Re: computer-go: rules

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

I haven't tried to explicitely code any of his rules, just 
thought about some of them.  One example:  I took a lesson
from him on crosscuts, and he has a rule about when to extend
and when to atari.  It goes something like:  If you can save 
both of your stones, extend and fight.  If you can't, choose one
to sacrifice, and atari both sides.  He went through a bunch
of examples, and it helped my game.  But coding this for a
computer is difficult.  Either do a lot of local reading to see if
the stones can both be saved, or recognize many different local shapes.
A strong player can usually tell without much reading if the stones
can be saved, but there is a lot of unconscious pattern recognition
going on there that has to be mad explicit for the program.

David

At 07:08 PM 11/19/99 -0800, Ray Tayek wrote:
>
>i''ve been hanging around mr. yang for a number of years now (he lives 
>relatively nearby) and naturally have heard many of his "rules" in lessons. 
>his set of rules does seem fairly small. the problem i have in games is 
>knowing what the priorities are in different parts of the game as well as 
>having a bad evaluator (in my head) sometimes.
>
>could you tell us which if any of his rules seem to work well when coded 
>up? or perhaps why the one rule exploded so much?
>
>thanks
>
>---
>Ray (will hack java for food) http://home.pacbell.net/rtayek/
>hate spam? http://www.blighty.com/products/spade/
>
>