Re: [computer-go] Vincent's Game against SmartGo

[David G Doshay <ddoshay@xxxxxxx>]

On 6, Dec 2004, at 2:22 PM, Anders Kierulf wrote:

> I think we agree that a 10 ply global search with some sequences 
> explored to
> depth 30 would be a LOT stronger than today's programs. Whether it 
> would
> create problems for professional players is open for discussion.
SlugGo's lookahead of 16 with no branching seems to be good for about 4 
or
maybe 5 stones improvement against Many Faces from the base GNU Go
code.

I think that the question is still the evaluation function.

SlugGo will start playing games with a higher branching factor after 
the first
of the year, and I hope that we eventually get good ideas for evaluation
functions.


Cheers,
David


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