Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Can anyone here beat the "best" prog in 40/120

Author: Uri Blass

Date: 00:53:11 11/09/01

Go up one level in this thread


On November 08, 2001 at 20:47:54, Vincent Diepeveen wrote:

>On November 08, 2001 at 08:58:30, Jonas Cohonas wrote:
>
>I already answerred that question by beating Tiger on a 1.4Ghz machine
>some time ago. If you would know tiger well you know that no matter
>what time you give it, if it decides for a move it will keep playing
>that move.
>
>I beated it by 'accident' kind of 2 times because i prevented it from
>castling. by doing that i lost a pawn, but then it didn't do anything
>to escape from that castling prevention (i'm sure shredder would have
>done it and some other programs too).
>
>so it lost.
>
>I do not understand why you can't make the induction step that if
>it can be beaten at 15 15 (which i use to practice, not to beat compu's),
>then why wouldn't it be possible at 40 in 2?
>
>somehow you're the only computer guy who can't make that induction step.
>
>Human obviously is better when time gets better for him.

It is possible that you can win against a computer by using the same tactcial
trick at blitz but you cannot use the same tactical trick at long time control.

Adams could not win against Deep Junior in dortmund because he went into an
endgame when he knew that he could beat Junior at blitz but Junior found better
move at long time control and drew the game.

I cannot do generalizations and say that humans are better at long time control.
There are humans who are better at long time control but there are humans who
can do better results in blitz because they go for tactics that the computer
opponent falls for them only in blitz.

The trojan horse against a program that is good on tactics(not shredder5) can be
a good example for it.

There are still cases when the sacrifice is too deep for programs to see but if
the programmer use special extension and does not try to solve it by evaluation
then I guess that even in these examples the program is not going to fall for
the trojan horse at tournament time control when it may fall for it at blitz.

Uri



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.