From: "Reply to: hahn@lds.loral.com" <HAHN@lds.loral.com>
Date: Fri, 20 Aug 93 13:51:57 PDT
To: cypherpunks@toad.com
Subject: Re: genetic algorithms for crypto analysis
Message-ID: <930820165108.47c@lds.loral.com>
MIME-Version: 1.0
Content-Type: text/plain


  
  It has been noted in this thread that a good crypto algorithm would
  require an attacker to locate a single spike in a problem space, rather
  than having to climb a hill (which is, of course, much easier).
  
  I recall reading (I think in Sci. Am.) that a theory under investigation
  now as to why nature has sexual reproduction as part of its repertoire
  is that this gives a solution-seeking population a better opportunity to
  located spikey solutions.
  
  From the point of view of genetic algorithms, sexual reproduction means
  that each offspring must be generated from two members of the existing
  population, each of which contributes half the information needed to
  generate the offspring.  In theory, this maintains a population that
  is spread over a wider terrain, and is thus more likely to find the
  spike.
  
  I don't know if such a strategy would help at all in crypto analysis,
  or whether any genetic algorithm programs currently in use employ this
  strategy.
  __
  |         (V)              |  "Tiger gotta hunt.  Bird gotta fly.
  |   (^    (`>              |   Man gotta sit and wonder why, why, why.
  |  ((\\__/ )               |   Tiger gotta sleep.  Bird gotta land.
  |  (\\<   )   der Nethahn  |   Man gotta tell himself he understand."
  |    \<  )                 |  
  |     ( /                  |                Kurt Vonnegut Jr.
  |      |                   |  
  |      ^                   |