Author: martin fierz
Date: 20:17:19 08/30/02
Go up one level in this thread
On August 30, 2002 at 21:24:59, Mark Rawlings wrote: >Hi Martin- > >Slightly different subject here, but in the Las Vegas checkers tournament, your >program lost a few games. Did you figure out what the losing moves were and >then run cake for a long analysis? I was wondering if these positions were too >deep for the programs to see. hi mark, yes, of course i looked at the games. the game cake lost to kingsrow was a book mistake. the game it lost to nemesis was because of too optimistic forward pruning. it worked well in all my tests... cake played another bad move in it's final game against kingsrow. of course, i started fiddling around immediately when i came home to see if i could make it avoid those mistakes. my "post-las vegas" version would avoid both of these mistakes in crossboard play now. of course, with the same book it would make the same book error. i'm in the process of having the book fix itself - and soon it will know about that other loss too. however, i don't have to be too ashamed for my program. i got along well with the other two programmers and asked them what their programs see in these positions without books - and they don't see that these moves lose either. to their credit, they both had the line i lost with against kingsrow in book though, so they would not have made that mistake. aloha martin
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.