Author: Bruce Moreland
Date: 00:01:04 09/10/98
Go up one level in this thread
On September 09, 1998 at 17:54:06, Serge Desmarais wrote: > Just that I am not sure about how good/valid blitz games testing can help to >improve things? And computers/programs are already better at playing blitz games > than about any human, not to speak about bullet games. Personnally, I would >like the programmers/companies to organize more matches against good GMs >involving only long games (40 moves in 2 hours and then 20 moves an hour). These are very good issues. On ICC you have to take what you can get. This has always been fine with me, I never much concerned myself with whether the games were against humans or computer or were slow or fast. If what is happening is that I am optimizing for blitz peformance against humans, I am probably also optimizing for enterprising play, and this is fine with me even if the concern is that my program will take too many risks at longer time controls. There are some problems that you have to solve when you play blitz on ICC, which will carry through to real games, I think. For instance, you have to be able to handle the Stonewall, and various attempts made to defeat your opening book. I don't know if there are any drawbacks from doing too much blitz. I don't get enough long games to know if my thing has any problems at longer time controls, but I don't have any reason to suspect that it does. 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.