garyebickford writes: I have a suspicion that TOR is nowhere near as private as is generally assumed. We can assume that some fraction of all the nodes out there are run by what I'll term 'spies' — entities who want to know things about whoever's using TOR. The question is, what fraction is sufficient to be able to reconstruct missing pieces, and figure out with a high degree of reliabillity what the 'real' source and destination are, assuming those 'spy' nodes can all talk to each other? There is some good math for doing such reconstructions of networks where most of the nodes are unknown. I suspect that the necessary fraction is somewhere near 10%. It's quite possible that your friendly neighborhood 3-letter spook shop knows a lot more about what's going through the TOR network than any of us, the great unwashed, realizes. So, how much of the TOR network needs to be 'cooperating' to significantly compromise privacy?