Computer Chess Club Archives


Search

Terms

Messages

Subject: A Machine Still Cant Create Reality in Chess but It Should

Author: Rolf Tueschen

Date: 13:51:13 02/16/06


In a recent debate with RYBKA author Vasik Rajlich I could enjoy a short
exchange about knowledge and Vasik's opinion that for example a chess GM has
more and better knowledge than an amateur. Vas differentiated more knowledge
from just more complex knowledge with a lot of unintereting stuff. I
contradicted that concept with my view on a more knowledgeable GM who must NOT
have more knowledge as such but a quality of relevant knowledge that allowed him
to make an optimal judgement about any concrete situation of a particular game.
Then I tried to take a look into the thought process of a machine. THe point is
that even if we add the famous 30000 chunks, a human GM learns during his
maturing process, it is almost impossible for a machine to refine the essential
aspects of a concrete position. Because how can a machine decide what is
important and unimportant if that depends on smallest details? Of course a
machine actually has a certain helpful depth of mere calculation results. But
also then it's neccessary to decide what is the best because a pure brute-force
is impossible if you want to go so deep that a correct decision is possible for
a move with a line and all its implications.

I wrote with a smile that imagination is also knowledge in a human brain. Our
brain isnt a machine. All depends of our perception. This plasticity of the
brain makes us independent of determinism. Experience builds new neurons and
their connections. This unconscious inner reality functions independent of outer
reality and its cues. In the amygdala of our limbic system we create
imaginations of fantasies about traces of our perception. And on this base  we
start our actions. This is a puzzling process. The unconscious isnt a memory
localised in the amygdala or in another region. It's a system of memory traces
which are NOT a mirror of outer reality, although our inner reality is a social
reality too.

Perhaps chessplaying readers could now understand a little bit better what they
are doing in a game of chess. He knows now that his fantasies about an attack
for a mate isnt daydreaming at all. It is the inevitable process in our brain.
We create a fantasised reality and begin to analyse how we could reach the goal
in a concrete position. On the other side a programmer of chess software should
understand that his machine or program should have a goal. Only then a game gets
some meaning. Until now we see machines who make step for step and suddenly they
stand in front of a situation a chessmaster had prepared and where a former
advantage turns into a practically lost position. A machine steps through a game
of chess but a master creates a meaningful whole that a machine cant understand
because a machine has no fantasies and has still no idea of this process of
creation. But perhaps also a machine could come closer to that ideal if the
programmer gives it a chance to connect the actual situation with a fantasised
outcome on the board. That would also solve the old horizon dogma.

With all due apologies for very premature thoughts...from a newbie.

Rolf



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.