Author: Amir Ban
Date: 03:34:00 12/29/98
Go up one level in this thread
On December 28, 1998 at 19:33:00, Detlef Pordzik wrote:
>On December 26, 1998 at 18:38:07, Amir Ban wrote:
>
>>On December 26, 1998 at 17:12:24, James T. Walker wrote:
>>
>>>Can someone explain why Junior 5 played h7?? in this position? Even if allowed
>>>to go to 62 ply it plays h7?
>>>
>>>[Event "Blitz 5'"]
>>>[Site "?"]
>>>[Date "1998.12.26"]
>>>[Round "?"]
>>>[White "Junior 5, PII."]
>>>[Black "Nimzo99"]
>>>[Result "0-1"]
>>>[WhiteElo "2555"]
>>>[BlackElo "2500"]
>>>[Annotator "Junior 5"]
>>>[SetUp "1"]
>>>[FEN "6nK/5k2/7P/8/8/8/1B6/8 w - - 0 1"]
>>>[PlyCount "4"]
>>>
>>>{20928kB, Jun-book.ctg=351287 pos}
>>>1. h7 {0.00/62 3} 1... Ne7 {15}
>>>2. Be5 {-85.66/62 0} 2... Ng6# {6}
>>>0-1
>>>Jim Walker
>>
>>This is the result of assuming insufficient material for mate.
>>
>>Go to 'engine parameters' and turn off 'insufficient material' if you don't want
>>to fall for these kind of positions.
>>
>>Amir
>
>What about the people, who want to LEARN from your Prog, Amir ?
>
>My PERSONAL opinion on this behalf is, that I sure do expect a top ranked
>program to KNOW - without button pressings, what a wrong bishop is and where
>insuff. material is correct - or not + some more of this.
>
There are a few positions where the insufficient material assumption fails. I
don't know of any way to know this except to search and find that it's true or
not, but this defeats the purpose of the heuristic, which is to make the engine
stronger by ADDING the knowldege (which granted is not perfect) that a class of
positions are not winnable.
This assumption is not much different from the assumption that a rook is better
than a bishop, or the null-move assumption. These are true in general, but fail
in some positions, and the fact that they are not always true is not a reason
not to make them. A program that doesn't assume that a rook is better than a
bishop will play very smartly in those positions where a bishop is better, but
in general it will be weak and stupid compared to one that correctly makes this
assumption.
>If this path will be followed continuously - by whomever, the next this, that
>will be spared out - are passed pawns - or something, because not so often
>seen during the gameplay.
>And so on ( again ) -
>the end will be them ultrafast searchers - with emphasingly 500.000 positions a
>minute or second or whatever - and with the basical knowledge implement
>of a Domino play engine.
>
I think the opposite applies here. This is a case of applying a simple piece of
knowledge about the game, which fails in this position.
>See : that's just MY opinion on this......I love the game. I'd love
>to increase my rather poor quality - but this can only be achieved - by
>knowledge, very fundamentally - correct ?
>
>DPordzik
>a.k.a. ELVIS
Amir
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.