Japanese Rules: Enabling Fun (2)

1 Aufruf
Direkt zur ersten ungelesenen Nachricht

Robert Jasiek

ungelesen,
24.08.2003, 08:16:0224.08.03
an
You thought that you understood example 10 of the J1989's official
commentary part I?

9x3 board
position after the game end

# O . . # O . O # .
. # # # # O O O # #
# # O O O O . O # .

intersections

b c d e # O . O # .
a # # # # O O O # #
# # O O O O . O # .

Is the black string b capturable-alive or dead? Certainly, b is
an enabling intersection. However, it is not clear why a, c, d,
or e should be enabling: #[dpapp], #[dpcpp], #[dpp], or #[dpepp],
respectively, allow Black to play a not captured stone there in
any case. So if b is the only enabling intersection, then the
black string b is dead: O[aBbdbapcpbe..]. (Capital letters are
ko-passes, p is pass.)

If you allow some intersections of a, c, d, e to be enabling or
some move-sequences to be enabling that capture b but play a
stone on some of these intersections, then you are in serious
trouble in case of other shapes.

Therefore I prefer to solve the problem by allowing capturable
life also if the player can create a two-eye formation with b
being one of the eyes. I.e. instead of playing a new not
captured stone the player may also create an uncapturable eye.

***

Now I really worry whether there are sekis in that a player
cannot play a new uncapturable stone but create an uncapturable
seki-one-point-eye of a string having another shared liberty...

--
robert jasiek

Bill Spight

ungelesen,
24.08.2003, 09:23:5224.08.03
an
Cher Robert,

> 9x3 board
> position after the game end
>
> # O . . # O . O # .
> . # # # # O O O # #
> # # O O O O . O # .
>
> intersections
>
> b c d e # O . O # .
> a # # # # O O O # #
> # # O O O O . O # .
>
> Is the black string b capturable-alive or dead? Certainly, b is
> an enabling intersection. However, it is not clear why a, c, d,
> or e should be enabling: #[dpapp], #[dpcpp], #[dpp], or #[dpepp],
> respectively, allow Black to play a not captured stone there in
> any case. So if b is the only enabling intersection, then the
> black string b is dead: O[aBbdbapcpbe..]. (Capital letters are
> ko-passes, p is pass.)
>
> If you allow some intersections of a, c, d, e to be enabling or
> some move-sequences to be enabling that capture b but play a
> stone on some of these intersections, then you are in serious
> trouble in case of other shapes.

Interesting. ;-)

Suppose that you restrict enabling intersections to those that cannot be
played *immediately*. Then 'a', 'b', 'c', and 'e' would all be enabling.
Each of 'b' and 'c' has a stone on it, 'a' dies when White replies at
'd', and 'e' dies in ko.

Best,

Bill

Robert Jasiek

ungelesen,
24.08.2003, 10:24:4724.08.03
an
Dear Bill,

> Suppose that you restrict enabling intersections to those that cannot be
> played *immediately*. Then 'a', 'b', 'c', and 'e' would all be enabling.
> Each of 'b' and 'c' has a stone on it, 'a' dies when White replies at
> 'd', and 'e' dies in ko.

Currently Bernd Gramlich believes in "immediately" but I have still
to see a precise definition and tests in some 50 standard examples:)

Best,
--
robert jasiek

Bernd Gramlich

ungelesen,
24.08.2003, 10:57:5924.08.03
an
Robert Jasiek wrote:

In absence of kos, my "definition" of a capturable living string reads:

A "capturable living string" is a string whose capture allows the
defender to play either another capturable living string or an
uncapturable stone on an intersection where he couldn't immediately
have played one in the starting position.

This works well with the "official" example you posted.

If there are kos on the board, you need some kind of localization in
order to avoid messy side effects. Localization using the defender's
kolessly uncapturable strings as borders seems to work well, but I
haven't applied it to all of the official bestiary yet.

--
Bernd Gramlich [bE6nd "gRamlIC]

Robert Jasiek

ungelesen,
24.08.2003, 11:08:0324.08.03
an

Bernd Gramlich wrote:
> In absence of kos, my "definition" of a capturable living string reads:
>
> A "capturable living string" is a string whose capture allows the
> defender to play either another capturable living string or an
> uncapturable stone on an intersection where he couldn't immediately
> have played one in the starting position.

This is not precise enough for me. You should explain
"allows", "uncapturable stone", "could not", "immediately".

--
robert jasiek

ro...@telus.net

ungelesen,
24.08.2003, 14:02:3324.08.03
an
On Sun, 24 Aug 2003 14:16:02 +0200, Robert Jasiek <jas...@snafu.de>
wrote:

>Is the black string

You mean, "unit"?

-- Roy L

Robert Jasiek

ungelesen,
24.08.2003, 15:03:5724.08.03
an

Since you ask, see below. It is not verified yet but should give you
a pretty good idea what I do intend to mean ;)

--
robert jasiek


Meta-language Conventions

Headlines are not part of the rules.
"" denote and bracket an axiomatic or a defined textual term.
Lower case and upper case letters are distinguished in a term.
Bold text indicates the place where a term appears as an axiom or is
defined.
[Sorry, here text in bold does not appear like it should.
It should be obvious what shall be bold.]
iff is if and only if.


Rules

Axioms


Let there be "Black", "White".

Let there be "empty", "black", "white".

Let there be "1", "2", ..., "361".

Basic Definitions


"player" is an element of {"Black", "White"}.

"colour" is an element of {"empty", "black", "white"}.

"intersection" is an element of {"1", "2", ..., "361"}.

Advanced Basic Definitions


"intersection-colour" is (I,C), where I is an "intersection" and C is a
"colour".

"horizontal" is {{"1", "2"}, {"2", "3"}, ..., {"18", "19"}; {"20",
"21"}, {"21", "22"}, ..., {"37", "38"}; ..., {"343", "344"}, {"344",
"345"}, ..., {"360", "361"}}.

