Author: Matthew White
Date: 12:46:06 10/20/03
Go up one level in this thread
On October 18, 2003 at 14:07:10, Russell Reagan wrote: >Omid recently told about his test on whether he should include checks in the >qsearch, and he said that when he included checks, his program butchered Crafty >when it got an open, tactical position, but that when it was a more closed or >positional position, his program didn't have such good chances. > >This made me wonder if it is possible to secure certain types of positions. For >instance, in the case of Falcon, it would be beneficial to secure positions that >are open and tactical when playing Crafty. So I would like to know if it is >possible for guarentee certain types of positions. Maybe via the opening book >and selecting to play gambits, or by altering the evaluation function. > >It is one thing to make the engine prefer open positions, but another thing to >have it achieve them. By playing gambits, it is always possible for the opponent >to decline, so that might not ensure a desired type of position. Also the answer >might be different depending upon whether you are playing white or black. As >black you are responding most of the time and it seems like white could keep the >game from becomming too open if he chose. I don't know a great deal about >opening theory though. In my experience, many programs have difficulty playing certain types of gambits. Since most gambits are designed to gain tempo and most programs don't have a very good understanding of tempo, gambits are often misplayed. I would only set up a program to play a gambit if I was sure that the program could take advantage of the specific imbalance created by the gambit. To answer your question, there isn't really a way to GUARANTEE a certain type of position, but it is very feasible to favor certain types of positions. If a program is a wild attacker, then playing a gambit might be fine (though if a gambit is seen as highly speculative by opening theory, it might be wise to avoid). If a program would rather grind an opponent down to an endgame, then a quieter position would be favorable. It's up to you (and your testing) to decide which your program is better at. Another type of position that might be useful to have in opening books is the type not just with an equal position, but rather good counterchances for both sides. Equal positions are often drawish, and many programs (and human players) have a tendency to get greedy in drawish positions, which leads to a loss. In a position with a great deal of play on both sides, the game is much more likely to be exciting, rather than a wait to see who screws up first. Regards, Matt
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.