Author: Bruce Moreland
Date: 11:06:06 05/13/98
Go up one level in this thread
On May 13, 1998 at 08:00:40, Vincent Diepeveen wrote: >We're talking about results of the latest version of Diep. >Not about a set of games somewhere in between. This post contains a huge number of excuses. Fact is that if you want to compare two programs, you have to compare what is there, not what would be there if you didn't have bugs, or you'd spent an afternoon writing a book learnier, or if your program was better at certain phases of the game. It is true that you can't judge a game solely by the outcome. You might have a big advantage during a game, and your opponent might benefit from figuring out how you got it, and perhaps getting it indicates that you were doing something right. But if you throw the advantage away, your opponent might deserve some credit for keeping itself in the game, and you have some work to do in order to be able to more precisely finish the game. There has to be some point to this. If your goal is to produce something that plays some moves very accurately, and you are willing to overlook that it screws up at other times, fine. If your goal is to play attacking chess, chess that is fun to watch, and you are willing to endure some losses due to over-speculation, fine. These are both fine goals, and users want programs with these features. Remember, though, that real chess strength is assigned on the basis of wins and losses. You can't expect your opponent to resign when you have achieved your goal (a great move or a nice looking position), if you want the point, you have to be able to checkmate. bruce
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.