[gnugo-devel] Weakness calculation

[Inge Wallin]

Consider the test 13x13b:43.

   A B C D E F G H J K L M N
13 . . . . . . . . O O X . . 13
12 . . . . . . O O O X X X . 12
11 . . . O O . O X O O X . . 11
10 O O O + . . X X . X X X . 10
 9 X O X X . X . X . X O O O 9
 8 X O X x . . X X O O . . . 8
 7 X X X X X O X O O O O O . 7
 6 . . . . . O X O X O X O . 6
 5 X X O . X O O O X X X O O 5
 4 O X X X X O X O . X . X O 4
 3 O O O O O X X O . X O O X 3
 2 . . O . O O X X . X X X X 2
 1 . . . . . O O X . . . . . 1
   A B C D E F G H J K L M N

X to move.

The problem here is that Gnu go plays J1 instead of J3.  But this is
not what I want to talk about here.

What I want to talk about is the weakness values for the dragons at A4
and J8.  Both dragons have weakness values of 0.13, both pre- and post
owl.

The total weakness value is calculated as a weighted sum of the
partial weakness values gained from the genus, the moyo and the escape
value.  Here is what Gnu go reports for A4:

> Computing weakness of dragon at A4:
>   moyo value 8.00 -> 0.20, escape 0.00 -> 1.00, eyes 2.50 -> 0.10
>  result: 0.13.

I have two questions:

1. Why would this rock solid group get a weakness value that is
   significantly greater than zero?  The genus alone should be enough
   to say that it is totally non-weak.

   The weakness values are interpolated from a table where genus 3.0
   gives weakness 0.0 and genus 2.5 gives weakness 0.10 and genus 2.0
   gives weakness 0.2.  Shouldn't genus 2.5 indicate a totally stable
   group and thus have a weakness value very close to 0?

2. Why is the moyo 8.0?  Intuitively, the moyo should be on the
   outside, i.e. not overlapping with the territory and eyes?.  In
   this case, it is also a bit strange that the weakness calculation
   takes into account the same points in both the eye value and the
   moyo value.

I propose to change the interpolation values for the genus, but before
doing so, I wanted to hear if there was some other, undocumented,
reason for the values as they are now.

In the end, this will lower the strategic effects, something that SP
Lee reported in the mail
http://mail.gnu.org/archive/html/gnugo-devel/2003-06/msg00213.html 

       -Inge