On 1 July 2013 01:55, Jacob Appelbaum <ja...@appelbaum.net> wrote: > > > I would like to see a tor configuration flag that sacrifices speed for > > anonymity. > > You're the first person, perhaps ever, to make that feature request > without it being in a mocking tone. At least, I think you're not mocking! > :) >
I would second that, it would be a desirable feature. As it happens, I have been pondering this very problem for a while now, even before information came to light about GCHQ's pervasive tapping of fibre cables. While I doubt any government agency is at the moment running any decent traffic analysis on the Tor network - as was alluded to in previous posts, it's hardly worth their while at the moment - conceptually it wouldn't take a massive leap to do so. If you have visibility of a large proportion of the internet with very accurate time stamps, it will almost certainly be possible to break the anonymity protection that Tor currently provides. There are some naive models that can combat that type of traffic analysis but they all introduce new problems as well. For example, if one creates a new mode of operation so that nodes forward entire messages instead of packets and that those messages have a lower and upper bound delay field, it would seem on the face of it that one could thwart traffic analysis because the data forwarding times are almost completely disassociated from the sender. However, because it is a larger message instead of packets, a new statistical bias is introduced in terms of message size and reduction in frequency of forwarding events. So in this naive model, it may actually have made the situation worse. So, yes, being able to sacrifice speed for improved anonymity is a desirable feature but I doubt it's going to be particularly easy to design or implement. There's also the problem of having applications that can utilise a mode of operation that has potentially much higher latency.
_______________________________________________ cryptography mailing list cryptography@randombit.net http://lists.randombit.net/mailman/listinfo/cryptography