Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!ames!hao!gatech!linus!mbunix!ecp
From: ecp@mitre-bedford.ARPA (Eric C. Pivnik)
Newsgroups: sci.crypt
Subject: Re: One time pads?
Message-ID: <25197@linus.UUCP>
Date: 24 Feb 88 14:34:01 GMT
References: <4209@june.cs.washington.edu> <1988Feb15.151522.5094@utzoo.uucp> <7272@brl-smoke.ARPA> <4104@hoptoad.uucp> <7315@brl-smoke.ARPA> <858@astroatc.UUCP>
Sender: news@linus.UUCP
Reply-To: ecp@mitre-bedford.arpa (Pivnik)
Organization: The MITRE Corporation, Bedford, Mass.
Lines: 34

In article <858@astroatc.UUCP> jojo@astroatc.UUCP (Jon Wesener) writes:
>
>	How about using the message itself, in some form or another,
>as the pad.  For example:
>
>	ABCDEFGHIJ       klmnopqrst
>	?ABCDEFGHI	 ?ABCDEFGHI
>	----------	 ----------
>	klmnopqrst	 ABCDEFGHIJ
>
>The 1st lines the message, the second line is your pad.  You give the
>pads 1st character as the key.  To decrypt it, you provide the first
>character and then use what you've decrypted as the rest of the pad.
	...
>jon wesener


The definition and the entire security of the one-time pad is that 
the KEY is random.  Having the key be obscure, statistically random, 
pseudorandom, or random enough, is not a one-time pad.

What you have described is close to a running key cypher known
as auto-key.  In an auto-key, the nth element of the key is formed
from the encryption of the (n-1)th plaintext and (n-1)th key element.
There would be a preagreed upon initial key element.

Running key cyphers do not provide much security by themselves even if you
do not compare them to a one-time pad.

See David Kahn's _The Codebreakers_ for historical information. 
See Friedman's _Riverbank Publication #16_ (published  in 1918) for
general solutions of running key cyphers.

ecp@mitre-bedford
Eric C. Pivnik