[computer-go] Choosing a move in UCT (was: Re: The physics of Go playing strength.)

[Álvaro Begué]

> This is something that I have been wondering about. It seems to me that
> since UCT is exploring moves by choosing the move with the most
> optimistic score (estimate plus confidence bound) that it would be very
> natural to play the move with the highest conservative score (estimate
> minus confidence bound) but I don't believe I've seen this mentioned
> much if at all. Has anyone tried this and does it help or hurt any?
>
> I suppose the highest conservative score may very well be the most
> visited move most of the time, if not always, so maybe it doesn't
> actually make any difference?
>
> Janzert

A few days ago I found a discussion of these things in this very
mailing list. This is one of the most relevant posts, but you should
look around that whole thread:
http://computer-go.org/pipermail/computer-go/2006-October/006889.html

I think trying to continue the search until the best move and the most
explored move are the same is the most reasonable option.