Author: Ricardo Gibert
Date: 09:33:08 09/09/00
Go up one level in this thread
On September 09, 2000 at 08:44:57, Antonio Dieguez wrote: >On September 09, 2000 at 02:13:19, Ricardo Gibert wrote: > >>On September 08, 2000 at 19:05:20, Antonio Dieguez wrote: >> >>>just for fun, can you do a search with your program in the initial position? >>> >>>[D]rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1 >>> >>>here is for amyan, unrealistic evals? 1/3 of a pawn just for one piece more >>>developed hum... may be better something a bit exponential? >>> >>>[2] -0.01 87 e2e4 e7e5 >>>[3] +0.35 426 e2e4 g8f6 b1c3 >>>[4] -0.01 2021 d2d4 d7d5 g1f3 g8f6 >>>[5] +0.32 7746 e2e4 e7e5 g1f3 g8f6 b1c3 >>>[6] +0.12 24010 e2e4 e7e5 g1f3 g8f6 f1c4 b8c6 >>>[7] +0.30 63896 e2e4 e7e5 g1f3 g8f6 b1c3 b8c6 d2d4 e5d4 f3d4 >>>[8] +0.16 282830 e2e4 e7e5 g1f3 g8f6 f1c4 f6e4 d2d3 e4d6 f3e5 >>>[9] +0.27 1086340 e2e4 e7e6 b1c3 g8f6 g1e2 f8b4 e4e5 f6d5 d2d4 >>> >>>just me. > >hello > >>Theoretically it should be about 1/6th of a pawn. If White is 1/3rd of a pawn >>better than "even", then Black is 1/3rd of a pawn worse than "even" or in other >>words, White is 2/3rds of a pawn better than Black, which can't be right. > >no, +1/3 for white means white is 1/3 of a pawn better than black. Are you sure? Lets say I am White. We are in the beginning position. I have the advantage, since it is my turn to move. Now if I drop a tempo, then it is now you who are effectively playing White and now you have the advantage. This different from the situation of my being a pawn up. Now If I drop a clear pawn, you do not assume the advantage of being a pawn up. The material is now equal. I would have to drop 2 pawns for the roles to be reversed. Do you see how it is different? Now consider what happens when White wins a clear pawn. His advantage of the move does not disappear, he then has the pawn PLUS the advantage of being White. If he were only exactly a pawn up, White would have to drop a ½ tempo. Do you now see why some say a pawn is worth about 2½ tempos? I think it is worth about 3, but I can see why there is the alternative interpretation. >very ugly speak this way about chess anyway :) > >>With White being 1/6th of a pawn better than "even", then Black is 1/6th of a >>pawn worse than even, so now White is now 1/3rd of a pawn better than Black, >>which is what you would expect. I remember some guy on ICC reported .17 with one >>of the commercial programs, which is consistent with 1/6th of a pawn. > >.17 doens't seem rare, i will implement an exponential function for development >and also for king danger patterns. May be the average between x and x^2, or >something like that. > >>With the eval oscillating with odd or even depth, you should be getting a ½ >>tempo better on odd ply and zero on even ply. You're showing instead about 1 >>tempo better on even ply and about a ½ tempo better on odd ply. > >hey, in the initial position white plays and wins! >ok I get it but is not so simple. > >>All this assumes a pawn is worth 3 tempos, but some would say it is worth only >>2½ pawns. Take your pick. > >mmmh...
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.