Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The reason why, beyond a point, deep searching doesn't pay off.

Author: Andrew Williams

Date: 02:45:55 05/15/99

Go up one level in this thread


On May 14, 1999 at 15:40:52, Timothy J. Frohlick wrote:

There have been a couple of articles on this point in ICCAJ. Both of them
suggested that a program searching to depth 14 will find a different best
move to the one it found when searching to depth 13 a significant proportion
of time. These experiments were done by Hyatt & Newborn in 1997 and Ernst
Heinz in 1998. The references are:

Heinz, E.A. DarkThought Goes Deep
ICCA Journal Volume 21 No 4

Hyatt, R.M and Newborn, M.M  Crafty Goes Deep
ICCA Journal Volume 20 No 2

Hope this is helpful.


Andrew


>I'm no mathematician but when you have a decision tree with many outcomes and
>you choose a particular path then the outcome is always contingent on unexpected
>events.
>
>We have all heard that grandmasters might say that they only look at a few moves
>and choose "the right move".  This is a "good enough" ploy and is found in most
>decision making; both animal and human.
>
>Going for a hike in the Alaskan wilderness can be most frustrating if you don't
>know what you are getting into.  Fighting the underbrush is exhausting.
>Similarly, getting into tactical complications with a computer only works if you
>can compute to 40 ply.  If you are in the brambles and you know that you will be
>there for another 50 miles then you will do what it takes to get through.  If
>you get hit by lightning or mauled by a bear that is a different story.
>
>The point is.... Chess is often won by unexpected but sound move combinations.
>The goal is to unbalance a position and lead your opponent into a risky venture
>while you are planning your own.  Whoever can program a computer to do this may
>become a very rich man/woman.



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.