On Tue, 24 Oct 2006, Heikki Levanto wrote:
> On Tue, Oct 24, 2006 at 07:26:34PM +0100, Jacques Basaldúa wrote: > > BTW: In my over 50K master games collection I have only seen 2 games > > with a triple KO. (The whole collection was played out by GnuGo 3.6 > > level 10 to verify/compute the final score.) And I have never seen > > other superkos than triple KO in real top games. In case of triple > > KO all SK definitions lead to the same, so we are probably discussing > > principles rather than facts. > > I believe this to be true for "normal" go, with a suitably large board. > The various ko rules seem to make a difference to pathologically small > board sizes, and other anomalies. Those may be an interesting field of > study, but seem not to be awfully relevant to the game of go, as we know > it. No. The probability of a ko arising is roughly proportional to the area, so it sort of grows quadratically with the side length of the board. I experienced this when generating life and death problems automatically that the percentage of problems with a status ko (given that these are unsettled positions where it matters who moves first) increases with the size of the problems. The same experience was made (and graphically displayed) when generating this data base of monolithic eyes up to size 11 that the positions in which the status is ko grew faster than the number of unsettled positions in general. Thomas > > Just my humble opinion, of course. I am sure to be corrected by more > knowing people on this list if I am wrong. Maybe even if I am not... > > -H > > > -- > Heikki Levanto "In Murphy We Turst" heikki (at) lsd (dot) dk > > _______________________________________________ > computer-go mailing list > [email protected] > http://www.computer-go.org/mailman/listinfo/computer-go/ >
_______________________________________________ computer-go mailing list [email protected] http://www.computer-go.org/mailman/listinfo/computer-go/
