Computer Chess Club Archives




Subject: Re: The Limits of Positional Knowledge

Author: Josť Carlos

Date: 02:13:42 11/12/99

Go up one level in this thread

  Positional play is just an aproximation. In chess, the only possible results
are 1-0 1/2-1/2 and 0-1.
  If it were possible, a program would only need these 3 "positional evals" to
play chess. Just scan the whole tree, and choose the move leading to victory or
draw, if no victory possible.
  But chess is too complex for this, so we have to aproximate "how far are we
from the victory, draw, or losing". This is what positional knowledge tries, and
the only limit I find for this is the time (and space in memory) you want to
spend on the evaluation. The more particular situations you take into account,
the more accurate the eval is. The deeper you go in the tree, the less you need
this "positional aproximation", cause you are closer to the real eval (1-0, 0-1,
1/2-1/2) you are.

  Just my opinion,

  Josť C.

This page took 0.01 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.