Newsgroups: comp.ai
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!hookup!news.uoregon.edu!news.bc.net!info.ucla.edu!csulb.edu!csus.edu!netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: Theoretical weaknesses of GPS?
Message-ID: <nagleDBsEy6.B56@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <3u9l16$2c@net.auckland.ac.nz>
Date: Sun, 16 Jul 1995 02:30:06 GMT
Lines: 19
Sender: nagle@netcom17.netcom.com

dnor01@cs.auckland.ac.nz (David Hikaru  Norman) writes:
>I'm wondering if anyone has written something on the theoretical limitations
>of Newell's GPS, as opposed to the practical limitations (hard disk, CPU speed,
>limited knowledge of an individual designer, etc). What weaknesses does the
>system have that means we should all have abandoned it by now in favour of
>GAs and neural nets?

>If you know of anything related to this, I'm interested. I'm looking for any
>well-written criticisms of symbolic manipulation approaches in general.

    Dr. George Ernst wrote a paper on the fundamental limitations of GPS
back in the late 1960s.  Don't have the reference, though.

    Like all approaches involving an evaluation function, it seems that for
many problems, too much manual intelligence is required to construct the
evaluation function before you can turn the algorithm loose to optimize
against it.

					John Nagle
