World Mind Sports Games Rules: News
Robert Jasiek
While the Simplified World Mind Sports Games Rules are simple and
clear, the official World Mind Sports Games Rules contain ambiguity
after the first succession of two passes. It particular is unclear
which intersections are "contestable", when a resumption is legal, and
which intersections are prohibited during a resumption. Participants
might want to read especially the section "Possible last minute rules
changes" below.
Apparently generally accepted aspects:
- the White passes first compensation rule and its yose value
- valuable 2-sided dame
- dame ko fights with dame as ko threats
- territory in (esp. asymmetrical) sekis
- superko fights
Things not to be resolved until the WMSG 2008:
- The missing positional / turn-given colour symmetry in shapes like
1-sided dame or 1000-year ko.
- bad but principally easily corrected wording of the rule text
- filling 1-point-eye instead of passing after a ko capture
- hane sekis and other pathological sekis
Possibly perceived problems by shapes:
- passing while there is unfinished ordinary yose, regardless of
whether
all the yose options might be pairs of miais, 2-sided dame, or
something else
- unfilled basic endgame kos (not to be confused with dead kos or
double
kos)
- 1-sided dame
- 1000-year ko
- dame that creates basic endgame ko versus sending-2-returning-1
Possibly generally perceived problems in the rules:
- what are contestable points
- yose after the first succession of two passes
Possible extreme approaches to treat contestable points:
- Robert Jasiek: Drop these rules.
- Nihon Kiin: Create more rules.
Classification of the shapes with endgame problems:
I) Each player can always strategically force resolution before the
first succession of passes:
- passing while there is unfinished ordinary yose, regardless of
whether
all the yose options might be pairs of miais, 2-sided dame, or
something else
- some positions with unfilled basic endgame kos
II) Not always each player can strategically force resolution before
the first succession of passes:
- some positions with unfilled basic endgame kos
- 1-sided dame
- 1000-year ko
- dame that creates basic endgame ko versus sending-2-returning-1
Examples for positions with unfilled basic endgame kos:
Type I:
. . . .
# # # #
. # O #
# O . O
O O O O
. . . .
Type II:
. . . . . .
# # # # # #
. # O # . .
# O . O # #
O O O O O O
. O . . . .
Possible last minute rules changes:
- "One may not pass if there is at least one unfilled basic endgame ko
on the board." This is a Chinese / Japanese compromise currently
discussed and likely to be applied during the WMSG2008. They speak of
"1 point kos" but what they mean is "basic endgame kos". Apparently
neither dead kos not double ko sekis shall be meant here. The
pathological case of "filling 1-point-eye instead of passing after a
ko capture" is ignored here at the moment.
- "After the first succession of two passes, filling 1-sided dame is
prohibited." It is unclear whether this or a similar rule instead will
be used. Thomas Hsiang, who is one of the tournament organizers and,
AFAIK, assistent chief referee, says though that this is his and other
referees' likely interpretation anyway.
Broken colour symmetry:
+--------------+
| . O O # . . .|
| # # O # . . .|
| # # O # # # #|
| . . O # O O O|
| O O O # O O O|
| # # # # O O .|
| . # . # O . O|
+--------------+
Black to move.
Variation 1:
+--------------+
| 1 O O # . . .|
| # # O # . . .|
| # # O # # # #|
| . . O # O O O|
| O O O # O O O|
| # # # # O O .|
| . # . # O . O|
+--------------+
2, 3 = pass.
White makes the first pass.
Score = (19+8) - (18+2) - 6.5 - 1 = 27 - 20 - 6.5 - 1 = -0.5.
Variation 2:
+--------------+
| . O O # . . .|
| # # O # . . .|
| # # O # # # #|
| . . O # O O O|
| O O O # O O O|
| # # # # O O .|
| . # . # O . O|
+--------------+
1, 2 = pass.
Black makes the first pass.
Agreement to end after 2.
Score = (18+8) - (18+2) - 6.5 - 0 = 26 - 20 - 6.5 - 0 = -0.5.
Variation 3:
+--------------+
| 3 O O # . . .|
| # # O # . . .|
| # # O # # # #|
| . . O # O O O|
| O O O # O O O|
| # # # # O O .|
| . # . # O . O|
+--------------+
1, 2, 4, 5 = pass.
