[liberationtech] Jacob Appelbaum's Ultrasurf Report

[Meredith L. Patterson]

On Tue, Apr 24, 2012 at 4:29 PM, Jacob Appelbaum <jacob at appelbaum.net> wrote:
> In the case of remailers, I view those as a different class of anonymity
> - namely, I have long viewed that if you use such a system, you should
> probably compose it with Tor - so few people use them that I'd want to
> disguise that I was using them. Also, you cannot use those systems for a
> lot of TCP communications that people regularly need to protect - has
> anyone written an XMPP or AIM client that uses mixmaster? I suspect not...

Nobody would; they'd be unusable. Mixmaster and mixminion are
extremely high-latency for a reason: a delay of 20 minutes to an hour
is perfectly acceptable for most email use-cases, and makes it far
more difficult to deanonymize the sender by timestamp correlation. But
that sort of delay is right out for anything with any kind of
real-time service expectations.

Andrei Serjantov's paper "From a Trickle to a Flood", available at
http://freehaven.net/doc/batching-taxonomy/taxonomy.pdf, describes
remailer mixing and pooling strategies in far more detail, and is
required reading for anyone interested in how high-latency anonymity
systems work and why they work the way they do.

Cheers,
--mlp