Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cuae2!ihnp4!inuxc!pur-ee!uiucdcs!uiucdcsp!forbus
From: forbus@uiucdcsp.UUCP
Newsgroups: comp.lang.lisp
Subject: Re: LISP for IBM compatible PC's
Message-ID: <80300003@uiucdcsp>
Date: Sun, 1-Feb-87 14:58:00 EST
Article-I.D.: uiucdcsp.80300003
Posted: Sun Feb  1 14:58:00 1987
Date-Received: Tue, 3-Feb-87 06:46:38 EST
References: <1048@sfsup.UUCP>
Lines: 40
Nf-ID: #R:sfsup.UUCP:1048:uiucdcsp:80300003:000:2111
Nf-From: uiucdcsp.cs.uiuc.edu!forbus    Feb  1 13:58:00 1987



>As I recall, the original query was specifically directed to PC
>environments.

Yes, and I pointed out that for PC's you would almost certainly be better
off with TI-SCHEME than GCLisp.  You brought up the AT environments.

>Further, I didn't say that our Symbolics users were ready
>to switch machines; I intended to indicate that they were impressed by
>the performance and the quality of the environment IN RELATION TO SYSTEM
>COST, which has been known to be a factor for some of us...

I suggest you might write more plainly then.  The unadorned statements "a
thing of beauty" and "makes their eyes bug out" would tend to imply that.

>Let's read replies in the context of the original message and be a little
>less quick with the derisive flames, hmm?
>						Peter C.
>/* End of text from uiucdcsp:comp.lang.lisp */

There is a larger context that one must also be aware of:  Many people are
being irresponsible in giving the impression that current micros can perform
at the level of real lisp machines (I'm not accusing Peter of this, his
response clearly shows him not of that ilk).  Right now they just can't.
(I'm certainly rooting for them, I'd very much like to get my research
machines fixed by throwing them into the back seat of my car and driving
them to the local ComputerLand!)  If people get the impression that that
kind of performance is already here, then if they fail to get the results
they want using a micro then they might conclude "AI is no good" as opposed
to "Okay, the computing power isn't here yet/isn't cheap enough yet".

I am a strong fan of attempts to get affordable, micro-based AI environments
distributed.  The combination of GCLisp-LM and AT's has enabled our AI
courses to provide much better "hands-on" experience than ever before.  I
would not have dreamed of developing my "Building Problem Solvers" course
without them.  HOWEVER:  They are also currently not powerful enough for
most serious AI research, and one should be realistic about that.  Given the
degree of AI hype floating around these days, I think one must be extra
careful not to mislead people.