[computer-go] an idea... computer go program's rank vs time

[Ray Tayek]

At 09:38 PM 1/21/2007, you wrote:
>On Sun, 2007-01-21 at 20:29 -0800, terry mcintyre wrote:
>...
> > Most programs reach scalability limits at some point.  ...
>
>Where most people go wrong is to assume that for a computer to beat
>a human it must be able to understand every single thing a human does
>but only better.    ...
>...

not everything and not better. but it's going to have to be "smart" 
about some go stuff and use heuristics (i am probably going off the 
deep end here).

take a look at some of the corner josekis, some of them has *many* 
variations (http://en.wikipedia.org/wiki/Taisha_joseki) and go for 
*many* moves (50+?). most humans can't choose the best variation that 
takes advantage of the stones in the adjacent corners (note that if 
there are stones on the side already, it is not a joseki).

since each player gets the same number of moves, the winner of a go 
game is the player whose stones surround territory most efficiently 
(modulo the number of groups).

this is why many players can play a player who is many stones weaker 
and give him *huge* amounts of time while using little time 
themselves. they know when a situation can be left and they move in a 
"bigger" place on the board. the stronger players stones will much 
more efficient so he wins easily.

thanks


---
vice-chair http://ocjug.org/