Newsgroups: comp.ai,comp.ai.nat-lang,comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!newsfeed.rice.edu!news.sesqui.net!news.coral-energy.com!newsrelay.courtave.net!news.localnet.com!news3.agis.net!agis!cancer.vividnet.com!hunter.premier.net!news.mathworks.com!howland.erols.net!netcom.com!nagle
From: nagle@netcom.com (John Nagle)
Subject: Re: Unofficial Cyc FAQ Version 2.0
Message-ID: <nagleE05MD4.FvC@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <whittenDzp2BF.7sv@netcom.com> <54l608$mmq@news.unocal.com> <whittenDzqnG7.3D7@netcom.com> <554r9a$6s0@singer.cent.gla.ac.uk>
Date: Thu, 31 Oct 1996 19:05:28 GMT
Lines: 19
Sender: nagle@netcom6.netcom.com
Xref: glinda.oz.cs.cmu.edu comp.ai:41796 comp.ai.nat-lang:5561 comp.ai.philosophy:48069

gpaa29@lenzie (Francis L.Burton) writes:
>In <whittenDzqnG7.3D7@netcom.com> whitten@netcom.com (David Whitten) writes:
>>Richard Ottolini (stgprao@sugarland.unocal.COM) wrote:
>>: Dr. Lenat might improve the marketability of CYC by making a sample
>>: accessible on the web.  For example, return the analysis of an assertion
>>: a browser types in.
>>: CYC is an tantalizing system thats been inaccessible for ten years.

   Some have accessed it, and not been impressed.  See
<http://retriever.cs.umbc.edu/~narayan/proj/cyc-critic.html>
for Prof. Vaughn Pratt's evaluation of Cyc.  From that, it
looks like Cyc does about what you'd expect from pouring a 
a few hundred thousand rules about the real world into an expert
system; if you put in query for which the rules on file easily
yield an answer, you get a good answer, and otherwise you don't.
There's no indication of some kind of "deep understanding" 
developing as a result of putting in a large number of rules.

						John Nagle
