[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [computer-go] Re: Random Go Challenge



Markus wrote:
> Gunnar Farnebäck wrote:
> > In order to get some testing of GTP implementations, and to some
> > extent because it's interesting in itself, I have started a Random Go
> > Challenge. For current standings and information about how to
> >
> > http://www.lysator.liu.se/~gunnar/gtp/random_go_challenge.html

I had planned to wait with announcing this to the computer go list
(the original announcement was to the GTP list) until it had
stabilized a bit more, but I guess it wouldn't hurt with some more
traffic.

> a very nice idea, though probably not a good benchmark for Go programs.
> The simple algorithm MaxAvgLib which tries to maximize the average number
>  of liberties outperforms all other programs.

Certainly not a good benchmark for real go strength, and I really hope
nobody was thinking it would be. Still I think it is of some interest
to see how different programs perform in this setting and I also
believe it's possible to improve the current top results
substantially.

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