Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornellcs!newsstand.cit.cornell.edu!portc01.blue.aol.com!news-res.gsl.net!news.gsl.net!news.mathworks.com!newsfeed.internetmci.com!news.ac.net!news.bconnex.net!news.lightlink.com!mv!usenet
From: "Scott Russell" <queuetoo@emrys.mv.com>
Subject: Re: Schema Theory and others
Message-ID: <01bb852f$c0d0ff20$420116cf@Emrys>
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Organization: QC
Date: Thu, 8 Aug 1996 13:42:50 GMT
References: <slrn4vt3he.pgp.zxmgv07@hp12.zdv.uni-tuebingen.de><Pine.OSF.3.91.960801114840.14670B-100000@leofric><stevem-0308960554520001@spk0a-0.iea.com><4u1d33$706@soenews.ucsd.edu> <u4uivavy5e6.fsf@circa.comp.vuw.ac.nz>
X-Newsreader: Microsoft Internet News 4.70.1141
X-Nntp-Posting-Host: emrys.mv.com
Content-Transfer-Encoding: 7bit
Lines: 23

[snip]
> My feeling is that crossover, as implemented in all GA/EC -- whatever you
want to
> call them -- is no more than a glorified mutation operator, and probably
more
> disruptive than the usual mutation operator at that.

I agree.  In my own (very limited) tests, crossover is simply extreme,
uncontrolled mutation.  Pure, controlled mutation and creep have roven to
give much more reliable and dependable results.  It will remain this way
until the organisms used in the crossover are complex (ie, BIG) enough, and
until we are able to develop a communications protocol as good as what
really happens in biological crossover.  But I suppose you already sum this
up in your next paragraph.

> Crossover as implemented in biological systems is incredibly
sophisticated by
> comparison. Indeed the adaptive capacity of natural systems makes the
whole GA
> research program to date look rather like apes banging rocks together.
(no >disrespect intended.)

- Scott
