Author: Mike S.
Date: 10:44:02 11/22/00
Go up one level in this thread
On November 21, 2000 at 19:53:00, kurt wrote:
>(...)
>Hi,I am not a programmer,just a long time chess player. (...)
>One more thougth- I can visualize a picture of a castle-somewhere-someplace,
>and can find the means to get there and build it. Computers can not.They must
>put block on block to get somewhere. You must guide this block laying as to
>how to do it and stay within its memory capacity. To play chess you gave
>the instruction to add the sum of its programed evaluation points and try to
>increase its own total. You do it by making allowable moves and adding a block
>with an equall or more weight without knowing where to go or where it will
>end up. Now desribe how will you reach an endgame position which will be
>andvandages for your program as Capaplanca and other players did.
>
>In defence of humans lets remove any opening libary which humans
>are not allowed to use (while playing OTB)to level the playing field
>between machine and human.
I'd like to answer although I am not a programmer either. Your last idea -
removing books - has been discussed often (unfair!, etc...). But the human
opponents would keep their opening knowledge! I don't think this is what you had
in mind (?), as it obviously gives one side a big advantage. Building databases
is the method to store knowledge in a computer, and the chess program uses this
knowledge, like a human player uses his memory.
Opening Book = opening knowledge. This discussion leads us nowhere... because I
have the impression either someone accepts the "artificial" opponent completely,
and right from the start, or he will always find argument after argument why
everything is unfair and that computers can't play chess (tell this a
Grandmaster who has just lost to one). Since I watch this discussion on the
internet, I have never seen somebody really change it's opinion about this. Me
neither.
But I would like to illustrate the other thing you mentioned, about visualizing
a picture. I think, this is one of the main differences between human thinking
in chess, and the method chess programs use. I have written an article some time
ago for the CSS magazine ("Visions versus Variations"), where I tried to
illustrate this by the following moves (starting from a training position):
[D]r4r1k/p5pp/b1p5/3pNp2/3Q1P1P/q2P2P1/P5B1/1R4K1 w - - 0 27
In the original continuation (Yates-Nimzovich, Karlsbad 1923) white followed the
idea h4-h5-h6, but without success. In my small training session (which was
against Hiarcs 4/P133 at 0:15 per move - still a superior opponent for me) I had
another idea now: To put the queen at h5, pinning the pawn h7, and eventually
get the chance to play Ng6+ etc.:
[Event "Training_0:15 per move"]
[Site "?"]
[Date "1996.??.??"]
[Round "?"]
[White "Scheidl, M."]
[Black "Hiarcs 4/P133"]
[Result "1-0"]
[SetUp "1"]
[FEN "r4r1k/p5pp/b1p5/3pNp2/3Q1P1P/q2P2P1/P5B1/1R4K1 w - - 0 27"]
27. Qf2 Qc3 28. Qf3 Rab8 29. Rxb8 Rxb8 30. Qh5 Bxd3 31. Ng6+ Kg8 32. Ne7+ Kf8
33. Ng6+ Kf7 34. Ne5+ Kg8 35. Qf7+ Kh8 36. Qe6 h6 37. h5 Qc2 38. Nf7+ Kg8 39.
Nxh6+ Kh8 40. Nf7+ Kg8 41. Ng5+ Kh8 42. Qg6 Kg8 43. h6 Qb1+ 44. Kh2 Qb7 45.
Qh7+ Kf8 46. hxg7+ 1-0
Unfortunately, there are refutations to my attack :-( - so there's more than one
reason why no program will ever play 27.Qf2. The main thing is, that the concept
of the attack with Qh5 shows, how vision forms a target position (to reach via
the maneuvre Qf2-f3-h5 in this example).
The question is, if a chess program should be programmed to use this, or a
similar approach for something like planning. To check , if a target position
can be achieved at all, it would have to do a traditional search - and that's
what it does anyway. It seems to me, that the target position, if not too far
ahead, should appear in the search tree anyway. So, the human concept of
"vision" does not quite fit into a chess program's method as far as I understand
it, being no programmer.
Btw. there's a test position included here after 35...Kh8:
[D]1r5k/p5p1/2p1Q3/3p1pNP/5P2/3b2P1/P1q3B1/6K1 w - - 0 42
Here, I remembered a well-known mating pattern (I think it's called smothered
mate in english). Pattern recognition. 43.Qe6 is quite difficult to find for
some programs; from 10 tested at P2-333, only 5 found the move at tournament
time setting (programs from 1999 and earlier).
Regards,
M.Scheidl
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.