From: Edward J OConnell <ejo@world.std.com>
To: Graham Toal <gtoal@an-teallach.com>
Message Hash: 7f3f706c40804c7b3ac91e9fc5c59fef37e1d28f5a6afdc6848fb7e5dd28ee48
Message ID: <Pine.3.87.9310291722.A1566-0100000@world.std.com>
Reply To: <4477@an-teallach.com>
UTC Datetime: 1993-10-29 21:39:07 UTC
Raw Date: Fri, 29 Oct 93 14:39:07 PDT
From: Edward J OConnell <ejo@world.std.com>
Date: Fri, 29 Oct 93 14:39:07 PDT
To: Graham Toal <gtoal@an-teallach.com>
Subject: Re: ID of anonymous posters via word analysis?
In-Reply-To: <4477@an-teallach.com>
Message-ID: <Pine.3.87.9310291722.A1566-0100000@world.std.com>
MIME-Version: 1.0
Content-Type: text/plain
One could also imagine a 'symantic scrambler' analogous to the word
analysis program, but designed to defeat it, by randomly altering the
syntax of a post. The output might have to be tweaked afterwards, in
order to restore some sense to it, but it would be a sort of ascii
version of the cutting the words out of magazines style of ransom note.
It might just be a huge lookup table of canned phrases that get swapped
in to replace your 'unique,' identifiable sentences. It seems like a
dedicated global search and replace, combined with some sort of die throw
to dictate sentence structure might be enough to screw up word analysis,
actually...
Actually, the scrambler wouldn't have to be random--it could always
produce output with the same word analysis signature. As long as a bunch
of people were using it, or had access to it, you'd have deniability.
E. Jay O'Connell____________________________________________________
"God does not play dice with the Universe"--A Einstein
"No, she plays SuperScratch-Card Wingo (TM)"--Me.
____________________________________________________________________
Information Wants to Be Free PGP Public Key available by Finger
Return to October 1993
Return to “Matthew J Ghio <mg5n+@andrew.cmu.edu>”