Re: [computer-go] Pattern matching - example play

["chrilly" <chrilly@xxxxxxxx>]

>
>Maybe Frank is the first person ever to write a program that plays
>better than the programmer?
>
This is in chess quite usual. For a book about computer chess I formulated
once a few laws of computer chess: One law is: Everbody - besides
Grandmasters - is able to write a good chess-programm. Deep Blue father Feng
Hsu wrote an article with the title: "Writing a GM-level chess programm
without knowing nothing about chess". He really did not know much about
chess, but this was also an ironic comment against H.Berliner.

>That is quite an interesting achievement by itself, I would say.
>
It is according to Chrilly´s law. There are also serious reasons behind
"Chrilly´s law".
a) Writing a piece of software is always a translation process from a
real-world problem, with an own language, semantic... into the language of
computing. As in every translation process it is easier to translate from a
foreign language into ones own language. So essentially one has to be a good
"hacker".
b) If I would program a walking robot I would not ask a sports-champion how
he walks. He does not know. I would ask my sister-in-law. She has
multiple-sklerosis and after each attack she must learn anew how to walk.
She knows much more about walking.
Being able to do something very good means, that one does not know how to do
it. It has become part of the unconsciousness.

The Dutch GM Donner wrote once an article called "Anti-Computer". One
central conclusion of Donner was: "The computer is an full-blown idiot. He
does not know anything and everything must be explained to him" (this is a
free translation, Donner formulated it more elegantly).
So probably another reason is: If someone is concerning the game an idiot,
he is on the same level.

Chrilly


_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/