Re: tagging attacks and forward-message/reply-message distinction

[George Danezis <gd@theory.lcs.mit.edu>]

On Wed, 3 Apr 2002, Zooko wrote:
> 3.  It isn't good for forward-travelling and reply messages to be distinguishable
> (although I haven't seen any explanation about how *bad* it might be).
> 

Basically there is nothing fundamentally "bad" about them being 
distinguishable. 

In case intermediate mixes can tell them apart, it is just the case that 
you immediately cut your anonymity set in two sets. If each mix wants to 
offer a particular amount of anonymity this might increase the latency 
for example (by a factor of 2). Of course it also adds to the complexity 
of handling them since there would be 2 different mechanisms to deal with 
packets (for intermediate mixes).

It could always be the case that they are indistinguishable for outsiders, 
so that at least only corrupt mixes can take advantage of any 
possibility to perform traffic analysis attacks.

George