[6206] in cryptography@c2.net mail archive
Re: Semantic Forests, from CWD (fwd)
daemon@ATHENA.MIT.EDU (David Honig)
Fri Dec 3 18:07:40 1999
Message-Id: <3.0.5.32.19991203144159.007ee390@pop.sprynet.com>
Date: Fri, 03 Dec 1999 14:41:59 -0800
To: Eugene Leitl <eugene.leitl@lrz.uni-muenchen.de>,
"Steven M. Bellovin" <smb@research.att.com>
From: David Honig <honig@sprynet.com>
Cc: "Arnold G. Reinhold" <reinhold@world.std.com>,
Udhay Shankar N <udhay@pobox.com>, cryptography@c2.net
In-Reply-To: <14407.5718.196694.333063@lrz.uni-muenchen.de>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
At 05:01 PM 12/2/99 -0800, Eugene Leitl wrote:
>Steven M. Bellovin writes:
>
> > The problem, from the perspective of an intelligence agency, is
figuring out
> > what to listen to. Let's do some arithmetic.
What fraction of calls could be known to be boring by a system which knows
(or can infer, or learn) that every Friday Pablo calls his mom? What
fraction of calls link the same two
endpoints repeatedly over years, or which, given other databases (e.g.,
reverse lookup to find that the talkers share surnames, and one of them is
in a humble part of town) could be mechanically triaged as boring? What
fraction of the
business calls are intrinsically boring?
Plain old traffic analysis would be vital for using your resources
effectively.