Author: Dann Corbit
Date: 18:41:29 04/27/04
Go up one level in this thread
On April 27, 2004 at 21:38:18, Andrew Wagner wrote: >On April 27, 2004 at 18:06:17, Dann Corbit wrote: > >[snip] > >> >>Test suites do not work for this purpose. They are good for judging tactical >>strength, but poor estimators of game strength. If you optimze for tactics, >>then the program will play poorly. I can generate a large boost in the tactical >>strength of Beowulf by tuning using test suites. Then it gets murdered in >>actual games. >> >>It may be that quiet moves could be a good indicator. >> > >I don't understand this philosophy. The only real way to make your program >better at tactics is to improve its depth. There is no certain connection. Make your evaluation just count wood, and you will reach double your current depth. And yet it will guess wrong all the time, tactically. >How does this actually hurt its >positional play? I assume that's what you're getting at here. But it's still >evaluating positions the same way, it's just evaluating them deeper from the >root. In my opinion, at least for my program, improvement of the search and >quiesce and related algorithms have been where I've gained the most improvement >in play. This could, of course, be because I still have a pretty crappy program >that's a slow searcher anyway, but that's been my experience. Andrew Search will add a lot of strength, as will evaluation. The hard part is finding the happy medium.
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.