"vertical" is {{"1", "20"}, {"2", "21"}, ..., {"19", "38"}; {"20",
"39"}, {"21", "40"}, ..., {"38", "57"}; ..., {"324", "343"}, {"325",
"344"}, ..., {"342", "361"}}.

"grid" is the union of "horizontal" and "vertical".

"line" is an element of "grid".

"intersection" I and "intersection" J are "adjacent" iff {I, J} is a
"line".

"path" is (I[1], I[2], ..., I[z]), where z>1 and I[j]<>I[k] for all
j<>k; j,k=1..z and (I[l],I[l+1]) is a
"line" for all 1<=l<z.

"position" is ((I[1],C[1]), (I[2],C[2]), ..., (I[j],C[j]), ...,
(I[361],C[361])), where I[j] = "j" and C[j] is a "colour" for j=1..361 .

Given a "position" ((I[1],C[1]), (I[2],C[2]), ..., (I[j],C[j]), ...,
(I[361],C[361])). Then "string" is the union of "intersection" I[j] and
all elements of {A | A is "intersection" and (X[1], X[2], ..., X[k],
..., X[y]) is a "path" and y>1 and 1<=k<=y and X[1] = I[j] and X[y] = A
and there is 1<=l<=361 so that X[k] = I[l] for all k and C[l] = C[j]}.

Given a "position" ((I[1],C[1]), (I[2],C[2]), ..., (I[j],C[j]), ...,
(I[361],C[361])), where C[j] = "empty". Then "empty-string" is the union
of "intersection" I[j] and all elements of {A | A is "intersection" and
(X[1], X[2], ..., X[k], ..., X[y]) is a "path" and y>1 and 1<=k<=y and
X[1] = I[j] and X[y] = A and there is 1<=l<=361 so that X[k] = I[l] for
all k and C[l] = C[j]}.

Given a "position" ((I[1],C[1]), (I[2],C[2]), ..., (I[j],C[j]), ...,
(I[361],C[361])), where C[j] = "black". Then "black-string" is the union
of "intersection" I[j] and all elements of {A | A is "intersection" and
(X[1], X[2], ..., X[k], ..., X[y]) is a "path" and y>1 and 1<=k<=y and
X[1] = I[j] and X[y] = A and there is 1<=l<=361 so that X[k] = I[l] for
all k and C[l] = C[j]}.

Given a "position" ((I[1],C[1]), (I[2],C[2]), ..., (I[j],C[j]), ...,
(I[361],C[361])), where C[j] = "white". Then "white-string" is the union
of "intersection" I[j] and all elements of {A | A is "intersection" and
(X[1], X[2], ..., X[k], ..., X[y]) is a "path" and y>1 and 1<=k<=y and
X[1] = I[j] and X[y] = A and there is 1<=l<=361 so that X[k] = I[l] for
all k and C[l] = C[j]}.

[I have not gone beyond "string" so far.]

Ted S.

ungelesen,
24.08.2003, 17:27:4524.08.03
an
Somebody claiming to be Robert Jasiek <jas...@snafu.de> wrote in
news:3F48AC82...@snafu.de:

>9x3 board
>position after the game end
>
># O . . # O . O # .
>. # # # # O O O # #
># # O O O O . O # .
>
>intersections
>
>b c d e # O . O # .
>a # # # # O O O # #
># # O O O O . O # .
>
>Is the black string b capturable-alive or dead?

Yes. It most definitely is one or the other, or both. :-)

Seriously, I don't get your point. Assuming White to move, the best White
can do is to capture the Black stone at B. Black has to pass, having no
ko threats, and White can't do anything (since White has no ko threats
either). Since "game stop" (or whatever term you wish to use for the time
just before any dame are filled and the score is counted) doesn't depend
on consecutive passes under Japanese rules (this isn't computer server
rules!), White can't get out of it with a second pass.

I highly doubt this position would be a problem in a normal 19x19 game,
since Black could just fill in a dame or empty point someplace in his own
territory.

Does anybody actually play the way you do, Robert, and do these positions
ever come up in your real games (eg. EGF games)?

I swear that one of these days, we're going to see you write "It depends
on what the definition of 'is' is."

--
Ted S.: change .spam to .net to reply by e-mail
Homer Simpson: I'm sorry Marge, but sometimes I think we're the worst
family in town.
Marge: Maybe we should move to a larger community.
<http://www.snpp.com/episodes/7G04.html>

Bill Spight

ungelesen,
24.08.2003, 17:54:5024.08.03
an
Dear Ted,

> >9x3 board
> >position after the game end
> >
> ># O . . # O . O # .
> >. # # # # O O O # #
> ># # O O O O . O # .

[snip]

> I highly doubt this position would be a problem in a normal 19x19 game,
> since Black could just fill in a dame or empty point someplace in his own
> territory.
>

In fact, positions like those in the left of the 3x9 board do occur in
real games on the 19x19 beard. (This one would be in a corner.) It is
unusual for them to remain on the board after the end of the game, but
that happens, too.

In theory, Black has 5 points of territory there, and that is the result
by the Japanese rules. Black does *not* have to capture the lone White
stone there. By the convoluted logic of the Japanese '89 rules, the
White stone is dead while the Black stone in the corner is alive. (If
not, Black has no territory there.) The wording of the article is
unclear, however. We are sure of this position because it has been
explained in the commentary.

Robert is attempting to find a rigorous set of rules that produce the
same results as the examples in the official Japanese commentary (and
maybe others where the ruling is fairly clear). While rigorous, such
rules may be even more convoluted than the Japanese '89 rules.

