Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Positional Play (Only 2700+ GMs need respond please)

Author: walter irvin

Date: 06:44:28 10/24/99

Go up one level in this thread


On October 23, 1999 at 02:21:01, Bruce Moreland wrote:

>On October 22, 1999 at 23:34:02, walter irvin wrote:
>
>>i think computers can play at and beyond gm level . if a computer is fast enough
>>then who is to say where tactics end and strategy begins .i mean if a program
>>could hit 25 to 30 ply what chance would anyone have?????? all a computer needs
>>to see is the end result before the human can see it .
>
>You'd beat such a thing by giving it a long-term positional weakness and kicking
>it as hard as you could.  Eventually the thing would see that it is going to
>lose material, and it would fall over, vomit a few times, and die.
>
>Another way would be to get an obviously winning attack and let it play itself.
>
>Imagine you are going down a dead-end street that has no "dead end" sign.  If
>you can see a long ways, perhaps you can see the end of the street, realize it
>is a dead end, and turn around, before you get permanently stuck in there.
>
>But if the street is longer, you can't see the end.  OK, so perhaps you learn to
>see better, so you can see to the end of this one, but there are always longer
>dead-end streets.
>
>A strength of the GM player is the ability to identify nasty dead-end streets.
>
>bruce

your right about the gm strength ,but what is the gm weakness ? tactics and
exact memory of positions .a gm learns patterns of positions , where a computer
can store billions of exact positions .now WHEN a computer is fast enough it
makes up for its short comings by looking much much deeper than the man .

i know you wont believe this but it is true . if a program looked at nothing
other than material and king attacks it would still beat the strongest human
player if it could look deep enough .because if it looks deep enough it will see
the end of the game itself .but that kind of depth is not needed . i think
around 30 ply or so with a average (current)eval and perhaps 40 to 50 ply with
only material .also programs should only use openings that lead to open games
.computers perform 200 points stronger in open positions .




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.