Author: Otello Gnaramori
Date: 12:18:29 07/08/01
Go up one level in this thread
On July 08, 2001 at 14:49:59, Gordon Rattray wrote: >No. For many positions, it is difficult to evaluate them correctly. It is >anything but straightforward. It is common for humans/computers to play towards >a certain position, only to find that they had incorrectly evaluated it. For >example, playing for an attack or an endgame position that isn't as good as it >appeared from afar. Exactly. The example of the attack is perfect to explain my point : Only after the calculus of "all" the possible variations you can decide which is the best move to play. >>King safety , doubled pawns , open files ,center domination etc. don't require a >>special effort to be evaluated. The major effort goes into calculus of >>variations, since IMO the evaluation of the goodness of the position is easily >>understood (by the computers too). > >Easily understood? You should either play for the world championship and/or >start writing a chess program if you think that positional evaluation is >straighforward... ;-) Why do you think that programmers keep adding more and >more knowledge instead of just trying to get their programs to calculate deeper? > >Gordon > That's not exact, since the effort of the programmers IMO are also in efficient search and propagation algorithms . They aren't focused only in the improving of the eval function that is obviously of key importance. What I meant is that the main concepts behind the eval are known , there is nothing special about that : king safety and so on... The main power of programs against the human beings is to outplay the human players in deep calculations. Regards
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.