Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Rubbish

Author: Sune Fischer

Date: 10:15:52 04/29/02

Go up one level in this thread


On April 29, 2002 at 12:31:57, stuart taylor wrote:

>>I prefer (at least) ONE game at tournament time control, g/120, 40/40 instead of
>>hundreds or thousands of bullet games.
>
>Well, in fact, I feel the same way, but I thought I was always in the minority.
>On the other hand, great programers like Chr. Theron seems to favour the
>multitude of games, even blitz (I don't know WHICH blitz), and, as far as
>computers themselves are concerned, some programs can play "instant" at a
>standard that is unimaginable to any human without quite alot of thinking time
>allowed.
>Simply, the whole concept of time is not the same for humans and computers. BUT,
>longer is (alot) better, even for computers, and as long as that remains true,
>my preference for those longer games makes Blitz a complete waste of exitement.

Neither program has solved chess, they both suffer from horizon effects which
will make all games a bit random.
In some sense it is better to play a lot of fast games to see which is better
_on average_. For Hiarch-Fritz you would might see a score like 45-55, so of
course one game would not tell you much. Getting this kind of statistic is
almost impossible with long time controls (obviously).

However for all we know, program X might do better against Y in fast games and
vice versa in slow games, so it probably doesn't show us that much anyway.

-S.



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.