Author: Russell Reagan
Date: 04:48:51 01/16/02
Go up one level in this thread
On January 16, 2002 at 07:41:28, Graham Laight wrote: >It has occurred to me that it is wrong to evaluate a position in terms of >relative pawns (the "de facto" standard - whereby an evaluation of 2 means that >you're approximately the equivalent of 2 pawns ahead). > >This means that many aspects of evaluation have to be squeezed into a dimension >which is not appropriate at all. > >A better way would be to evaluate "winning probability". If a position was a >draw, the value would be 0.50 (or 50%). If the player should win 3 out of 4 >times, the eval should be 75%. If the player must win from here, then the >evaluation should be 100%. Right there is one problem I spot right off the bat. Assuming you can compute this winning probability number (I haven't even the slightest idea how you would), and the position is such that player 1 "must win from here", then how exactly are you going to compare evaluations? If you know "this position is a win" that doesn't do much good unless you can compute the same "is this a win?" data for any given position. If you could do that, then your program would just evaluate the opening position and we would know once and for all whether chess is a win for white or if it's a draw or a win for black. The point is, it's not possible to accurately compute this kind of evaluation any more accurately than it is using the "pawn" method. You could do your evaluation in "elephants" if you wanted to, and it wouldn't make much difference. As long as the side that is approximately winning has a higher evaluation than the side that is approximately losing, the units of your evaluation do not matter. Pawns, percent, queens, dollars, monkeys...all units will work equally well if you can compute an evaluation in terms of them from a chess position. >It seems strange when you think about it that all programmers have chosen to >adopt the traditional "pawn equivalence" standard. > >-g
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.