Author: Rolf Tueschen
Date: 17:05:30 12/05/02
Go up one level in this thread
On December 05, 2002 at 17:32:13, Bob Durrett wrote: >Rolf: > >Before offering you a more complete expression of any [hopefully useful] "chess >programming related" thoughts I might have on your topic, we need to clear up a >point of confusion. I'm not clear what you mean when you say "Positional >Position." [Also, what is it, exactly, that is "impossible"?] Example, a position without tactics, meaning without ordinary eval advantages. Either no attack, no sacrifice, no pawn flow etc. For example you have the chance to exchange N for B or vice versa. But not wih a "later" pawn win or such. Now my point is that if you tak the position it might be that a program makes the exchange. But - has it understood that here N or B is "better". Now my point is that you go back a few moves and the prog should show if it knows that the later exchange should be worthwhile. Because then we know that this was intended. > >What do you mean when you say "a position without tactics"? I am no GM, but I >have never been able to completely separate the two to my personal satisfaction. Of course you are right. You can define it this way or that way. > Exploitation of "positional" weaknesses seems always to involve tactics at some >point. But "my" positions require that you have no instant visible advantage in the evaluation. >Perhaps a GM could come up with a counter example? Similarly, tactics >often is associated with a positional weakness, which is to be exploited [or >defended]. > >Going a bit further on this point, please let me say that all chess positions >come about from the previous moves of the game. Creation of a positional >weakness might have been accomplished by tactics or maneuvering. Sometimes they >originate by blunders or mistakes. In that sense, no positional feature is >independent from what came before in the game. But, I doubt that is what you >are saying. : ( I fear yo must prepare yourself that you must help me a bit with all this. I'm not an autonome thinker. :( > >Incidentally, there are more than two "Bob's" here. Several "good" ones, in >fact. : ) We need all of them. Rolf Tueschen > >Bob D. > >_ _ _ _ _ _ _ _ _ _ _ _ _ _ > > >On December 05, 2002 at 11:43:36, Rolf Tueschen wrote: > ><snip> > >>I see what you mean. But I think you won't deny that in a game of chess there >>are no isolated positions. > >A chess game can be thought of as a sequence of chess positions related by the >condition that each, except for the initial position of the game, is obtained >from the previous by a legal move. > >But I don't think that's what you meant. Care to offer clarification regarding >the interrelationships you refer to? > ><snip> > >>I was talking about "positional" positions. Positions that are allegedly >testable in the mentioned WM (=Wch) test by Gurevich. And the header of my >message says >>it exactly what I meant: positional test positions are physically impossible. >>Because, and that is important, the development is so important. > >Can you come with an example, which makes the correctness of your observation >more evident? Please try. I'm not sure what all you would include in "this >development." Incidentally, I am a chess amateur not yet at the GM level, so >please try to make your explanation understandable at the chess amateur level. > >>I don't mean >>development in the INFORMATOR language! I mean the development of the chess >game >>up to that particular positional position. That is a position without tactics, >>so without a Kings attack, but also without clear goals at all! At least for >>amateurs and chess programs. > ><snip> > >>our program could have never attempted to get to that position >>earlier on because it simply cannot ""see" the advantages of such a position. > >This may be a bit too much thinking of chess playing programs as being human, or >having human attributes. They really are NOT human. [Alas! They are more like >"aliens from outer space" : ) ] > >My impression is that part of the code in any modern chess engine is devoted to >evaluation of chess positions. Crafty, for example, has about 4000 lines of >source code just for that purpose. The programmer is at liberty to make that >code evaluate *anything* he/she desires. [The execution of this code may >include sequential processing similar to that in search algorithms, or it may be >completely different. After all, PCs are sequential, unlike humans.] > >Why cannot the 4000 lines of "position evaluation" code in Crafty do what you >want? Maybe "the other Bob" can give the answer? > >I prefer not to think of the programming task you discuss as being "impossible." > At worst, it should be "difficult." That, anyway, is my preference. I hate >the word "impossible." That's one of my personal quirks. > > >>So I was concluding, that if you want to test machines for such positional play >>you must take earlier position. But because of the actual blindness that can't >>work either. > >Incidentally, I'm curious: WHY is it so impossible to examine ANY chess >position and determine what move(s) are best? Or, did you not mean that? Is >there something inherent is chess programs which make them incapable of doing >that? If [?] humans can do it, why not computer programs? Or is it impossible >for humans too? > ><snip> > >>Rolf Tueschen > ><snip>
This page took 0.01 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.