README

1. What is komoku

Komoku is my attempt to write a program which is able to play go. As a general
idea, it uses monte carlo simulations of random games and tries to find out 
"the best move" by looking at the statistics. The best move (at a given situation)
should be the move with the best winning percentage (taking the randomly generated
follow-up games into account).

The strongest computer go programs right now use monte carlo methods, see 
MoGo (http://www.lri.fr/~teytaud/mogo.html) and Zen (http://senseis.xmp.net/?Zen)
to name a few. Zen has a rank of 3d at the Kiseido Go Server, which is a
strong amateur rank. This is very strong for a computer go program. With this 
in mind, I started implementing my own Go engine in a very nice language,
which is accidentally also called Go (www.golang.org).

But it seems my idea has been far too naive. I spent quite some time testing my
move generation implementation and profiling it, thinking that my program will 
not be terribly bad when I finally let it play against me. I didn't expect it
to be strong at all, but I thought that it might be a challenge to someone who
is just beginning to play go.

It seems my expectations have been far too high. I recently made komoku able to 
play, and it is so bad that I'm loosing hope and begin to think that it might be
better to either stop working on it or 'look for help'.

So you can find it here on github now. Perhaps someone will find something
useful in it. I didn't want to throw everything away.