Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Programs vs. the "Web", any statistics available?

Author: Brian Kostick

Date: 13:55:02 03/30/01

Go up one level in this thread


On March 30, 2001 at 16:07:31, Dann Corbit wrote:

>On March 30, 2001 at 15:20:05, James Robertson wrote:
>
>>I've noticed that a lot of programs are now battling the web. Does anybody know
>>how many people are actually voting though? Gambit Tiger, for instance, only has
>>the percentages, and not the actual numbers of voters.
>>
>>Out of curiosity, are these programs actually playing the "Web" or are they in
>>reality just playing a few dozen chess freaks from this site? Does anybody know
>>how many voters (even a very rough average will be interesting) there are for
>>the games Deep Fritz, Deep Shredder, and Gambit Tiger vs. the web?
>>
>>I'm hoping the numbers are not too low. The fact that Deep Shredder's game was
>>advertized on kasparovchess.com and TWIC is really good. Maybe this will attract
>>substantially more people.
>
>"X verses the Web" is an old gimmick.  Actually, most of the time, "the web"
>plays much more poorly than a single good opponent.  Imagine, a big committee
>voting on which move to make.  Are most of them spending 24 hours of computer
>time analyzing a move or GM's themselves?  Surely not.  In general, it is a
>farce (to my way of thinking) but a nice way to gather publicity.  I think the
>Kasparov match was different, however.  For some reason, a very large group
>galvanized resources very well, and organized fairly efficiently.  The
>commentary by the experts was quite good (and gave me new respect for I. Kush).
>At any rate, that sort of high quality match [despite the attempted sabotage] is
>_by far_ the exception rather than the rule.
>
>If you want to see very high quality chess, you will get far, far better by a
>match between two highly rated opponents than one high quality opponent verses
>the web.  I think I could probably beat most people on the internet (which isn't
>saying much) but I am absolutely sure that a really good player like Vincent,
>Djordie, or Come would slaughter me repeatedly.  So you get ten thousand morons
>to collectively make their choice -- how good is that choice going to be?  Even
>if experts give excellent analysis -- will they even understand it?
>
>I suspect [snicker] that the higher the number of votes received, the LOWER the
>quality of the move generated.  If a move got a million votes, it is almost SURE
>to be a real dog.
>
>IMO - YMMV.

Dann,

   You write as if it's really ProgramX vs. the Web (where the Web describes
thinking, chess playing individuals?). In reality I think the winning vote ends
up being ProgramX vs. ManyPrograms, with 'ManyProgams' thinking maybe 16
min./move average but that's really just a guess on my part. The major vote
discrepancies I've observed seem due to opening book lines. Regards, B.K.



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.