Black makes the first pass.
Disagreement and resumption after 2.
Score = (19+8) - (18+2) - 6.5 - 0 = 27 - 20 - 6.5 - 0 = 0.5.
According to Thomas Hsiang, move 3 of variation 3 is prohibited.
Presumably also the resumption. Due to the ambiguity of the rules, one
cannot derive this from them clearly. Rather one has to accept the
official interpretation. A consequence is a broken colour symmetry:
+--------------+
| . # # O . . .|
| O O # O . . .|
| O O # O O O O|
| . . # O # # #|
| # # # O # # #|
| O O O O # # .|
| . O . O # . #|
+--------------+
Black to move
Variation Inverse-1:
+--------------+
| 2 # # O . . .|
| O O # O . . .|
| O O # O O O O|
| . . # O # # #|
| # # # O # # #|
| O O O O # # .|
| . O . O # . #|
+--------------+
1, 3, 4 = pass.
Black makes the first pass.
Score = (18+8) - (19+2) - 6.5 - 0 = 26 - 21 - 6.5 - 0 = -1.5.
This is a different rules treatment because White is never
endangered with possibly having to fill a 1-sided dame after
the first succession of passes.
It is against traditional go theory to treat a tsume-go
position and its colour-inverse tsume-go position differently!
Development of precedental rules if that should be wanted:
Treatment of 1-sided dame:
Replace the first succession of two passes by the first succession of
three(!) passes. (One might also allow a dynamic variation of either
two or, if a player requests it, three passes.)
Maybe 1000-year kos can be treated similarly (I have not checked this
yet).
Treatment of positions with basic endgame dame or unfilled basic
endgame kos:
Define (this may refer to two-eye-formation, shapes in question in
between independently-alive groups, and so depends on hypothetical
play) or presume definitions of "basic endgame dame" and "basic
endgame ko". Use a rule that prohibits both players (i.e., either
player) to pass until all basic endgame dame and all basic endgame kos
are filled.
Treatment of other special cases:
No treatment yet.
First attempt to interpret Thomas Hsiang's interpretation of rules
Section 5:
- The first succession of two passes is followed by the verbal phase
and then by either the short game end or the long game end.
- The short game end consists of the players' agreement on all the
stones to be removed, their removal, and then the determination of the
score.
- The long game end consists of the players' disagreement, the
continuation of alternating moves to the second succession of two
passes, and then the determination of the score.
- The stones removed during the short game end are called "dead". All
the stones on the board just after the second succession of two passes
of the long game end are called "alive".
- For all other purposes during the verbal phase, the status of each
string's stones is either alive or dead but the status assignment is
ambiguous intentionally.
- During the verbal phase, each empty intersection belongs to either
Black, White, or both players but the property assignment is ambiguous
intentionally.
- During the verbal phase, it is the players' duty to make status and
property assignments according to their playing strengths and
available rules knowledge. They may have a disagreement only
unintentionally, i.e., if an assignment question is not necessarily
obvious for all players of their playing strengths and levels of rules
knowledge.
- During the verbal phase, the "omniscient player" is an imagined
player with perfect playing strength, perfect rules knowledge, and
impartial best play for both Black and White.
- The omniscient player identifies all the unequivocally alive black
stones, all the unequivocally alive white stones, all the
unequivocally dead black stones, all the unequivocally dead white
stones, all the empty intersections that are Black's property
unequivocally, and all the empty intersections that are White's
property unequivocally. The remaining stones and empty intersections
are called "equivocal".
- During the long game end, it is prohibited to play on an empty
intersection that is equivocal or that is part of a connected region
of empty intersections adjacent to at least one equivocal stone or
contains at least one equivocal empty intersection.
- During the determination of the score, each intersection of a
connected region of empty intersections containing at least one
equivocal intersection is evenly divided between the two players.
Concluding, it is not impossible to clarify the rules (after further
work about all the details) but, for comparison, the Simplified WMSG
Rules are clearly simpler:
"After the first succession of two passes, the players may a) either
agree on all the stones to be removed, remove them, and then determine
the score or b) disagree, continue to alternate moves to the second
succession of two passes, and then determine the score."
