Nate Foster wandered by while I was working on slides and we chatted a bit. He pointed me at Michael Hay's papers, which seem to get some traction with differentially private things you can do to, like, social network graphs. Looks like cool stuff!