Interesting position attached, white to play. Both Many Faces and Mogo
utterly miss the correct move (white H2). And they don't even get it
after black fills F4 - they both still believe they are winning (though
Many Faces' static score gets it right at that point, as does gogui
scoring).

Many Faces keeps playing on the left (even though set for Japanese rules
(*)). Mogo (at 225s per move) shows why it isn't the perfect test
position - it plays forcing moves at the top (F6, F7, G7, G8). It then
plays A4. But then after black F4 it still says it is winning with 98%
confidence. (the game continues H8, G6, E8, F8, B9, C8, F9, E9, A9, and
it is still very sure it has won.)

It's possible I don't have very latest version of either program. If
other people see it play the correct move please let me know what
version you are using.

Darren

*: Though if it doesn't see the seki then white has a 2.5pt win, so this
is just the standard MCTS playing for a 0.5pt win.

-- 
Darren Cook, Software Researcher/Developer
http://dcook.org/mlsn/ (English-Japanese-German-Chinese-Arabic
                        open source dictionary/semantic network)
http://dcook.org/work/ (About me and my work)
http://dcook.org/blogs.html (My blogs and articles)

Attachment: bad_for_uct1.sgf
Description: application/go-sgf

_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/

Reply via email to