[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [computer-go] I know we disagree,but I choose to do nothingabout it.
The problem is that there is a difference between knowing that status of a
group, and correctly playing out a sequence to take stones off the board.
This gives programs a huge disincentive to pass when they are behind. The
opponent might make a mistake or lose on time. This would cause computer go
competitions to have games with dozens or hundreds of extra moves.
I for one, am more interested in working on go playing ability of the
program, and not working on playing well in the many extra moves after the
program would normally pass.
David
> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx
> [mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx] On Behalf Of
> William M. Shubert
> Let's say two programs disagree on the game situation. The
> standard GTP protocol has no way for them to figure out which
> is right, and which is wrong. So rather that favoring a
> protocol over understanding the game, what I'm doing is
> providing a protocol by which programs can determine which
> one of them had the correct understanding of the game
> situation, and which was wrong. The program that has the
> correct understanding of the game situation will be the one
> that benefits, as long as it can use the protocol...but on
> the other hand, isn't that the case for GTP (or any other
> protocol) in general? After all, you may have the best go
> playing program in the world, but if you don't implement the
> GTP protocol, then all your game understanding won't help you
> a bit in a GTP-based tournament. Same thing here; your
> program may be able to perfectly analyze living/dead groups
> at the end of the game, but if you don't have a way of
> showing why they are living and dead, then you won't win.
>
> _______________________________________________
> computer-go mailing list
> computer-go@xxxxxxxxxxxxxxxxx
> http://www.computer-go.org/mailman/listinfo/computer-go/
>
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/