Anonymity -> Untraceability -> High Latency?

[Nelson Minar <nelson@santafe.edu>]

I've been trying out various mechanisms for anonymity: remailer
chains, HTTP proxies. There's one problem that makes them inconvenient
to use regularly: latency.

A good Type I remailer chain takes at least an hour to deliver email,
instead of the 15 seconds I'm used to. Mixmaster-style takes even
longer; the delay is important to the security of the system.
Forwarding all my HTTP requests through a proxy adds an extra hop and
some processing to all web transactions, noticeably slowing down
browsing. I'm not much for waiting for computers.

The problem is that that these anonymity schemes rely on untraceability.
And to be untraceable, we have to have centralized servers take our
traffic and forward it along, stripping out identifying information,
burying it in the noise of lots of other traffic. But that forwarding
process seems guaranteed to add latency to the communication.

Back in the old days (ie: six months ago, before Web search engines
were big on the scene) I was reasonably happy with the
needle-in-a-haystack anonymity of the unorganized Internet. I posted
personal things to Usenet, fairly sure that only the members of that
Usenet group were going to see my messages. I ftped files from all
over without worrying that my transactions would be logged permanently.

But now, with the amazing success of Web searching, I no longer feel
that obscurity is sufficient security. Are there other approaches to
anonymity that don't impose the latency that forwarding messages
around does?