Author: Andrew Dados
Date: 00:23:38 02/20/00
Go up one level in this thread
On February 19, 2000 at 22:28:25, Tom Kerrigan wrote:
>On February 19, 2000 at 18:35:43, Thorsten Czub wrote:
>
>>the content of the game, the HOW and why and style and all the stuff that
>>cannot be valued in the 3 stages above, is the game itself.
>>this is the quantum-mechanics !
>
>Quantum mechanics is a branch of physics. It deals with physical things. Hence
>the name.
>
>There are some physical things in chess, like chess pieces and chess boards.
>
>But the game itself has absolutely nothing to do with physics.
>
>So please don't draw confused, esoteric links between the two.
>
>Personally, I think you can get your point across just as well by using words
>like "beauty" and "opinion."
>
>-Tom
Let me be a dumb determinist and pseudo-scientins and try to define 'beuty' in
chess.
But first somehow long prologue.
Game of chess has finite number of states (positions). The game rules, as
defined, limit transitions from one state to another ('making moves'). Finite
number of states (although huge number) and clearly defined transition rules
assure that it is a deterministic game. It can be solved in several ways (by
backsolving, for example - same way TBs are created). I don't argue if it will
be solved anytime soon. I just state that each position has a property (let's
call it 'evaluation') which determines game outcome if optimal moves were made.
Each position is in one of 3 states: win, draw or lost. Making a move can not
shift state 'up'; it can not change 'objective' outcome of perfectly played game
from, say, draw to win. It can only retain position state or worsen it.
So: if game was 'draw', then after my move it can be 'draw' or 'lost', never
'won'.
Now: a 'blunder' is the move which worsens position 'objective' evaluation.
There are two very important properties of positions which we try to maximize
(or minimize) during game: probablities of blunder for opponent (OB) and
myself(MB).
So, for example, we try to make moves which transpose
draw(OB1,MB1)->draw(OB2,MB2), when either OB1<OB2 or MB1<MB2 or both. But we can
also try moves which transpose draw(OB1,MB1)->lost(OB2,MB2), where OB1<<OB2, so
objectively we blunder, but we maximize chances that opponent will go wrong and
we will win anyway. (we don't know 'objective' score, anyway).
Now my _guess_ is that what most call 'beauty' is taking a chance that move we
made can be a blunder but opponent will go wrong.
For computerchess it means evaluation which tries to foresee events beyond
alpha-beta horizon in advance. There is little beauty in 'simple', 'technical'
endgames, but it's easiest to make evaluation for those positions (example:
famous overevaluating connected passers) . For middlegame it means sacs lowering
king safety with no material back within computable event horizon.
If you take time to ponder on the above, you will understand that
'outsearching the opponent' is a nonsense-term. Opponent blundered, that's it. I
just found that narrow way to explore it. So: if bean-counter won, long live
bean-counter. I blundered. I was not 'outsearched'; I just made a losing
'mistake' (Or I could not keep my won position - anyway: I blundered). Always
blame yourself in the games...
All the above is N/A for life, quantum phisics, etc....
...at least for me :)
-Andrew-
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.