Message-ID: <3218F724.5927@baste.magibox.net>
Date: Mon, 19 Aug 1996 16:22:12 -0700
From: Brig Klyce <bklyce@baste.magibox.net>
Organization: Acorn Enterprises, Inc.
X-Mailer: Mozilla 2.01 (Win16; I)
MIME-Version: 1.0
Newsgroups: comp.ai.genetic
Subject: Q: programs that innovate?
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: baste.magibox.net
Lines: 16
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!godot.cc.duq.edu!newsgate.duke.edu!news.mathworks.com!enews.sgi.com!news.sgi.com!news-out.microserve.net!news-in.microserve.net!imci2!pull-feed.internetmci.com!chaos.magibox.net!

In biological evolution, the genotype acquires new genes, and the 
phenotype acquires new features as life goes from bacteria to eukaryotes 
and so on up to birds, reptiles, mammals, etc. Each new step builds on 
the previous stage. According to the modern version of Darwinism, this 
process has no input except random mutations in the genotype and natural 
selection of the phenotype. 
	I am aware that random mutation can make a variety of shapes in 
Dawkins's "Biomorph" program. I know that natural selection in TIERRA can 
yield mutants with a reduced instruction set. And I know that boolean 
networks can exhibit complex patterns. But I am not aware of any computer 
program that can, by any means, with limitless iterations, create the 
analog of new genes. That is, that can produce wholly new, longish 
strings of code with unforseeable new, useful meaning.
	If someone has an example, please describe. Thanks.
Brig Klyce  (bklyce@baste.magibox.net)

PS	"Please e-mail, I'll summarize."
