Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Is it better for a program to beat other programs or Human GM?

Author: Jorge Pichard

Date: 03:41:58 07/24/05

Go up one level in this thread


On July 24, 2005 at 06:15:54, George Tsavdaris wrote:



FF: We went for a long walk and I asked him how much time he spent tuning his
programs to play against other programs. He said 60-80% of his development time.
Then I asked him how much we, ChessBase, profited from this time. He said, well,
we beat most of the other programs. Finally, I asked what would happen if we
spent all that time teaching it how to play against human beings. Frans thought
for a minute and said, “I think we should be doing that.”

http://www.chesscafe.com/mig/mig.htm
_______________________________________________________________________________
This was my sugestion of how to accomplish this task:

My sugestion to prevent human AntiChess is to incorporate an algorithm, in which
the program would force to exchange at least three pawns early in the game
before it reach the middlegame, even if it lose one pawn for the cause. I notice
that in most AntiChess game, the position always ended up being blocked by at
least 5 pawns.

Jorge

Yes, but implementing this algorithm would *maybe* weaken the engine in engine
vs engine matches, in which more people are interested.

But all in all it is a good plan, but not in the engine in general. It should be
included as an option in the parameters, something like Tiger's "Anti-Human"
style. "Anti-AntiChess" lol.






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.