Author: Pete R.
Date: 17:47:21 07/15/00
Go up one level in this thread
On July 15, 2000 at 19:11:44, pete wrote: >>This may be so, but how much greater is a GM's knowledge really? Being >>realistic and not worshipping the talented, that is. ;) > >It is immense I think . > >If you look at GM judgements of positions you often see even material , no >forced things and the GM says : Black wins ! No prog can do this yet ( and be >right most of the time :-) ) . I just remember a very famous position from >Alekhine from Nottingham36 and the comments in tourney book , if interested I >can search for it . > >I had an interesting experience once ; being about a 2100 player then ( far away >now ;-) ) I played against an IM in team game , chose my favourite opening , >reached my favourite type of position and was ( very ) happy . > >Some 40 moves later I lost . > >Interesting was post-mortem analysis ; I searched for my mistakes move per move >and was critical with nearly every move I made . Opponent listened a little then >explained to me :" Whole setup was wrong ! Position is passive allowing opponent >to play with a clear plan without adequate counterplay !" This isn't rocket science though, it is still a simple judgement based on factors that are difficult to code. It's like seeing which side of the board you should play on, or which your opponent intends to. Very hard to code I imagine, but nothing earth-shaking for a human. Again a strong player like that has a lifetime's collection of simple rules of thumb, but the key is that his intelligence allows him to decide which are relevant to the current position. That kind of malleable evaluation would be very difficult to code. I.e. you can't just weight all the evaluation factors and come up with something meaningful in all positions. Everything is conditional, and the factors have to be weighted differently depending on the nature of the position. *That* is what a human can do that would be incredibly complex to program. If I had the foggiest idea of how to code a program I think my general approach would be to consider pawn structure first, and make that the condition that changes how I weight the evaluations. Piece of cake. ;)
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.