Best regards,

Bill

Robert Jasiek

ungelesen,
25.08.2003, 04:34:1025.08.03
an

"Ted S." wrote:


> Somebody claiming to be Robert Jasiek <jas...@snafu.de> wrote:
> >9x3 board
> >position after the game end
> >
> ># O . . # O . O # .
> >. # # # # O O O # #
> ># # O O O O . O # .
> >
> >intersections
> >
> >b c d e # O . O # .
> >a # # # # O O O # #
> ># # O O O O . O # .
> >
> >Is the black string b capturable-alive or dead?
>
> Yes. It most definitely is one or the other, or both. :-)
>
> Seriously, I don't get your point.

If this position is on the board during scoring, then one has to
score it. The Japanese 1989 Rules fail to explain correctly how
to score this position while explaining at the same time how to
score

example (2)

. # . O #
# # # # #
O O O O O
. O . O .

intersections

. # a b #
# # # # #
O O O O O
. O . O c

Since it has to be possible to score both positions due to the same
rules, one has to write or interpret them in a way that allows that.

Let us restrict discussion to the black corner stone in example (1)
and the single white stone in example (2). It is necessary to
determine whether either is either capturable-alive or dead.

The J1989 Rules make that distinction by referring to the word
"enable" while the [string] in question is captured [during the
same hypothetical-sequence of hypothetical-analysis]. The J1989
Rules fail to define "enable". Therefore one has to provide an
interpretation of "enable". That is what I am trying to do.
Without such an interpretation one can never in general distinguish
between strings that are capturable-alive or dead; one would have
to judge them "capturable-alive or dead"; the score of the game
would be undefined.

So let us try to interpret "enable":
(Capital letters are ko-passes, p is a regular pass.)

Interpretation (1): The player whose string is captured may play
a new uncapturable stone anywhere.

- In example (1): O[aBbdbpp] or O[aBcbAecdpp]. Black has succeeded
to play a new uncapturable stone at d. So Black's corner string is
capturable-alive.
- In example (2): #[acpp]. White has succeeded to play a new
uncapturable stone at c. So White's single stone is
capturable-alive.

Obviously, interpretation (1) fails in example (2). Therefore
we have to have a better interpretation. Next attempt:

Interpretation (2): The player whose string is captured may play
a new uncapturable stone under the original string.

- In example (1): O[aBbdbapcpbe..]. Black fails to play a new
uncapturable stone under the original string. So Black's corner
string is dead.
- In example (2): #[app]. White fails to play a new uncapturable
stone under the original string. So White's initial single stone
is dead.

Obviously, interpretation (2) fails in example (1). Therefore
we have to have a better interpretation. That is what I am
looking for.

> Assuming White to move, the best White
> can do is to capture the Black stone at B. Black has to pass, having no
> ko threats, and White can't do anything (since White has no ko threats
> either).

Please note that during J1989 scoring the PASS-FOR-KO-RULES APPLY!

Therefore you have to forget about all you know about ko threats.
Essentially during scoring the only ko threats are a pass for the
particular ko.

> Since "game stop" (or whatever term you wish to use for the time
> just before any dame are filled and the score is counted) doesn't depend
> on consecutive passes under Japanese rules (this isn't computer server
> rules!), White can't get out of it with a second pass.

Get out of what? We are AFTER the "game end" anyway.

> I highly doubt this position would be a problem in a normal 19x19 game,

THIS position would not be a problem because THE OFFICIAL
COMMENTARY LISTS IT AS A PRECEDENT.

However, if we do not understand rules application for also
this shape in general, then we cannot apply the rules to
various other shapes like in example (2).

> since Black could just fill in a dame or empty point someplace in his own
> territory.

See above, this would be interpretation (1). It fails.

> Does anybody actually play the way you do, Robert,

Are we discussing playing styles or whether it is possible to
apply Japanese rules in general?

> and do these positions
> ever come up in your real games (eg. EGF games)?

Example (1) occurs occasionally, shapes like example (2) very
frequently.

> I swear that one of these days, we're going to see you write "It depends
> on what the definition of 'is' is."

For a discussion of "as is" see earlier RGG articles:)

--
robert jasiek

ro...@telus.net

ungelesen,
25.08.2003, 14:03:1225.08.03
an
On Sun, 24 Aug 2003 21:03:57 +0200, Robert Jasiek <jas...@snafu.de>
wrote:

>Rules

Hmmm. What a very odd and misleading and counter-intuitive use of the
word, "string." What language is it from? It seems to refer to what
English speakers not familiar with what you are defining would
normally call a "unit." If you are going to just assign an arbitrary
word to the concept, why not call it a "Jasiek"? At least that would
be clearer than calling a round, solid, lenticular object a "string."

-- Roy L

Stephan Zidowitz

ungelesen,
26.08.2003, 03:41:5626.08.03
an
Leaving aside the text of rules for a moment the intended meaning of
capturable-alive stones is IMO that the WHOLE area occupied by this
stones will eventually become part of a unconditionaly living group.

The problem seems, how to express this 'global' idea of a living group
in local terms of 'capturable/death stones'. Thinking about what can
happen I come up with following (quite similar to what robert stated
earlier):

A union of stones is Capturable-alive, if - after capturing - on ALL
intersections occupied by the Capturable-alive stones finally are
uncapturable stones of the same color OR empty intersections which could
not be occupied by the opponent (due to not allowens of suicide), i.e. a
'living eye'.

Cheers.

Robert Jasiek

ungelesen,
26.08.2003, 04:13:0426.08.03
an

Stephan Zidowitz wrote:
> the intended meaning of
> capturable-alive stones is IMO that the WHOLE area occupied by this
> stones will eventually become part of a unconditionaly living group.

This is OC completely wrong. It is not the intended meaning at all.
E.g., the official example II.4: None (!) of the string's initially
occupied intersections becomes part of an old or new uncapturable
string.

. # . . # # #

# # O O O O O
. O # # # # #

. O # # . # .
# O # # # # #
# O # O O O O
# O # O . O .

