Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Computer Go Report

Author: Albert Silver

Date: 07:41:44 09/16/05

Go up one level in this thread



>>>>Chess is 80% combinational 20% positional game.
>>>>While Go is 99->100% positional game.
>>>
>>>How did you calculate the percentage?
>>>
>>>>Computer Go can't beat human because of this.
>>>
>>>No
>>>The only reason is that programmers did not write programs that are good enough
>>>to do it.
>>>
>>>Uri
>>
>>Programmers do not write strong Go programs, because they don't know how to do
>>that for Go.
>>
>>And why don't they know?
>>
>>Because Go is 99->100% a positional game. :)
>>
>>Dave
>
>No this is not the reason that they do not know (otherwise they could have also
>problems with Backgammon).
>
>In BackGammon positional knowledge is more important than calculations but it
>does not prevent the top programs to play well.

While its true that it is essentially a positonal game, since even at 1-ply
(meaning no look-ahead) the strongest program is on average as strong or
stronger than the best players in the world, it is also a far simpler game to do
this. With 2 moves of look ahead, it clearly becomes stronger (on average) than
the best human players.

It was done with neural nets, and is extremely well documented on the internet,
however the depth of backgammon cannot possibly be compared with that of Go.
Think of how many move are required to complete a game, not to mention the
number of possible moves that can be played at any given moment.

                                         Albert

>
>I do not know much about go and I do not know if the relative advantage of
>humans relative to computers is mainly better static evaluation or knowledge
>which lines to search.
>
>Uri



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.