Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: LCT-II Fin8

Author: Vincent Diepeveen

Date: 19:20:08 08/29/99

Go up one level in this thread



On August 29, 1999 at 21:55:49, Howard Exner wrote:

>On August 29, 1999 at 21:15:56, Vincent Diepeveen wrote:
>
>>
>>I wonder why this position is in a testset for computers,
>>as when i analyzed the win here some years ago
>> i needed some 40 moves before
>>any evaluation of a computer could see it as a win.
>
>It would be great if you happened to save your analysis. Did you?
>A few of us have been trying to work this difficult endgame out
>assisted by computer programs (in my case at least).
>>
>>that's 80 ply, as white king has to move a lot from one side
>>to another and such
>
>Just a guess but maybe it was testing whether programs had some kind
>of code for toatally blocked positions. As humans we know that without
>the pawn break, c5 the score is 0.00. Most programs most likely see
>the position before the break as approxiamately +1.XX. Would programs
>therefore chose a move like c5? Maybe in the future. Even now some
>cunning programmer might have code that detects such blocked
>positions as a draw?

have code like that in diep but that never makes up for sacking a pawn here
such meta code can never change scoring that much as code is not real
knowledge that is always true.

in dutch open diep lost vs patzer cuz some meta code 'gave it the insight' to
keep pos closed....   ....happens. after that i got rid of that.
influence of a heuristic may be not that big.
this will remain a big advantage for mankind: meta meta code or in short
having an idea... ...and recognizing its importance. human can sack pawn here
without seeing line

gotta dig for line





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.