1993-10-01 - steganography & fidonet

Header Data

From: doug@netcom.com (Doug Merritt)
To: cypherpunks@toad.com
Message Hash: ba2935ae191f56303c56d917b429c8e1a9d9a0327076f157e608d4fb5bef766e
Message ID: <9310010206.AA27468@netcom5.netcom.com>
Reply To: N/A
UTC Datetime: 1993-10-01 02:07:07 UTC
Raw Date: Thu, 30 Sep 93 19:07:07 PDT

Raw message

From: doug@netcom.com (Doug Merritt)
Date: Thu, 30 Sep 93 19:07:07 PDT
To: cypherpunks@toad.com
Subject: steganography & fidonet
Message-ID: <9310010206.AA27468@netcom5.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain


In regard to things like Fidonet sysops noting forbidden encryption
pasing through their systems:

Encrypted messages could be encoded in a final pass as English sentences.
A 2000 line C program with a 200,000 word dictionary could encode and
decode sentences that were roughly grammatical, even if they sounded
really weird on average.

Although I've never heard of this being done, it's pretty obvious in one
sense, so I'm sure it's no surprise to the spooks. It surely wouldn't
be obvious to snoopy Fidonet sysops, though, so it may have its uses.

BTW a more complex program + dictionary could confine the encoded utterances
to topical words and therefore sound even less weird. With enough
sophistication, such an encoding could generally pass muster as
"confusing and poorly worded jargon" to anyone but the most devoted analyst.

I've got enough (or almost enough) sw & dictionaries & word clusters on hand
to implement such a thing, but I've personally no purpose to use it for.
	Doug





Thread