Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!pipex!uknet!dcs.gla.ac.uk!udcf.gla.ac.uk!gapv64
From: gapv64@cent.gla.ac.uk (Brian Ewins)
Subject: Re: GA's for corewar
Message-ID: <gapv64.781785431@cent.gla.ac.uk>
Sender: news@udcf.glasgow.ac.uk (News)
Organization: Glasgow University Computing Service
References: <36pva0$i8u@ns.mcs.kent.edu> <mreddy-041094101901@mreddy.comp.glam.ac.uk> <1994Oct6.210307.27146@inca.comlab.ox.ac.uk>
Date: Mon, 10 Oct 1994 10:37:11 GMT
Lines: 23

I'm just an innocent bystander, so take this as being a stoopid
question:

When you talk about the redcode in corewar being critically independent
on some values, is this because the code that's produced by just
writing in 'random' values is useless (as redcode) or is it something
that is fairly independent of the instruction set you use? (is this
why you use your 'paracode'?)

I only ask because I read about Tierra, and they said there that 
they developed their redcode because most strings of 'real' machine
code are useless: the Tierra redcode is supposed to 'work' (well,
maybe) for most strings. Sounded fairly similar to the problem
you described.

Tierra was supposed to be used to evolve better programs, as you are
doing... does anyone know if they have done anything cool recently?

	if this was dumb, consider my head slapped and
	I'll act suitably chastened.

			Baz.