Robert Jasiek
Regulations for Go Tournaments. It contains this additional ko rule:
>>
For controversies over half-point ko before the endgame, the owner of
the ko must end the ko and make the last play, despite the outcome of
the ko.
<<
This rule suffers from bad translation, wrong contents, doubtful
terms, and ambiguous semantics. So what might it want to express?
The most obviously, "endgame" is a wrong translation and should be
"game end".
Under WMSG Rules, "half-point ko" is a doubtful assessment of values.
A basic endgame ko can assume the deire values 1 or 3 when there are
or are not any dame. Dividing 1 by 2 gives "half-point" but why should
one divide by 2? With which model? Even then, it might also be the
other case of 3 divided by 2.
Speaking of "half-point ko" makes some sense (simplistic 50%
likelihood model of either ko connection or ko capture and connection)
under Territory Scoring but WMSG Rules do not use Territory Scoring.
Maybe we should forget about values and assume that "half-point ko" is
just an informal term for what might also be called "basic endgame
ko".
Who is the "owner of the ko"? The player for whom the ko is open? The
player who can win the ko? Something else? Let us assume the simplest
guess: The player for whom the ko is open. (I.e., the one who can
connect the ko stone by filling the ko.)
Why the last play? There can be several basic endgame kos on the
board. Does the rule apply only if there is only exactly one basic
endgame ko left? This would be a dangerous ruling because one would
have to conclude that the rule did not apply otherwise - when there
are more than one basic endgame ko still on the board. Let us be
pragmatic and assume implicitly that the rule has consequences for
several basic endgame kos: a transformation to the case of only one
basic endgame ko is expected.
What is "end the ko"? Probably just "fill the ko". For basic endgame
kos, the only reasonable ways to end something is a) to fill the ko or
b) to pass. Since the rule does not mention "pass", presumably (a) is
meant.
"must [...] make the last play": Why "last"? There might be, e.g.,
several 1-sided dame for the opponent. Ok, let us not be so devious
but consider a more ordinary case: The ko has been part of a dame ko
fight, the opponent has just filled the second last dame, the player
has run out of ko threats and so has to fill the ko (otherwise he
loses the ko), then the opponent plays the last dame - i.e., the
player cannot fulfil the rule's requirement to be the one to make the
last play. But let us see below for further arguments.
Here the condition "despite the outcome of the ko" comes in.
("outcome" is a bit ambiguous but we might pretend to understand it:
either a player fills or the opponent captures and the fills. The
outcome is about who might fill.) The condition, I am guessing, might
be interpreted somehow like this: The ko is not fought. If there are
ko threats, they are ignored and the players may not play them.
Instead the last basic endgame ko is filled without fighting the ko.
If there are dame (regardless of whether there might be a dame ko
fight), the dame are filled and then the open player fills the ko. (If
he gets the last two-sided dame, the opponent gets the first pass, and
then the player gets the "last play", where - I haven't mentioned this
further guess yet - approach moves to supposedly dead stones do not
count as possibly being last plays.)
What are "controversies"? Maybe "ko threats", "dame as ko threats",
"1-sided dame", etc. Maybe verbal disagreements? One cannot tell.
Maybe the rule means much less: After a player has won the ko, the
opponent passes (likely the game's first pass), and then the player
has to fill the ko.
Guessing again, dead kos or kos in other disturbing kos are not meant
to be "half point kos".
It is really tough to guess the intended contents of the rule. All the
above is pretty much nothing but guesswork. So what is the rule meant
to say?
Judging from hearsay, my guess of what the rule wanted to express is:
"A player may not pass if there is at least one _fighting ko_ on the
board and he has at least one legal play other than filling his own
liberty of a two-eye-formation."
Here, "fighting ko" is meant in the rough sense of being the opposite
of "disturbing ko". Note that I am close to defining fighting ko, but
this requires about half a dozen or dozen of pages of definitions.
Defining a more specialized type (basic endgame ko) is yet more
difficult. But if we assume that the rule's authors didn't care about
formalized theory and that they overlooked the "he has at least one
legal play other than filling his own liberty of a two-eye-formation"
side condition, it becomes:
"A player may not pass if there is at least one half point ko on the
board."
