Do You Know Where Your Headers Are? Comparing the Privacy of Network Architectures with Share Count Analysis

Online privacy is more important now than ever. Using encryption goes a long way by hiding application data from third parties, but some amount of private information is still exposed in packet headers. Tools like Tor are designed to address this concern, and, more recently, research efforts to replace IP have begun to consider how a network architecture itself can improve privacy. Unfortunately, we do not have a good way to quantitatively compare network architectures (in terms of privacy and in general). We take the first steps in this direction by presenting share count analysis, a methodology for measuring "how private" an architecture is. We describe our design and implementation and present initial results indicating that this approach is promising.