RE: computer-go: Engineering (was: Most simple Go rules)

[Jean-Pierre Vesinet <jpvesinet@xxxxxxxxxx>]

> From: Mark Boon [mailto:tesuji@xxxxxxxxxxxxxxxxx]
> 
> - The Japanese rules work very well in real life play, as we all have
> experienced. For online play there's the problem of removing the dead
> stones. Current solutions that require an extra process of 
> the two players
> deciding which stones are dead I find very impractical. And 
> if there's a
> disagreement, there's a need for a way of determining which 
> one is correct.
> 
> So now you may understand better what I'm trying to 
> accomplish, which is
> adding a simple protocol that allow the Japanese rules (but 
> not just the
> Japanese rules) to be used more easily online and in addition 
> allowing for a
> simple dispute resolving mechanism.

  I like the way disputes are resolved in Goliath. I have not played Goliath
much and have no experience with any situation where the automatic life and
death detection fails. But in about twenty games I played against HandTalk
(19x19) it happened several times. Some seki positions are misevaluated as
dead, and some won semeais are misevaluated as sekis. HandTalk (and Wulu)
use a bypass: the user can click groups to change their status
life/death/seki at the end of the game. Of course this is not possible in
computer/computer games.
  I don't believe it is possible to ever fix completely these issues without
switching to area scoring. Japanese rules require perfect play or agreement
- without agreement, there is no alternative to perfect play. The problem of
accurately reading L&D with full Japanese 1989 rules (including pass for ko
and the tricky "enabling" part of the definition of life) is not solved
AFAIK: no one programmed it correctly, and even if someone did, it would be
far too slow to be used in practice. It is impossible to program Japanese
rules correctly IMHO.
  BTW, I agree with David Fotland's assertion of being able to compute the
correct score in _almost_ all games. But that's only statistical evidence. I
certainly don't want to belittle his efforts, but no amount of effort will
ever be able to fix the broken Japanese rules.

Jean-Pierre Vesinet - hopeless tsumego addict and part-time go programmer

Humpty Dumpty sat on a wall.
Humpty Dumpty had a great fall.
All the king's horses,
And all the king's men,
Couldn't put Humpty together again.