Apparently the only way to - hopefully - find out - at least part of -
the intention is to hear some explanation at the WMSG...
In particular, it is still an open question whether the first pass may
serve as a tedomari.
Robert Jasiek
application of the rules to very ordinary positions:
Preliminary note:
The 1st WMSG - Supplements to Regulations for Go Tournaments contain
this rule II.2:
"For controversies over half-point ko before the endgame, the owner of
the ko must end the ko and make the last play, despite the outcome of
the ko."
I refer to this rule as "the additional ko rule".
Questions related to the additional ko rule:
1) What is the intention and the meaning of the additional ko rule?
Note: Additionally, an application to these positions would already
give some hints at least:
. # # O O .
# # . # O O
. # # O O .
Black to move
. # # O O .
# # . # O O
. # # O O .
White to move
2) In the additional ko rule, is "endgame" a mistaken translation for
"game end"?
3) In the additional ko rule, what is a "ko"?
4) In the additional ko rule, is "half-point ko" a bad choice of words
for "basic endgame ko"? (Notes: Under Area Scoring or WMSG Scoring, it
does not make sense to speak of half-point ko in general because a
basic endgame ko can have the deire values 1 or 3. Half-point ko is an
appropriate term under Territory Scoring and there for a simplistic
likelihood model.)
5) In the additional ko rule, what is the "owner of the ko"?
6) In the additional ko rule, what is "end the ko"?
7) In the additional ko rule, what is "the outcome of the ko"?
8) In the additional ko rule, what is the "last play"? Shouldn't it
rather be called "last play of endgame value"?
9) In the additional ko rule, what are "controversies" and which may
occur?
10) When applying the additional ko rule to a position where, say,
Black is the owner of the ko and White might play on, e.g., three
1-sided dame, how can it be Black to make the last play and why is
White allowed to make plays later than Black's "last" play?
11) When applying the additional ko rule to a position with an
ordinary ko fight or a dame ko fight about a basic endgame ko, after
the owner of the ko has made the "last play" to end the ko, may the
opponent then still fill a two-sided dame?
12) How is the additional ko rule applied to a position with several
basic endgame kos?
13) Does the additional ko rule apply to hidden kos (like a) a hidden
ko next to a sente dame or b) a bent-4-in-the-corner)?
14) May the owner of the ko take the first pass and make the "last
play" later than that?
15) Wouldn't the additional ko rule's intention be expressed more
clearly by saying the following? "A player may not pass if there is at
least one 'fighting ko' on the board, unless the player's only legal
plays would fill a liberty of a two-eye-formation." Note: Although
"fighting ko" is still undefined, it is a clearer concept than "basic
endgame ko" because the latter is a subtype of the former.
Other questions:
16) What is a "contestable point" in general?
17) What in general distinguishes the "contestable points" from the
"not contestable points"?
18) Is it really so that 1-sided dame for Black are treated
differently from 1-sided dame for White? Notes: This is a consequence
of the game's first pass possibly contributing the game stopping's
succession of passes. These positions are interesting:
. O O # . . .
# # O # . . .
# # O # # # #
. . O # O O O
O O O # O O O
# # # # O O .
. # . # O . O
Black to move
. # # O . . .
O O # O . . .
O O # O O O O
. . # O # # #
# # # O # # #
O O O O # # .
. O . O # . #
Black to move
19) In which types of positions will the game's first pass serve as
tedomari? In particular, application of the rules should be
demonstrated to these positions:
. . . . . .
# # # # # #
. # O # . .
# O . O # #
O O O O O O
. O . . . .
Black to move
. # . # .
O # # . .
. O O # #
O . O O #
. O . O #
Black to move
Robert Jasiek
expresses about the intended meaning:
"When there are no two-sided dame on the board and there is exactly
one basic endgame ko on the board, the player who can connect the ko
stone has to do so on his next move, which therefore may not be a
pass."
Note that for the purpose of this rule, "basic endgame ko" should be
understood in a somewhat broad sense. In particular, a basic ko nearby
teire is also considered a basic endgame ko aka "half-point ko",
regardless of the ko's value.