From: cme@ellisun.sw.stratus.com (Carl Ellison)
Date: Sat, 21 Aug 93 07:02:24 PDT
To: cypherpunks@toad.com
Subject: Re: genetic algorithms for crypto analysis
Message-ID: <9308211614.AA07811@ellisun.sw.stratus.com>
MIME-Version: 1.0
Content-Type: text/plain



The problem I see for finding spikes is that this is still a brute force
approach.  The advantage that biological populations have is in growing
new test processors exponentially, so that there is a chance to
attack something which is merely exponential.

An algorithm or breaking method which helps against modern cryptosystems
would have to turn the spikes into gentle hills by some sort of mapping of
the problem space.  Biological attack machines would still run up against
limits (e.g., the amoung of carbon in the solar system), so we have to spread
the spikes.  In fact, we have to spread them over the whole space since
enlargement of the key space can take a gentle but limited-width hill
and make it look/act like a spike.