Author: Uri Blass
Date: 10:56:00 12/17/00
Go up one level in this thread
On December 17, 2000 at 12:12:00, Christophe Theron wrote: >On December 17, 2000 at 11:09:25, Enrique Irazoqui wrote: > >>On December 17, 2000 at 04:30:27, Christophe Theron wrote: >> >>>Don't show this mainline and score (0.00) to Thorsten!!! >>> >>>:) >> >>By the way, did you take a look at this position >> >>[D] r1b1k3/1p1p1pp1/1p1P3p/pP6/P7/8/5PPP/R5K1 w >> >>and at how all programs, yours included, believe that black is doing great? >>Since there are 1001 cases like this, why do you think programs can put the plan >>before the search when they can evaluate so horribly? Asking programs to behave >>like people is it not like putting the cart in front of the horse, like asking a >>lion to fight like a snake? Sunday questions... :) > > > >Planning does not work 100% in current programs but it will come. > >To be fair you should also mention all the positions that would totally confuse >human players and that chess programs would play very well in the blink of an >eye. > >Programs and human players do not play with the same weapons. Programmers try to >cover the cases where programs look stupid to the human eyes. It's more a >question of cosmetics. You need to play at least 1000 games to see a position >like the one you have posted happening in a real game. > >So solving it would not improve a program significantly (from a very pragmatic >point of view). > >But it is true that solving it would be much more satisfying from an >intellectual point of view. It will certainly come... > > > > Christophe I can add that you need to know to evaluate the position correctly not only at the root. I can imagine a program that does a preprocessing and evaluate the position correctly but the preprocessing does not help it in a previous move and it is happy to go for this position with black. 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.