From: nobody@alumni.cco.caltech.edu
Date: Thu, 17 Jun 93 23:50:08 PDT
To: cypherpunks@toad.com
Subject: Computer Shopper encryption
Message-ID: <9306180648.AA04035@alumni.cco.caltech.edu>
MIME-Version: 1.0
Content-Type: text/plain


If Kragen's RNG from Computer Shopper is called repeatedly with range=256,
so that the resulting values are in the range 0-255 for xor'ing,
it is very weak.  That will mean that the RNG will repeat with period
of at most 256, and so the only question is which of the 256 possible
starting points was used.  In other words you only need to do 256 trial
decryptions (just try seeds from 0-255) and you've got it.

Using the low-order bits of an LCM RNG like this one is a bad idea.  You
should use the high order bits, or use a range which is not a power of
2 so you end up using all the bits.  Even then LCM RNG's aren't crypto-
graphically strong, although from what I have seen the techniques of
breaking them are what a layman would call complicated.  Compared
to breaking, say, DES, though, they are no doubt trivial.

Hal Finney
74076.1041@compuserve.com