***

For territory scoring rules one could define death however one
likes, however, as soon as one regards the particular case of
those territory scoring rules that are Japanese rules one loses
a lot of freedom for otherwise possibly reasonable definitions.

--
robert jasiek

Stephan Zidowitz

ungelesen,
26.08.2003, 05:39:1026.08.03
an
I don't get the point - Sorry not so familar with official examples -
what is the defined outcome ?

Stephan Zidowitz

Robert Jasiek

ungelesen,
26.08.2003, 05:56:2026.08.03
an

Stephan Zidowitz wrote:
> I don't get the point - Sorry not so familar with official examples -
> what is the defined outcome ?

See my preliminary study below.
U = uncapturable-alive
C = capturable-alive

In proof play for and capture of the upper white string a new
uncapturable stone appears on the left side. In proof play for
and capture of the left white string a new uncapturable stone
appears on the upper side.

--
robert jasiek

Example II.4
"a miai position"
7x7
F5
final-position

. # . . # # #
# # O O O O O
. O # # # # #
. O # # . # .
# O # # # # #
# O # O O O O
# O # O . O .

intersections

. # a b c d e


# # O O O O O

f O # # # # #
g O # # . # .
h O # # # # #
i O # O O O O
j O # O . O .


The black string in the upper left is U: O[bdgiaecedepp], etc.
The black string c-d-e is C: As an example, O[bdgiaecedepp] puts a
new uncapturable black stone on e.
Likewise, the black string h-i-j is C.
Black capturing the upper white string leads to a new uncapturable
white stone on i: #[abdgeicpp].

enabling type S: enabling intersections W for the upper white string

W W W W W W W
W W W W W W W
W W W W W W W
W W W W W W W
W W W W W W W
W W W O O O O
W W W O . O .


Presumably, under the Japanese 1989 Rules any of the intersections W
are enabling for White when his upper string is captured.
Since the new uncapturable white ston on i is on an enabling
intersection W, the upper white string is C.
Likewise, the left white string is C.
Score = O.

enabling type T: enabling intersections W for the upper white string

. # W W W W W
# # W W W W W


. O # # # # #
. O # # . # .
# O # # # # #
# O # O O O O
# O # O . O .


Since the new uncapturable white ston on i is not on an enabling
intersection W, the upper white string is dead.
Likewise, the left white string is dead.
Score = 27.

alternate moving before the first game stop with Black to move

. # . . # # #
# # O O O O O
. O # # # # #
. O # # . # .
# O # # # # #
# O # O O O O
# O # O . O .


Under rules with enabling type S: #[abdgeicfpp]. Prisoner-difference
= 0. Board-score = 5. Score = 5.
Under rules with enabling type T: See below.

alternate moving before the first game stop with White to move

. # . . # # #
# # O O O O O
. O # # # # #
. O # # . # .
# O # # # # #
# O # O O O O
# O # O . O .


Under rules with enabling type S: O[pabdgeicfpp].
Prisoner-difference
= 0. Board-score = 5. Score = 5.
Under rules with enabling type T White must encourage Black to
dissolve the shape so that O[pp] does not occur: O[bdgipefj..],
O[bdgiae..]. However, White fails, whatever he tries. So O[pp] is a
best choice for White.
Conclusion:
Under rules with enabling type S, the shape is dissolved before the
final-position.
Under rules with enabling type T, the shape is not dissolved before
the final-position.
Unlike in example II.2, here in example II.4 White wants a new
uncapturable stone not only beyond opposing U strings but also at
the
same time under some former own stone!

Bernd Gramlich

ungelesen,
26.08.2003, 07:11:1426.08.03
an
Stephan Zidowitz wrote:

> A union of stones is Capturable-alive, if - after capturing - on ALL
> intersections occupied by the Capturable-alive stones finally are
> uncapturable stones of the same color OR empty intersections which
> could not be occupied by the opponent (due to not allowens of
> suicide), i.e. a 'living eye'.

This works well with snapbacks, but not with other precedents of
Japanese rules. Take e.g. the one we discussed recently in
<3F3FBD74...@snafu.de>:

+-------------+


|. # O . O # .|

|# . O . O # #|
|O O O O # # .|
+-------------+

The 1989 Nihon Ki'in rules try to unify the behaviour of snapbacks and
many of these precedents with just one "simple" rule. This works quite
well (except for "Three points without capturing" and some shapes with
ko and doubleko seki), but has two major drawbacks:

1. Despite its simple wording, the rule is not easy to understand.
2. Some pathological examples with absurd results exist (e.g. the
one by James Davies where a doubleko seki gives capturable life
to every string on the board with at least two liberties).

Now Robert tries to reword this rule in order to avoid these drawbacks.

Bill Spight

ungelesen,
26.08.2003, 07:53:0626.08.03
an
Dear Stephan,

Robert Jasiek:


>
> . # . . # # #
> # # O O O O O
> . O # # # # #
> . O # # . # .
> # O # # # # #
> # O # O O O O
> # O # O . O .
>

In case this is not clear, it is a 7x7 board.

Robert likes to give full board diagrams of small boards without
identifying them as such.

Best regards,

Bill

Bill Spight

ungelesen,
26.08.2003, 08:24:3526.08.03
an
Dear Stephan,

> > . # . . # # #
> > # # O O O O O
> > . O # # # # #
> > . O # # . # .
> > # O # # # # #
> > # O # O O O O
> > # O # O . O .

[snip]

> I don't get the point - Sorry not so familar with official examples -
> what is the defined outcome ?
>

Under Japanese '89 rules, assuming that play has ended, this is just
what it may look like: jigo with seki.

However, that is debatable. In fact, zero is not the theoretical value
of the position.

I trust that it is fairly obvious that White cannot capture either of
the Black groups of three without dying. Suppose that White captures
both of them and Black prevents 2 eyes, yielding this position:

. # . O . # .

