Re: On Game Space Size

[Henrik Rydberg <rydberg@xxxxxxxxxxxxxx>]

Several people have during the last week or so been interested
in the question of how many legal states there are in Go space.
I guess the answer has been around for a long time, even.
So what's the point?

Clearly, a Go game without capturing stones, or threatening
to, is no Go game at all. Moreover, it would be very, *very*
easy to predict.

Hence, the illegal states are important. They are the points
in Go-space where a drastic transition occurs. Every time we
encounter an illegal state, it means something was captured,
or we entered a Ko, or something else exciting happened.
The rest of Go-space is smooth like a baby.

Now, since it is these illegal states in Go-space that make
life so complicated for a Go programmer, it's a natural first
question to ask: How many are they? How are they distributed?
The contributions of Antii Huima, Pieter Cuijpers, Mousheng Xu,
Avijit Ghosh, John Aspinall and latest Christopher Birk
(hope I did'nt forget anyone) all add to the picture,
and above all, show (IMHO) the *right spirit*. Good work, guys!

My own ultimate goal, as a Go programmer, is to understand
enough about Go space so as to be able to transform the game
of Go into a form which is more computationally favourable than
changing board states and counting liberties. I bet we all
have our own favourite goal.

Henrik

-- 
Henrik Rydberg (http://fy.chalmers.se/~rydberg),
Department of Applied Physics, Chalmers University of Technology.