Author: Tony Werten
Date: 02:45:51 03/17/00
Go up one level in this thread
On March 16, 2000 at 17:30:51, blass uri wrote: >Another problem is that tactical test suites do not test tactical ability of >chess programs. > >The right move is usually a sacrifice and it is possible to improve the results >of programs in test suites by dividing the value of all the pieces by 2. > Actually, it's even worse. You can have your eval return a random score. Since the sollution move will give your apponent few moves to respond, while you have a lot of moves ( thus more chances of a good score ) you will still find the key move. ( I had a bug that did this ) Tony >The program will not be better in tactics(the opposite it may do stupid >sacrifices for positional advantage that is not enough) but the program is goign >to score better at tactical test suites. > >Uri
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.