# # O O O O O
. O # # # # #

O O # # . # .

. O # # # # #

# O # O O O O

. O # O . O .

Now, with White to play:

. # 1 O 4 # 2

# # O O O O O

3 O # # # # #
O O # # . # .

. O # # # # #

# O # O O O O

. O # O . O .

White cannot save either group.

OTOH, Black to play can capture one of the White groups of five but not
both.

. # 1 2 # # #

# # O O O O O
. O # # # # #
. O # # . # .
# O # # # # #
# O # O O O O
# O # O . O .

. # # O . 3 .

# # O O O O O
. O # # # # #

4 O # # . # .

# O # # # # #
# O # O O O O
# O # O . O .

. # # O . # .


# # O O O O O

6 O # # # # #
O O # # . # .

. O # # # # #

5 O # O O O O
. O # O . O .

If B5, W6 kills. So,

. # # O . # 5


# # O O O O O
. O # # # # #

O O # # . # .

. O # # # # #

6 O # O O O O
. O # O . O .

W6 lives.

Black can kill the White group of five on the top, but doing so enables
White to play uncapturable stones on the left (W4 and W6). Therefore,
the group of five on the top are alive. Likewise, so are the White
stones on the left.

Under Japanese '89 rules Black should capture one of the White groups
before play stops, or resume the game and do so later. If she leaves the
position as is, it is seki.

Best,

Bill

Ted S.

ungelesen,
26.08.2003, 13:23:0826.08.03
an
Somebody claiming to be Robert Jasiek <jas...@snafu.de> wrote in
news:3F49CA02...@snafu.de:

>>Seriously, I don't get your point.
>
>If this position is on the board during scoring, then one has to
>score it. The Japanese 1989 Rules fail to explain correctly how
>to score this position while explaining at the same time how to
>score
>
>example (2)
>
>. # . O #
># # # # #
>O O O O O
>. O . O .

Well, I don't see why any ruleset would need to make a specific
explanation for this position.

BTW: This thread is titled "Japanese Rules: Enabling Fun". I don't see
how these rules discussions enable fun for anybody. ;-)

Robert Jasiek

ungelesen,
26.08.2003, 14:03:1626.08.03
an

"Ted S." wrote:
> Well, I don't see why any ruleset would need to make a specific
> explanation for this position.

This is one important reason why I search for rules that do not
make extra rulings for any position. The Japanese 1989 Rules,
however, use and need extra rulings for specific (classes of)
positions.

--
robert jasiek

Symeon

ungelesen,
26.08.2003, 15:12:5526.08.03
an
Robert Jasiek <jas...@snafu.de> writes:
> This is one important reason why I search for rules that do not make
> extra rulings for any position. The Japanese 1989 Rules, however,
> use and need extra rulings for specific (classes of) positions.

I thought "precedents" were no longer necessary in this most
enlightened of rulesets. Somebody is confused (probably me).

/Symeon

Roy Schmidt

ungelesen,
26.08.2003, 16:20:4226.08.03
an
"Ted S." <fe...@bestweb.spam> wrote:

> Somebody claiming to be Robert Jasiek <jas...@snafu.de> wrote in
> news:3F49CA02...@snafu.de:
>
> >>Seriously, I don't get your point.
> >
> >If this position is on the board during scoring, then one has to
> >score it. The Japanese 1989 Rules fail to explain correctly how
> >to score this position while explaining at the same time how to
> >score
> >
> >example (2)
> >
> >. # . O #
> ># # # # #
> >O O O O O
> >. O . O .
>
> Well, I don't see why any ruleset would need to make a specific
> explanation for this position.

Without any special exceptions, the isolated white stone would have to
be captured and removed from the board. Using territory scoring, this
would produce jigo. With a special dispensation, the white stone is
removed during scoring without further play, and Black wins.

Using area scoring, there is no need for a specific explanation for
this sort of situation, because the score is the same either way.

Cheers, Roy

--
my reply-to address is gostoned at insightbb dot com
-------------------------------------------------
Roy Schmidt
Part-time Translator for Yutopian
Full-time Professor of Business Computer Systems
Bradley University

Robert Jasiek

ungelesen,
26.08.2003, 17:06:3426.08.03
an

Symeon wrote:
> > The Japanese 1989 Rules, however,
> > use and need extra rulings for specific (classes of) positions.
> I thought "precedents" were no longer necessary in this most
> enlightened of rulesets.

This ruleset makes the pretence to be without precedents, i.e. the
rules text itself does not have a reference to an explicitly
included precedents section like the J1949 and WAGC1979 Rules had.
However, the J1989 Rules are unclear about various parts of its
text so that quite some examples in the official commentary serve
the same purpose as a list of precedents. Some like example I.16
even contradict the rules text so that some intentions of the J1989
authors become a little clearer.

--
robert jasiek

Bill Spight

ungelesen,
26.08.2003, 17:47:0626.08.03
an
Cher Robert,

> However, the J1989 Rules are unclear about various parts of its
> text so that quite some examples in the official commentary serve
> the same purpose as a list of precedents. Some like example I.16
> even contradict the rules text so that some intentions of the J1989
> authors become a little clearer.

Please be clearer. Even if someone wants to look up the example you
mean, it is not clear which one it is. And even if they do find the one
you mean, how can they tell that it contradicts the rules text? Have
some mercy on your readers, please.

Thanks,

Bill

Ted S.

ungelesen,
26.08.2003, 20:57:1726.08.03
an
Somebody claiming to be "Roy Schmidt" <r...@anti-spam.com> wrote in
news:tiP2b.204741$It4....@rwcrnsc51.ops.asp.att.net:

>Without any special exceptions, the isolated white stone would have to
>be captured and removed from the board. Using territory scoring, this
>would produce jigo. With a special dispensation, the white stone is
>removed during scoring without further play, and Black wins.

