"Solved" means that you can play perfectly in any position. A few games have been solved.
"Weakly solved" means that you have a complete tree that cannot be defeated from the root position. Checkers has been weakly solved. Some forms of Go-moku have been solved. I do not think that 9x9 Go will be weakly solved in 50 years. I do think that a combination of great books and powerful software will approach that, within 5 years. Like Reversi (Othello). Brian From: [email protected] [mailto:[email protected]] On Behalf Of Don Dailey Sent: Wednesday, July 13, 2011 9:03 AM To: [email protected] Subject: Re: [Computer-go] 19x19 opening books Better terminology is, "approach perfect play" which I think is what was meant when someone claimed "weakly solved." Of course that is also an ambiguous phrase, but it probably captures the intent of the statement more accurately. Suppose that at 6.5 komi black wins. If a computer wins the majority of the games as black against humans we have evidence that the program is very strong but we still don't know how strong the human players are so this does not imply that the program is nearly perfect. However, if you take a candidate program that you believe is close to perfect and it wins 95% (or some arbitrarily high percentage) of it's games against itself when playing black (with 6.5 komi) then you have a much stronger belief that you are approaching perfect play. Since this is no proof, you cannot make any claims about the game being almost solved. Of course if such a strong player is based on massive opening book engineering, choosing a different komi probably invalidates the book work you have done so any claims much be based on a specific komi unless all reckoning and play is based on maximizing points on the board, certainly a much more difficult task. Don On Tue, Jul 12, 2011 at 11:42 PM, Álvaro Begué <[email protected]> wrote: On Tue, Jul 12, 2011 at 10:05 PM, Aja <[email protected]> wrote: > For Zen and CrazyStone, they might not be interested on 9x9, because 19x19 > is their arena. Mogo is maybe the best candidate. In the TAAI > conference last year in Taiwan, Olivier stated that Mogo will solve (or > weakly solve?) 9x9 by winning 4 out of 7 games against some top professional > player. > > Aja Ein? That's not what solving a game means. http://en.wikipedia.org/wiki/Solved_game Ąlvaro. > > > ----- Original Message ----- > From: Brian Sheppard > To: [email protected] ; 'Aja' > Sent: Wednesday, July 13, 2011 12:01 AM > Subject: RE: [Computer-go] 19x19 opening books > >>I think 9x9 go, even though compared to chess in complexity, is still more >> complex than chess and that the book will have a little less impact, >> although still a lot. > > > > > > My projection is the opposite: I think that 9x9 will be "played out" within > 5 years. Not weakly solved, exactly, but close to it. Zen and CrazyStone > have the ability to start on that project already. > > > > My impression is that the opening books are routinely worth a few hundred > rating points in 9x9 CGOS. > > > > I would cite Valkyria, which has a version that is playing near the top of > the CGOS ladder most of the time. A comparable version was playing ~200 > rating points within the last year, and I suspect that the opening book > knowledge that comes from its long-term memory is the dominant contributor. > > > > I also cite the Little Golem server, which is dominated by programs that > have opening books. > > > > Based on the work of Mogo and Valkyria, I suspect that if you take a pretty > good player and create a feedback system then you get a great opening book. > With an effective branching factor of maybe 2 to 3, you can get pretty far > into the game. > > > > Brian > > > > _______________________________________________ > Computer-go mailing list > [email protected] > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go > _______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
