On 07/21/2010 07:44 AM, Darren Cook wrote:
I think it is entirely plausible that even 5.5 komi favours white. I've had a program chugging away since December analyzing 9x9 games, that I hope will give strong evidence one way or the other, but it needs a few more months/years yet.
Could you describe your program and the preliminary results that make you think 5.5 favors white? If you have some kind of database or book that you could snapshot and publish that would be even better, as it would be possible to verify your results with other programs.
Right now I'm running the public version of MoGo against itself, no book, at 6.5 komi, with 100k playouts, and so far Black has the advantage (60.5%, 703 games). I plan to let it go to 3,000 games.
Admittedly this is a weak test, as there's no book and 100,000 playouts isn't that much. The games usually complete in under 5 minutes. More playouts with a book, and using other programs like Fuego, would be good to add to the mix.
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
