Author: Mridul Muralidharan
Date: 13:21:20 04/22/05
Go up one level in this thread
Just to humour your argument , please take a loot at any of these position's. [D]1nb1kbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQk - 0 1 [D]1nbqkbn1/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1 [D]1nb1kbn1/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1 [D]2b1kb2/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQ - 0 1 I seriously doubt if any program will show a mate for these (the last position - maybe it might find a mate). Does that mean that the eval of 22+ pawns for these positions is wrong ? Definitely not ! The program can see itself winning and making progress. We are not talking about bean counter programs here which have very poor positional and material knowledge - I am sure others might have tried to explain this to you :) Just consider this - when will you say a position is better than other ? Only when you can see the mate ? When there is a certainity of victory even though you can possibly get to the mate in realtime (not just humans , but even computers) ? So if you are a queen down in an endgame , will you continue to play till the bitter and miserable end ? (assuming the opponent is as competent as you) - if you do this in a tourny , people will term you as obnoxious :) But trust me , I have seen people do that even in dead lost endgames with lone kind against 3 passed pawns with no chance to stop the pawns :) One of the reasons why programmers usually let their programs play to the bitter end is that no one is actually wasting any mental energy thinking , also that it will be instructive to look at the variations taken so that the programmer can better analyze the game latter (like going from a mate in 28 to mate in 19 - should improve engine to prevent this - that sort of thing). I am not sure if I conveyed what I wanted to. And my program also says 3.6 pawns adv - which slowly keeps rising after each iteration. Regards Mridul
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.