Author: Vincent Diepeveen
Date: 07:55:43 01/08/01
Go up one level in this thread
On January 08, 2001 at 10:42:36, Uri Blass wrote: >On January 08, 2001 at 10:34:38, Uri Blass wrote: > >>On January 08, 2001 at 10:09:34, Vincent Diepeveen wrote: >> >>>On January 08, 2001 at 10:02:44, Uri Blass wrote: >>> >>>>On January 08, 2001 at 08:12:25, Vincent Diepeveen wrote: >>>> >>>>>On January 07, 2001 at 20:57:46, Ricardo Gibert wrote: >>>>> >>>>>>One well known rule of thumb is: It is often a bad idea to grab more material >>>>>>when already ahead in material. >>>>>> >>>>>>For instance, when I play, I generally avoid increasing my material advantage >>>>>>unless it does not cost me anything (time, structure, open lines for opponent, >>>>>>etc.) to do so or I have nothing else to do that is constructive. >>>>>> >>>>>>In fact, a reasonable rule of thumb for open positions might be: The 1st pawn >>>>>>grabbed is worth 3 tempi, but the 2nd pawn is only worth 1 tempo. Something I >>>>>>observed while I studied some of Morphy's games. >>>>>> >>>>>>My question is: Do any programs take this into account or do they all consider >>>>>>the 2nd pawn they grab to be as valuable as the 1st? >>>>> >>>>>why do you assume all programs are preprocessors? >>>> >>>>I did not read this assumption. >>>> >>>>Uri >>> >>>He says: "if you make a move capturing a second pawn" >>> >>>So his way of seeing a positions evaluation is move based. >> >>I think that he meant that if you increase your material advantage from 1 pawn >>to 2 pawns and it is not move based. >> >>Uri > >I mean that he meant to say that the positional advantage should not be added >linearly. > >It means that if you see that white has 1/2 pawns positional advantage for >reason A and 1/2 pawns positional advantage for reason B the total advantage is >more than 1 pawn. > >Uri This is of course your interpretation and i also already interpreted it possibly as this, but obviously wasn't the question, nevertheless this interpretation is already used by many programs using 'lazy evaluation' which in a sense is the same principle. If a lot of material up they don't watch the positional score anymore and they simply prune. Now here you do something similar: if a lot of material up, do as if you are less material up, like dividing by 0.75. Now basically this is an attempt to win a game 'safely', nevertheless it has big risks. A program will simply easily give away a second pawn with only some real vague compensation. Just like many speculative things, that might work against humans but in computer-computer games it will not be very rewarding. You'll now lose with a dead sure garantuee instead that you might save your ass with just 1 pawn less. Greetings, Vincent
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.