Re: [computer-go] An [open] question on game tree search theory

[Gunnar Farnebäck <gunnar@xxxxxxxxxxxxxx>]

Antii wrote:
> I understand this result but in go it seems to be mostly irrelevant. In
> go almost any vacant intersection can be played (with suicide rule,
> *every*), so both players have the same number of options available.

Actually one-stone suicide tends to be forbidden also in rulesets
allowing multistone suicide, see http://senseis.xmp.net/?Suicide. In
any case it's kind of useless to have multiple ways to effectively
pass.

> If black can play to N positions, white can play to N-1 next, or N+K
> if black captures. But then white should in general be worse, not
> better off...

Sounds like a problem.

Otherwise, assuming at least one-stone suicide to be illegal, I would
expect minimax over random evaluations to favor forming small eyes
and, more crucially, not filling in small eyes. This is probably a
significant advantage over a pure random player, although still way
below measurable strength by human standards. If nothing else it
should improve the chances of getting a game against itself to finish
in some kind of reasonably settled state.

But why speculate when it can be tested?

/Gunnar
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/