You mean the two players don't simply agree that the isolated White stone
is dead, and remove it?

I guess it's too bad that I don't really play go, but a similar variant
that requires some common sense rather than guessing the intent of arcane
rules written down decades ago. Then again, this might explain why I'm
still a double-digit kyu player. ;-)

Stephan Zidowitz

ungelesen,
27.08.2003, 02:21:4527.08.03
an
Thx, getting into it :-)

Awkward positions seems to come up, if stones can be captured, but in
doing so one must admit a loss on other parts of the bord - wich would
not occur without the capture. It realy seems to make sense to classify
them as capturable-alive (as well to enforce an outplay of the situation
if on side if a caputure would mean a net win).

Pondering this I come up with something like this:

Stones are capturable-alive, if there is no sequence of plays to capture
them which did not enable the placement of a living stone (or another
capturable-alive stone) of same color at an intersection where the stone
could not be placed as living or capturable-alive stone as first stone
in any sequence of plays starting at the given position.


Leaving one with the problem of proving the non existence of a 'better'
plaing sequence - hence the analysation of all possible combinations...

Stephan Zidowitz

Robert Jasiek

ungelesen,
27.08.2003, 02:53:2927.08.03
an
Dear Bill,

Sorry for being short when I write much.

The standard references for the Japanese 1989 Rules are:

The Go Player's Almanac
http://www-2.cs.cmu.edu/~wjh/go/rules/Japanese.html
http://home.snafu.de/jasiek/rules.html

The Go Player's Almanac is a safer source for the original text
in its English translation because Fred Hansen's webpage contains
typos in the diagrams and even replaces some diagrams by others.
In particular I have had to notice that the example I am
referring to has a different diagram there!

When I refer to examples of the official commentary, then "I.x"
means "Official commentary on the Japanese 1989 Rules, Part I,
titled 'Commentary on the Japanese Rules', example x" and "II.x"
means "Official commentary on the Japanese 1989 Rules, Part II,
titled 'Examples of Confirmation of Life and Death', example x".

Example I.16 is (upper left corner)

. . . . # O . .
. . . O # O . .
# # # # # O . .
O O O O O O . .
. . . . . . . .
. . . . . . . .
. . . . . . . .

This is an example where a position serves the function of a
precedent for the reasons given in the inline attachment below.
It is also an example that contradicts the wording of the
Japanese 1989 Rules.

If that is still not clear enough - ask! :)

Best,
--
robert jasiek

Example
6x3
F1
final-position

# # # # # #

# . # . O #
# # # # # #

intersections

# # # # # #

# a # b c #
# # # # # #

The Japanese 1989 Rules describe: "Empty points surrounded by the
live stones of just one player are called eye points."
From the description it is clear that a is an eye point.
From the description it is already unclear whether b is "surrounded"
by the live stones of just one player. So it is unclear whether b is
an eye point.
From the description it is clear that c is not an eye point since c
is not empty.
The definition should be: "An intersection is an eye point if it is
of a set of intersections that is connected by lines, has no living
stone on one of them, has - if at all - only stones of one colour on
them, is only adjacent to intersections with living stones of one -
possibly the other - colour, and is adjacent to at least one living
stone."
The definition should be like this since otherwise the Japanese 1989
Rules continue with "[...] Eye points surrounded by stones that are
alive but not in seki are called territory [...] each player removes
any opposing dead stones from his territory as is [...]" and
therefore one could not remove the stone from c since the
intersection c is not part of Black's territory because it is not a
black eye point.

Example I.16
<14>
5x3
F1
after the game end

. . . . #
. . . O #

# # # # #

The black string is uncapturable. The white string is dead.
The official commentary says: "[...] the white stone in Black's
territory [...] is dead, so [one removes] it [...]" This shows the
intention of the authors of the Japanese 1989 Rules. However, they
failed to write a rules text that would have fit their intentions.
Strictly by the wording of the rules, the intersection under the
white stone is not territory since it is not an empty intersection.
Since by the wording of the rules dead stones are removed from
territory, one cannot remove the dead white stone that is not on
territory.

Here are notational conventions:

General

Text and Diagrams are shown in a Fixed Width or True Type font.
Often trivial facts are not explained precisely. Only some
move-sequences are verified to be hypothetical-sequences. Terms
related to territory are explained only occasionally since the
difficult part of the rules is the hypothetical-analysis that
determines alive or dead.
To save space, most diagrams are on smaller boards than 19x19.
Positive scores are favourable for Black.
Negative scores are favourable for White. However, in diagrams the
minus is omitted.
For a final-position or another shon starting position of an example
and unless stated otherwise, the prisoner-difference zero is
assumed.
[Positional themes that are neither standard, classic, nor
explicitly
attributed to other persons have been invented by the author.]

Example Headers

mxn = size of a rectangular board
Fn = frequency class n
Fn? = frequency class n in games with strategic mistakes

Diagrams

# = intersection with black stone
O = intersection with white stone
. = empty intersection
A blank space separates horizontally adjacent intersections.
A line break separates adjacent grid lines vertically.
a,b,c... = denoted intersections

Sequences

[] = move-sequence of alternating moves of both players
#[] = Black moves first
O[] = White moves first
x = move on intersection x
p = pass
t = tenuki (possibly on an imagined bigger board) or intersection t
X = ko-pass for the ko that includes the empty intersection x
[xy..] = obvious part omitted
[]* = repeated infinitely
[][] = move-sequence followed by another move-sequence

Misc

[---] U = uncapturable-alive
[---] C = capturable-alive
[---] HS = hypothetical-sequence
[---] <V> = agrees to rules version V

Frequency

Frequency of a shape in actual games is estimated as follows:

F1 = very frequent = ca. 1:1 to 1:10
F2 = frequent = ca. 1:10 to 1:100
F3 = occasional = ca. 1:100 to 1:5,000
F4 = rare = ca. 1:1,000 to 1:100,000
F5 = very rare = ca. 1:20,000 to never

Needless to say, generally the more frequent cases should have a greater
relevance for designing rules. Surprisingly, most Japanese rules
precendents have been about rare shapes. Maybe their study is more
entertaining. However, one must not overlook the behaviour of frequent
cases. This is much easier said than done! Therefore, many different
classes of shapes are studied carefully. Some rare shapes are still
important if they exhibit and test crucial theoretical aspects. So the
reader can expect detailed explanations for some of the difficult cases.

