From: jdd@aiki.demon.co.uk (Jim Dixon)
Date: Fri, 12 Aug 94 06:31:02 PDT
To: lcottrell@popmail.ucsd.edu
Subject: Re:  RemailerNet
Message-ID: <5445@aiki.demon.co.uk>
MIME-Version: 1.0
Content-Type: text/plain


In message <199408120649.XAA07108@ucsd.edu> Lance Cottrell writes:
> >Actually, the odds are better than this, .8^5, about 0.33.  You will be
> >compromised "only" 1/3 of the time.
> >
> >But if you are sending regular messages to another party, then traffic
> >analysis will quickly show that you are communicating, because even if
> >the boys at Langley are really dumb, you won't make send more than
> >two or three messages without having all the cherries lining up.
> >
> >You will be protected if you have encrypted your messages, but using
> >a remailer network offers little additional protection.
> 
> I am not sure I see why you think that the "cherries" will line up.

Remember that the original assumption was that you were choosing five
remailers at random, on each transmission. I argue against this
strategy; I think that if you know someone is reliable you should
stick with them.

80% of the remailers are compromised, so 2/3 of single messages get
through OK, 45% of two message sequences, 30% of three message
sequences, etc.

If a population of users selects five remailers at random and sticks
to their initial selection, 1/3 of the population will be compromised
immediately.  Langley will have to try harder to get the rest.
--
Jim Dixon