Computer Chess Club Archives


Search

Terms

Messages

Subject: algorithms for human brains !?

Author: Werner Kraft

Date: 04:23:22 05/06/05


In Reply to: Re: Any alternative ways of evaluating positions ?
<message.html?424272> posted by Uri Blass <mailto:uri@agur.co.il> on May 04,
2005 at 18:16:58:
I mean that is all very nice...
It is relativy easy to do an internet search fur human be able to take the roots
of very large number.
Simple method to find e.g approximate squre root of 276 000:

 500 squared =  250 000
600 squared = 360 000

therefore squre root closer to 500

Intelligent guess : 520 ?

250 000 +  2* 500*20 =  270 000 , ergo sth like 253, 254

 Iused the formula :  ( a + b ) square + approx  a squrae + 2ab

Now, that is a first approximation for  evaluating chess positions quickly in a
human brain -
we need to find algorithms, that a human player can use, to find quick ,
approximate solutins

Mathematical solutions:

Topology was just a kind of intuitive guess.
New solutions ( e.g for tough problems in theoretical physics )  come usually
from  bizarre areas of mathematics .

Next guess : What about chaos theory (  attractors - and a small change in
initial conditions can lead to big variation further down the tree ? )
Posted by Werner Kraft <mailto:warnerk2@yahoo.com> (Profile
<../profile.html?warnerk2@yahoo.com>) on May 04, 2005 at 14:13:10:
In computer chess , the method to find  a move is based on " tree - searching "
- you
have a move ( 1.e4 ) - then the 20  reply moves for black . SO you build up
trees - every variation move is a new branch on it.
Now - as  a complete analysis of a chess game involves 10-power 120  positions -
you introduce evaluation system ...

Was there ever another way of looking at that -  may be from exotic branches of
mathematics - topology e,g ?

I was also thinking about new methods for humans to beat very powerful machines-
I mean , there are some gifted individuals , who can calculate roots from large
numbers - and they must use some kind mathematical system - shortcuts, that
allow them to approximate etc.

Now , what if there would exist some methods to allow human chess players the
same thing - a kind of a human way to calculate " Naidorf tables " in the head ?
Would that be the tool for The ELO 3500 player of the future ? :-)





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.