Author: Uri Blass
Date: 16:18:28 06/12/04
Go up one level in this thread
On June 12, 2004 at 18:01:19, Mike S. wrote: >On June 11, 2004 at 10:02:02, Uri Blass wrote: > >>>I agree and my opinion is that the way to test strategic insight of chess >>programs is games and not test positions. > >1. Here, ONE position out of 100 (!) is being discussed (and btw. it's from the >King Attack section of that test, not from the positional section). King attack can be often positional section because often you find the move for positional reasons. If there is no forced mate or forced win of material that was proved after the move then I see it as positional test. > >2. How many positions of the WM-Test did you study yet? I admit that I looked only at few positions a long time ago but my impression was that there were often cases when there was no clear proof that the move is the right move. Maybe better players than me can be convinced about it but it is not important because my opinion is that test positions with moves to find are to test tactics. If you want to test positional elements then the only logical solution except games is not to give the program positional tests with moves to find but to give the program positions and see if it evaluates correctly if one side is better and if the advantage is decive or only small advantage. 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.