And do not forget to set your browser to read fixed width font and
character set Western, ISO-8859-1 :)


Robert Jasiek

ungelesen,
27.08.2003, 03:14:1827.08.03
an

Stephan Zidowitz wrote:
> Stones are capturable-alive, if there is no sequence of plays to capture
> them which did not enable the placement of a living stone (or another
> capturable-alive stone) of same color at an intersection where the stone
> could not be placed as living or capturable-alive stone as first stone
> in any sequence of plays starting at the given position.

You have understood part of the problem of definition now.

It is valid to try to define capturable-alive recursively like in
a construction like yours:


"capturable-alive, if there is no sequence of plays to capture
them which did not enable the placement of a living stone (or another
capturable-alive stone)"

However, your other recursion
"[...] at an intersection where the stone


could not be placed as living or capturable-alive stone as first stone
in any sequence of plays starting at the given position"

will not work logically because this is not a recursive construction
but rather an attempt to refer to a previous definition of
capturable-alive while defining it. You better construct such a
second part by omitting the word capturable-alive there.

Then to avoid a stone being uncapturable-alive and capturable-alive,
it is safer to define "Stones that are not uncapturable-alive are
capturable-alive [...]"

> Leaving one with the problem of proving the non existence of a 'better'
> plaing sequence - hence the analysation of all possible combinations...

This is the "proud" feature of Japanese style rules:)

--
robert jasiek

Simon Goss

ungelesen,
26.08.2003, 21:13:2226.08.03
an
Robert Jasiek writes

>Some like example I.16 even contradict the rules
>text so that some intentions of the J1989 authors become a little
>clearer.

Uh? By I.16 do you mean diagram 16 of the first official commentary? How
does that contradict the rules text?
--
Simon

Robert Jasiek

ungelesen,
27.08.2003, 06:06:1027.08.03
an

Simon Goss wrote:
> Uh? By I.16 do you mean diagram 16 of the first official commentary? How
> does that contradict the rules text?


Let me cite parts of one of my earlier messages again:

>>
Example
6x3
F1
final-position

# # # # # #

# . # . O #
# # # # # #

intersections

# # # # # #

. . . O #

# # # # #

The black string is uncapturable. The white string is dead.
The official commentary says: "[...] the white stone in Black's
territory [...] is dead, so [one removes] it [...]" This shows the
intention of the authors of the Japanese 1989 Rules. However, they
failed to write a rules text that would have fit their intentions.
Strictly by the wording of the rules, the intersection under the
white stone is not territory since it is not an empty intersection.
Since by the wording of the rules dead stones are removed from
territory, one cannot remove the dead white stone that is not on
territory.
<<


The rules text states:
1. "Empty points surrounded by the life stones of just one
player are called 'eye points'."
2. "Eye points surrounded by stones that are alive but not in
seki are called 'territory'."
3. "[...] each player removes any opposing dead stones from his
territory as is [...]"

Let us apply this to the intersection-point under the single
white stone in example I.16:
1. This point is not an eye point since it is not empty.
2. This point is not territory since it is not an eye point.
3. The dead stone is not removed from this point since this
point is not territory.

The official commentary states:
"[...] the white stone in Black's territory [...] is dead [...]"

Let us verify whether this conforms to the rules text:
- The white stone is dead. Ok.
- The white stone is in Black's territory. Wrong. Application
of the rules text proves the opposite. Therefore the official
commentary contradicts the rules text.


Also compare my 1997 commentary on the J1989 Rules.


--
robert jasiek

Simon Goss

ungelesen,
27.08.2003, 06:41:0527.08.03
an
Robert,

>Let me cite parts of one of my earlier messages again:

<...>

Apologies. I put mine in the out tray last night but didn't connect till
this morning. Now I understand what you're driving at.
--
Simon

Simon Goss

ungelesen,
27.08.2003, 06:47:1827.08.03
an
Ted,

>You mean the two players don't simply agree that the isolated White
>stone is dead, and remove it?

No, Robert doesn't mean that. He's pointing out that the J1989 rules are
badly worded to describe this behaviour.

>I guess it's too bad that I don't really play go, but a similar variant
>that requires some common sense rather than guessing the intent of
>arcane rules written down decades ago. Then again, this might explain
>why I'm still a double-digit kyu player. ;-)

Oh, you do play Go. It's just that J1989 doesn't accurately describe how
you do it ;)
--
Simon

Roy Schmidt

ungelesen,
27.08.2003, 09:33:5927.08.03
an
"Ted S." <fe...@bestweb.spam> wrote:

> Somebody claiming to be "Roy Schmidt" <r...@anti-spam.com> wrote in
> news:tiP2b.204741$It4....@rwcrnsc51.ops.asp.att.net:
>
> >Without any special exceptions, the isolated white stone would have
to
> >be captured and removed from the board. Using territory scoring,
this
> >would produce jigo. With a special dispensation, the white stone
is
> >removed during scoring without further play, and Black wins.
>
> You mean the two players don't simply agree that the isolated White
stone
> is dead, and remove it?

That is normal practice, but the J1989 rules text is unclear about the
status of the white stone. White could well claim that the stone is
not in Black's "territory," and in fact the rules text would support
his claim. I think Robert already posted an explanation of this
problem. You could convert the spot on which the white stone stands
to "territory" by capturing the stone, but...

