20060428 newest contents,
20090117 last update,
20060428 first day online,
Robert Jasiek
version 3; 20060414 day of the original
INDEX  GO  RULES

WAGC rules  FLAWS  MODEL
 INFORMAL rules  PRECEDENTS
 WAGC intention
Model for the World Amateur Go Championship Rules
In principle, the World Amateur Go Championship 1979 Rules (WAGC79) can
be modelled in a design similar to the Japanese
2003
Rules (J2003). However, modelling WAGC79 is significantly
more difficult
than modelling the Japanese 1989 Rules (J1989). Therefore a formal
model
for WAGC79 has not been found yet. Here the remaining key problems are
described: "alive" and "partially perfect pass".
Alive
The problem to define "alive" for a model of WAGC79 is already solved:
Step 1
WAGC79 tries firstly to define concepts of "eye shapes" and secondly to
derive the concept "two eye formation". This approach fails because
there
are arbitrarily many different "eye shapes" but only one kind of "two
eye
formation". The inverse order would work: firstly define "two eye
formation"
and secondly derive all "eye shapes". However, it turns out that
concepts
of "eye shapes" are entirely superfluous for defining "alive".
Therefore
a model for WAGC79 should omit all concepts of "eye shapes".
Step 2
The concept "two eye formation" works for so called independent life 
but what about so called coexisting life (seki)? J2003 has already
provided
suitable concepts: uncapturable, capturable1, and capturable2.
Step 3

In a position, a twoeyeformation is a set of one
or several strings
of the same player and exactly two empty intersections so that each of
the strings is adjacent to each of the two intersections, none of the
strings is adjacent to another empty intersection, and each of the two
intersections is adjacent only to the strings.
 In a position, a string of a player is twoeyealive
if the opponent
cannot force no intersection of the string with a twoeyeformation on.

J2003alive is defined like in J2003 as either
uncapturable, capturable1,
or capturable2.

In a position, a string is WAGCaliveinseki if it
is J2003alive
and not twoeyealive.

In a position, a string is WAGCalive if it is
either twoeyealive
or WAGCaliveinseki.

In a position, a string is WAGCdead unless it is
WAGCalive.
Step 4
Chris Dams has proven:

WAGCalive equals J2003alive.
Hence also the terms twoeyeformation and twoeyealive are
superfluous
for a model of WAGC79. It suffices to use J2003alive only. In other
words,
also for WAGC79 the terms uncapturable, capturable1, and capturable2
are the best choice to describe alive.
Partially Perfect Pass
WAGC79 does not simply allow passes but restricts them by restrictions
for direct kos, dame, teire, etc. For the researcher, the problem of
modelling
all this is a nightmare. It is as difficult as finding and defining
"capturable2"
in J2003. Even worse, several restrictions have to be considered. The
following
is a draft for them.
Restrictions

J2003like direct kos.

Prohibiting possibility of direct kos other than J2003like direct kos
and if localscoreimprovement is possible.

Prohibiting possibility of both such a direct ko and its colourinverse
direct ko on the board. Define a local environment to separate both.

Filling of all dame in between independently alive strings.

Filling of all 1sided or 2sided dame in sekis.

Removal of prisoners from sekis while maintaining their seki status.

Answer removal of prisoners from sekis while maintaining their seki
status.

Prohibiting possibility of surviving invasion in territory.

Prohibiting possibility of surviving invasion under twoeyealive
stones.

Prohibiting the opponent's possibility of scoreimprovement for the
opponent
over all the scoringpoints in all the player's stricttwoeyeregions
considered together. Consider also the prisoners made there during
possible
continuations. Possibly require at least one new permanent stone of the
opponent during a possible continuation and inside at least one of the
opponent's (?) stricttwoeyeregions?

Prohibiting the opponent's possibility of scoreimprovement for the
opponent
over one particular stricttwoeyeregion of the player. Presumably the
player might have to defend with more plays than the opponent uses to
invade.
The opponent's invasion stones, as far as they are in the particular
stricttwoeyeregion
of the player, all have to be removed as prisoners. Globally the
scoreeffect
is neutral (otherwise the previous defense case applies).

WAGC rules speak of defense inside territory. If
all independentdame
are filled and there is some teire, then this teire is not
surrounded
by twoeyealive stones, so WAGC rules have the mistake of not also
requiring
to obtain territory per then
(immediately or a little later)
existing
local strictterritorystring. Such a defense to obtain territory is
necessary.
Terms
Besides, many related terms like scoreimprovement,
scoremaintenance,
independentdame, stricttwoeyeregion, etc. have to be defined so
that
those terms can be used by rules for the restrictions.
Force
Because WAGC79 sets life and death requirements for both the
alternation
and the analysis, defining one type of "force", as it is done in J2003,
might not be enough. This does not make research easier. At least, all
kinds of "force" are related. Whichever definitions will be necessary
here,
they can be derived from that in J2003. While it could make a model for
WAGC79 much longer, it is a comparatively easy remaining problem for
the
researcher.
Comparison
A good model for WAGC79 will be like J2003 with additional rules and
terms
to achieve a partially perfect pass. Concluding, necessarily
application
of WAGC79 is more difficult than application of J2003, which models
J1989.