Author: Reynolds Takata
Date: 19:48:59 01/19/99
Go up one level in this thread
On January 19, 1999 at 22:14:08, Dann Corbit wrote: >On January 19, 1999 at 21:40:06, Reynolds Takata wrote: > >> Is it possible to take a large database of a specific GM's games, then >>manipulate a programs settings so that in all of the games in the database, the >>comp finds the moves of the specific GM's to be absolutely best. Next set the >>programs settings to the average of the piece values and other settings etc., >>that one used to make the comp find all of the GM's moves to be strongest. >>This all for the purpose of truly making a program play ina specific GM's style. >> I'm not sure if conceptually this works out, and definitely see some problems >>with it. I'm just searching for input to make this conceptualization work out a >>little better(if possible at all). Thanks >The Chessmaster series tries to do this by some means. While you can change a >program to make it more attacking or favor a particular opening, I don't think >we can *really* make it play like a chess superstar. I think that the concept >is a fun idea, but imagine trying to make a program play like *you*. I think >when you think of it that way the difficulty becomes more obvious. Of course, as i mentioned even if it is possible i think it would quite difficult. In the manner i proposed above, of a sort of "average evaluation" scheme, the average doesn't neccesarilu mean that a computer will play like the proscribed GM. However, it might be possible to bring about play with certain tendencies perhaps, and in that vain said to play in a similar style. I'm not sure if CM's GM personalities are accurate or what method they use. I do know that the Tal personality will almost always sack a night for two pawns in front of the king and maintain a pin on knight with the bishop. Is this like Tal? No but it is attacking looking :).
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.