment but makes it quite difficult to gather together a consistent approach/position. There's also so many ethical problems it'll make your head spin.