Re: [computer-go] Third KGS tournament: game-end protocol

[Gunnar Farnebäck <gunnar@xxxxxxxxxxxxxx>]

Peter McKenzie wrote:
> There seems to be a feeling that it is unreasonable to expect programmers to 
> write programs that follow this protocol but, speaking as a programmer, it 
> doesn't seem that hard.

If you don't care about losing points it's not hard at all, just play
as many liberties of the opponent stones on the board as you can in
some arbitrary order. The stones remaining after that are
unconditionally alive with no room for uncertainty.

If you do care, it's a lot harder than it may seem and I doubt that
it's time well spent for most people, unless they happen to enjoy that
particular problem.

> Note that with the rules as listed above you can still happily enter a 
> program that plays to the death and kills everything (in a computer 
> tournament I will probably just have my program play that way anyway).  
> Having a program that passes before killing all stones is simply a finese 
> that is available under these rules.  The key thing is that ALL programs 
> should be able to kill dead stones if required (or forfeit those points) and 
> I don't see a problem with that.

If you design your program to play under Japanese rules it's not
particularly natural to play until dead stones have been removed and
in certain cases you are not even expected to be able to do that.

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