Hi,

I'm wondering about a particular case--let me explain. From your threat model you assume that the adversary has suspicions about encrypted traffic and may block them without strong justification. You also take as given that the adversary may be state-level. From the adversary objective this is because the adversary wants to know who and what this communication is about. In the limitations you state that the adversary (counter-intuitively) has strong socio-economic reasons to not block bittorent. It does not follow... In China it's not uncommon to hijack torrent sites or ban them entirely. They perform mitm even for encrypted sites like github. They have a one-strike policy that they don't normally enforce regarding file sharing. The golden shield is sophisticated enough to correlate the use of a bridge across multiple users. Which means you need strength in numbers. Then again, outside China, bittorrent is commonly subjected to traffic shaping. I'm unclear about how this helps the censored user. Under such circumstance wouldn't it be possible to have a common peer show up in multiple unique torrent swarms?


--leeroy