Re: computer-go: RE: beat 9d before 2010 - no, size does matter

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

This is exactly the approach that most go programmers have taken.  Use
a domain expert, usually about amateur 6 dan, and try to capture the
knowledge in a program as a set of rules, usually expressed as a mixture
of C code and patterns.  Of course usually the domain expert and the
programmer
are the same person (bad), and usually the programmer has little if any
expertise in expert system techniques.  But no one has come close at
all to representing the amateur knowledge, and several have put about as
much time into it (10 person years) as you mention below.

I think you underestimate the difficulty of the problem, and the difficulty
of dealing with a system with so many rules :)

David

>
>BUT I am saying that we can build a Go playing program that does not come up
>against these BIG numbers at all. I am saying that we can capture the
knowledge
>that a 9 Dan Professional has in his head. Then our program will be a 9 Dan
>Professional. The program will beat the professional because the program will
>not get tired, misread a sequence, forget to take something into account or
>have an off day etc.
>
>The BIG numbers that concern me are things like:
>
>1. How much knowledge do I need to represent.
>2. How long will it take to elicit and structure.
>3. How powerful is the human brain compared to my PC.
>
>AND I think the answers are - VERY approximately:
>
>1. Not more than 100,000 rules
>2. About 5 years of full time effort for a team of two + good access to a
9 Dan
>Professional
>3. About the same to within an order of magnitude.
>
>Cheers
>
>David Elsdon
>
>
>
>
>