Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Human-computer news

Author: blass uri

Date: 10:52:44 05/10/00

Go up one level in this thread


On May 10, 2000 at 12:42:46, KarinsDad wrote:

>On May 10, 2000 at 12:34:51, blass uri wrote:
>
>>On May 10, 2000 at 12:14:04, Enrique Irazoqui wrote:
>>
>>>On May 10, 2000 at 12:09:16, blass uri wrote:
>>>
>>>>On May 10, 2000 at 11:30:53, Enrique Irazoqui wrote:
>>>>
>>>>>Next July, Junior will play at Dortmund, one of the strongest tournaments of the
>>>>>year. Kramnik, Anand and other first rate GMs will be there too. More
>>>>>information at http://www.chessbase.com/News/newsframe.htm
>>>>>
>>>>>Together with Ed's games and the Israeli league, we can finally get a more
>>>>>precise idea of the strength of chess programs. I miss another kind of event
>>>>>where programs would have to face weaker opposition, say an average of 2200
>>>>>players. Playing only high rated players can give an inflated estimate.
>>>>>
>>>>>Enrique
>>>>
>>>>Playing everyone gives realistic estimate.
>>>>Playing more against 2200 players who know to play against computers(like
>>>>hasidovski who drew against 3 computers) gives only a lower bound.
>>>
>>>In the same way that playing only very high rated players will give you a higher
>>>bound.
>>>
>>>>I do not see a reason to assume that playing only high rated players will give
>>>>an inflated estimate.
>>>
>>>The Elo of the elite players is supposed to be quite inflated, maybe by 50 to
>>>100 points, due to the fact that they play only among themselves. When was the
>>>last time that you saw Kasparov playing in a category XIII tournament?
>>>
>>>Enrique
>>
>>I do not see a proof that kasparov's rating is higher because he does not play
>>against 2500-2600 players.
>>
>>My rating is close to 2000 and I do not think that I lost rating from my games
>>playing agaisnt 1700-1800 players.
>>
>>Uri
>
>The point is that if you only play amongst the elite players, you NEVER have the
>chance to lose to a player 400 points below you and lose a LOT of points. The
>worse that will happen to you is that you will lose a few points to another
>inflated player.
>
>Just go to FICS or ICC and see how ratings are inflated due to filtering who
>your opponents are.

ICC is not a good example because people can choose time control and choose
against who they play not only based on rating and the rules also say that if
you play against a player that is clearly weaker than you then you cannot win
small rating even if you win.

Uri
>
>KarinsDad :)



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.