On Tue, 2008-10-14 at 15:27 -0200, Mark Boon wrote: > On 14-okt-08, at 14:02, Don Dailey wrote: > > Mark Boon went off on a tangent here when he talked about a "swath of > > information available" and his imaginative discourse on how it > > might be > > used. He really launched into a different discussion and I don't > > disagree with him. It was just something else he was talking about. > > > > I have no idea whatsoever whether it's possible to build a better > > scoring algorithm based on taking advantage of a "whole swath of > > relevant information" but we were not talking about that. We were > > talking about using "margin of victory." > > When I read back this thread it does look like I misunderstood. I > thought that we were talking not only about the "margin of victory" > but also about the ownership map. My comments need to be read with > the latter in mind rather than the former.
The ownership map is very interesting to me. I made a lot of use of it in my old simple go program. But I probably could have squeezed out even more than I did. I didn't use it in the playouts themselves, but as separate information about which move to play. The primary thing I did was use it to make minor corrections to the score of each move "after the fact", once all the playouts were complete. It did help quite a bit in that context. Another use of it is for end of game dead stone agreement. The kgs version could pass early and report dead stones using the map created. It would pass early only when consecutive maps agreed (for 2 or 3 moves) and the certainty of them was reasonably high. And it was even correct most of the time :-) I think I remember that Mogo or some other strong program uses those maps somehow to improve the playouts a bit. It obviously makes sense not to move to points already decided, so perhaps that is how it can be used. But one must be pretty careful not to produce a self-fulfilling prophecy in this regard. If you make decisions based on the maps you could in some cases be enforcing a misconception (or be gradually changing the misconception in somewhat random ways.) > > I think that the further the game progresses, the more useful > ownership information becomes because areas become more independent > from each other. The closer to the end, the more a single playout > becomes equivalent to several runs. > > Mark > > _______________________________________________ > computer-go mailing list > computer-go@computer-go.org > http://www.computer-go.org/mailman/listinfo/computer-go/
signature.asc
Description: This is a digitally signed message part
_______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
