Newsgroups: comp.ai.alife
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!news.alpha.net!uwm.edu!cs.utexas.edu!news.cs.utah.edu!dog.ee.lbl.gov!overload.lbl.gov!emf.emf.net!hilbert.dnai.com!nbn!news
From: Ron Macken <rmacken@calon.com>
Subject: Re: Evolving code?
Message-ID: <D4sttJ.EMp@nbn.com>
Sender: news@nbn.com
Organization: North Bay Network's news posting service - not responsible for content
References:  <Ben.19.00276AC0@ins.netins.net>
Date: Thu, 2 Mar 1995 05:56:06 GMT
Lines: 27

Ben@ins.netins.net (Ben Sinclair) wrote:

> why can't real code be evolved? 


I've given some thought to this problem.  One challenge is building
complex stuctures out of simple building blocks.  For example, DNA has
only four basic components but produces very complex structures.

The language needs to have a fairly small set of possible commands.
A set of commands could form a "block" with functionality similar to 
a proceedure in a traditional language.  These blocks could work with
other blocks to form even higher levels of complex actions.

The fitness criteria for evolving useful programs poses more challenges.
At the very least, a programmer will need to specify the desired
outcomes of the program.  Then each program could be evaluated on how
closely it matches the desired outcomes.

I would love to discuss this problem further with other interested
people.  The future of programming will center around evolved code.
Programs have already become too complex for humans to "design" line
by line and have the desired outcome.  We need help from the
evolutionary forces that have shaped the world.

---
Ron Macken
