Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!henry
From: henry@utzoo.UUCP (Henry Spencer)
Newsgroups: net.crypt
Subject: Re: Code Breaking
Message-ID: <6649@utzoo.UUCP>
Date: Sun, 4-May-86 00:26:32 EDT
Article-I.D.: utzoo.6649
Posted: Sun May  4 00:26:32 1986
Date-Received: Sun, 4-May-86 00:26:32 EDT
References: <113@radha.UUCP>, <331@brl-smoke.ARPA>
Organization: U of Toronto Zoology
Lines: 19

> The second most hilarious encryption scheme I ever heard of used just
> a short key to get started, then when the key ran out it used the
> already-encrypted plaintext to provide the rest of the key...

This and variations on it are known as "auto-key" systems, as I recall.
They actually aren't bad if the bad guy has *no* *idea* that you might
be using such a scheme.  If he once suspects it -- either from hearing
about your scheme, or from breaking part of a message and noting the
self-similarity -- it's all over.

In a not-really-related vein, one way to make your messages hard to break
is to do data compression on them first.  Cryptanalysis exploits redundancy;
data compression considerably reduces redundancy, and makes the patterns
of the remaining redundancy much more subtle.  HOWEVER... don't use a
compression method that starts its output with a distinctive header, or
you've just handed any smart snoopers the keys to the vault.
-- 
Join STRAW: the Society To	Henry Spencer @ U of Toronto Zoology
Revile Ada Wholeheartedly	{allegra,ihnp4,decvax,pyramid}!utzoo!henry