I suspect that other strong programs put a higher emphasis on the center than Many Faces. The old engine evaluation (which biases the search) tries to maximize the score, so it is a little territory oriented. On 13x13, the center is small, so being territory oriented gives an advantage.
David > -----Original Message----- > From: [email protected] [mailto:computer-go- > [email protected]] On Behalf Of Aja > Sent: Monday, May 30, 2011 11:22 PM > To: [email protected] > Subject: Re: [Computer-go] CrazyStone Western version > > Hello Ingo, > > > But there is a Monte-Carlo engine inside. > > Manually, I played a few test games against ManyFaces 12.022 which is > > the current commercial version of David Fotland. > > Giving approximately same times (which is difficult with CrazyStone, > > see above), CrazyStone seems to have (on "normal" dualcore PC's) > > the upper hand on 9x9 (1 game) and 19x19 (1 game). On 13x13, ManyFaces > > seems to be stronger (estimated from the flow of two games). > > It looks reasonable. ManyFaces was the 13x13 winner in last Olympiad. It > features moderate joseki playing, which seems to be a very helpful element > on 13x13. > > Aja > > _______________________________________________ > 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
