On Tue, Jan 19, 2010 at 12:41:17PM -0800, terry mcintyre wrote:
> It's all about expectations -- at this point, we don't expect any SE to say
> at move 10: "This is a half-point win for white." ( I recall a pro making
> such an observation; I was willing to accept his expertise on the matter. )
>
> But if it says at move 160: "White wins by 5 points", it should be in the
> ballpark - or at least not so far out of the ballpark that a 16 kyu player
> wonders whether it might be playing in the wrong field entirely.
It is not a question of move number, but of the board status - how
unstable ownership of some board parts is (either due to a fight between
groups - middlegame, or too few stones in the area - fuseki). If some
board area is stable, it is rather easy to assess how many points it
will be worth.
The problem for SE is to deal with the unstable areas; perfect player
would score them accurately, imperfect player can do its best; it can
use bouzy maps (for empty areas), static life evaluation rules, tactical
tree search, montecarlo final owner statistics, or whatever else.
IOW, it is not about move numbers and score differences at all, but
about how complicated situations should the SE be able to resolve.
KGS SE is lacking in both the area of assessing stable areas which
should be easy (miscounting of seki positions and not handling
non-japanese rules properly) and of course in the really hard tasks
as well.
--
Petr "Pasky" Baudis
A lot of people have my books on their bookshelves.
That's the problem, they need to read them. -- Don Knuth
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
