Dear list members,

We are happy to present the preleminary analysis results from TTTT:

http://arxiv.org/pdf/1410.1823v1.pdf

The conclusions are:

1. We have found that AS-path prediction techniques significantly overestimate ASes and IXes traversed by Tor traffic, calling into question the results of previous evaluations of the network-level adversary threat.

2. Our simulations show that Tor may currently be less vulnerable to an AS or IXP adversary than has been previously found. We also find that proposed path-selection algorithms to avoid a network adversary still leave a significant chance for users to be deanonymized over time due to the errors in path prediction, and these algorithms lead to significant rates of connection failures, even though in the large majority of cases the failures could be avoided with better measurement.

3. Thus our results suggest the importance of accurate measurement both for understanding Tor security and for improving it.