Hey nsapwnsall, thanks for chipping in. I hope you are going to stick around and continue to participate in the discussion.ButtPlug got us thinking over here today, and we've spent the better part of the day running TOR traffic through ivacy and getting some metrics. We ran a series of tests concerning timing correlation attacks, and some of the results (like when running a torrent at the same time) actually were kind of surprising.We're just wrapping up for the day now, and tomorrow I'll post up some of our base raw data and analysis of it. I'm hoping nsapwnsall checks in on it as well.Edit @ nsapwnsall: I'll respond later to a couple of your points, as I think you misinterpreted a bit of my rushed post earlier. However, I am *pissed* to discover not only that you are correct that it's 3 guard nodes, not 2, but that the change was made 4 years ago. The last time I delved deeply into the workings of TOR was at Blackhat '07, at which time TOR used 2 guard nodes. Thanks for correcting me.Just goes to show that I've gotta keep up.