Author: Robert Hyatt
Date: 20:22:33 01/16/02
Go up one level in this thread
On January 16, 2002 at 22:38:01, Ricardo Gibert wrote: > >Special graph? You said, "It could be linear" *and* you said it would be a >"straight line", so resorting to using soemething like a graph with a >logarithmic (or whatever) scale on one side does help. You would still need to >redefine the terms "linear" or "straight line" to make what you said to work. As >for whether it is 2 or 1 pawns that is equivalent to a 75% WE, I'll let you take >your pick. It makes no difference. I tend to agree. I never said it _was_ a linear function. I said it _could be_ a linear function, perhaps with a slope of something other than 45 degrees. Didn't say anything else. Didn't intend to say anything else. Clearly beyond a certain point it can't be linear as if I am 2 queens ahead my winning chances are no worse than when I am 3 queens ahead... But in the "reality segment" of this, it might or might not be linear, I don't know. IE the probability of winning when 2 pawns ahead might well be some constant times the probability of winning when 1 pawn ahead... and so forth. Haven't given it much thought. Don't really intend to since the material + positional scores are more straightforward to think about... Although obviously my program _does_ have some "probability of winning" stuff in the evaluation so that it knows that KRB vs KR is not winning, for example... > >But it would >>be a pointless exercise, because if that can be done, then it would say that >>one could be derived from the other, and therefore the original score is >>"good enough"...
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.