Latency vs. Reordering

[hughes@ah.com (Eric Hughes)]

This horse isn't dead yet.

The distinction between latency and reordering is if primary
importance to the cryptanalysis of a remailer network.  To repeat yet
again: reordering provides security and latency is a by-product of
reordering.

I assert that anyone who's given a modicum of thought about how to
cryptanalyze a remailer network understands this distinction well.  I
also assert that those who haven't thought about cryptanalysis don't
understand the distinction, even if they do believe in it by
authority.

One of the oldest maxims in the book is "Don't design ciphers until
you've tried to break some."  A remailer network is intended to be a
cryptographic object, a new kind of cipher.

I assert that if you don't understand the distinction between
reordering and latency, you've not thought enough about the
cryptanalysis of remailers and shouldn't be designing them.

Therefore, in the future, from here on out, I will label the promoters
of latency as "sellers of snake oil."  It's the same fallacy as
creating a new cipher by putting lots of complicated operations inside
it without understanding where the security comes from.

Eric