How secure is an anonomous mail-server

[Peter Honeyman]

one way to defeat traffic analysis is to maintain a steady stream of
traffic.  to this end, avi and i are considering a "remailer tax" -- if
you send a message through a remailer, you pay a tax of (say) ten
additional messages sent and ten received.  so if you send a message
via a remailer, your software is obliged to send out an additional ten
nonce messages, spaced out over time.  that's not so onerous ... but
those messages have to *go* somewhere, so you will also be taxed by
receiving ten nonce messages.

this remailer tax is not completely thought out ... what do you
think?  if remailers catch on, the nonce traffic can be cut back.

i'm also still enamored of the probabilistic remailer notion i proposed
a few months ago, where a remailer flips a coin to determine whether to
deliver a message to it's destination or whether to throw it back into
a pool of like remailers.

	peter