Re: computer-go: WinHonte paper

[Gunnar Farnebäck <gunnar@xxxxxxxxxxxxxx>]

Darren wrote:
> P.S. Can anyone explain diagram 4 to me: I don't see how the white cut
> at B can do anything, and why black shouldn't have played A instead of
> his strange first line move.

The position is like this

.....O..XOO     .....O..XOO
...OOXX.XO.	...OOXX.XO.
.OOOX....XO	.OOOXB...XO
XOXX...X.XO	XOXX...X.XO
O.O.X....XO	O.OAX....XO
..........X	....*.....X
-----------	-----------

and black (X) to play the best endgame move on the left edge. If X
plays the move A, O can play hane in sente since X must connect
against the cut at B, giving this result:

.....O..XOO
...OOXX.XO.
.OOOX....XO
XOXX.X.X.XO
O.OXX....XO
..OOX.....X
-----------

If X instead plays at *, O can play atari at A. X defends and we have

.....O..XOO     .....O..XOO
...OOXX.XO.	...OOXX.XO.
.OOOX....XO	.OOOX....XO
XOXXX..X.XO	XOXXX..X.XO
O.OOX....XO	O.OOX....XO
....X.....X	..DCX.....X
-----------	-----------

Now O can of course descend at C and get a final result which is one
point better than the first diagram, but ending in gote instead of
sente, which usually is not worth it. Instead X has the option to play
D and O doesn't have liberties enough to block at C so X gains a
couple of points. (Ending in gote however, but if that turns out to be
undesirable X can still exchange C for D in sente, doing no worse than
in the first diagram.)

/Gunnar