RE: [gnugo-devel] trying some OWL tuning

[Arend Bayer]

On Fri, 13 Sep 2002, Portela Fernand wrote:

> Arend wrote:
>
> > Lunch is not what you think it is, and it is GNU Go's fault you
> > misunderstood :-) Lunch is supposed to be a critical worm whose capture
> > would guarantuee an eye. I think the problem here is that the worm at K3
> > isn't considered dead. (This is due to a misanalysis by small_semeai()
> > here.)
>
> Well, I might misundertand some things, but
>
> http://www.public32.com/regress/?tstfile=13x13&num=20&trace=1
> shows this :
>
> K3 : B dragon size 6 (6.92), genus 0, half eyes 0, escape factor 0, crude
> status dead, status dead, moyo size pre owl 0, moyo size post owl 0, moyo
> territory value 0.00, safety dead, weakness 1.00, owl status dead
> ... neighbors L5 M1
> ... adjacent worm M1 is lunch
>
> It seems GNU Go thinks it's dead.

This is the result from the owl analysis. However, the dragon analysis for
the L5 is mostly based on the _tactical_ analysis (i.e. by what is found
out in reading.c) for the worm K3:
K3 : (dragon K3) black string of size 6 (6.92), genus 0: (3,0,0,0) - is
a cutting stone
- attackable at J1, attack code = 3
- attackable at N2, attack code = 3
- attackable at L1, attack code = 3
- attackable at D9, attack code = 3
- attackable at F5, attack code = 3
- defendable at L1, defend code = 3
... adjacent worm M1 is lunch

The line "defendable at L1" is incorrect (it should mean that the worm
can gain 5 liberties, or that it cannot be tactically captured within
the reading depth limits). If the worm K3 had no defense reported at L1,
it would be considered tactically dead already before the dragon analysis
starts at all. In this case I am pretty sure L5 would not be treated as
WEAKLY_ALIVE.

Arend