> I guess it's too bad that I don't really play go, but a similar
variant
> that requires some common sense rather than guessing the intent of
arcane
> rules written down decades ago.

No, you're OK. It's the rules authors that screwed up. As a result,
they had to publish commentaries that dwarf the rules just to explain
what they really meant to say.

Bill Spight

ungelesen,
27.08.2003, 10:27:5927.08.03
an
Dear Robert,

Thanks for the clarification. :-)

> Example I.16
> <14>
> 5x3
> F1
> after the game end
>
> . . . . #
> . . . O #
> # # # # #
>
>
> The black string is uncapturable. The white string is dead.
> The official commentary says: "[...] the white stone in Black's
> territory [...] is dead, so [one removes] it [...]" This shows the
> intention of the authors of the Japanese 1989 Rules. However, they
> failed to write a rules text that would have fit their intentions.
> Strictly by the wording of the rules, the intersection under the
> white stone is not territory since it is not an empty intersection.
> Since by the wording of the rules dead stones are removed from
> territory, one cannot remove the dead white stone that is not on
> territory.
>

If you are going to make such precise analysis, you should consult the
original Japanese. Even the English here gives a clue: "the white stone
*in* Black's territory", not "*on* Black's territory".

Here is the translation:
<<
Article 10. Determining the result

1. After agreement that the game has ended, each player removes any
opposing dead stones from his territory as is, and adds them to his
prisoners.
>>

A more precise translation might be: "from *within* his territory"
rather than "from his territory". The original Japanese says, "ji no
naka", not simply, "ji".

There is, perhaps, some ambiguity. Arguably, the dead white stone is not
*within* Black's territory, but *beside* it. I really do not know what a
Japanese logician would say. :-)

BTW, the "as is" (sono mama) refers to the dead stones, *not* the
territory. A more precise translation might be: "each player removes any
opposing dead stones as such from within his territory".

> Example
> 6x3
> F1
> final-position
>
> # # # # # #
> # . # . O #
> # # # # # #
>
> intersections
>
> # # # # # #
> # a # b c #
> # # # # # #
>
>
> The Japanese 1989 Rules describe: "Empty points surrounded by the
> live stones of just one player are called eye points."
> From the description it is clear that a is an eye point.
> From the description it is already unclear whether b is "surrounded"
> by the live stones of just one player.

It should not be unclear. The live Black stones surround both point 'b'
and 'c'.

I think you are better off questioning this position:

O O O O O O O O O Middle of the board.
O # # # # # # # O Outer stones alive.
O # . O O O . # O
O # # # # # # # O
O O O O O O O O O

The empty points are surrounded by live Black stones. They are adjacent
to live White stones, but not surrounded by them. Yet they are
considered to be dame.

There is an answer to that. Although they are surrounded by live Black
stones, they are also surrounded by a combination of live stones of both
colors, so that they are *not* surrounded by live stones of only one
player. E. g., the point on the left is surrounded by three live Black
stones and one live White stone.

Note that that does not apply to point 'b' in your diagram. It is also
surrounded by a combination of three Black stones and one White stone,
but the White stone is dead.

The Japanese '89 rules are written in ordinary language, with its
attendant imprecision. Some articles seem poorly stated. I agree that
they need revision.

However, if you are going to critique them at the level of the
imprecision of ordinary language, you really should go to the original
Japanese. Translation adds its own imprecision.

Best,

Bill

Robert Jasiek

ungelesen,
27.08.2003, 13:14:0927.08.03
an
Dear Bill,

> the English here gives a clue: "the white stone
> *in* Black's territory", not "*on* Black's territory".

[...]


> A more precise translation might be: "from *within* his territory"
> rather than "from his territory". The original Japanese says, "ji no
> naka", not simply, "ji".
>
> There is, perhaps, some ambiguity. Arguably, the dead white stone is not
> *within* Black's territory, but *beside* it. I really do not know what a
> Japanese logician would say. :-)
>
> BTW, the "as is" (sono mama) refers to the dead stones, *not* the
> territory. A more precise translation might be: "each player removes any
> opposing dead stones as such from within his territory".

Many thanks!

> It should not be unclear. The live Black stones surround both point 'b'
> and 'c'.

Nevertheless, "surround" should be defined by the rules, especially
to distinguish adjacency to both some alive and some dead stones of
the same player.

> O O O O O O O O O Middle of the board.
> O # # # # # # # O Outer stones alive.
> O # . O O O . # O
> O # # # # # # # O
> O O O O O O O O O
>
> The empty points are surrounded by live Black stones. They are adjacent
> to live White stones, but not surrounded by them. Yet they are
> considered to be dame.

Funny! :)

> However, if you are going to critique them at the level of the
> imprecision of ordinary language, you really should go to the original
> Japanese.

Right, if only I could read (not to mention: understand) Japanese:)

***

What would be your precise translation of §9.2 sentence 1?

Best,
--
robert jasiek

Bill Spight

ungelesen,
27.08.2003, 15:43:3627.08.03
an
Cher Robert,

> Nevertheless, "surround" should be defined by the rules, especially
> to distinguish adjacency to both some alive and some dead stones of
> the same player.

I agree. :-)

> What would be your precise translation of §9.2 sentence 1?
>

Let me bow to John Fairbairn's discussion of a couple of years ago:

> Article 9.2
> ----------
>
> JAPANESE
> Dai-ku jou (shuukyoku)
> Taikyoku no teishi go, souhou ga ishi no shikatsu oyobi ji o kakunin shi,
> goui suru koto ni yori taikyoku wa shuuryo suru. Kore o 'shuukyoku' to iu.
>
> STANDARD TRANSLATION
> Article 9 (end of the game)
> 2. After stopping, the game ends through confirmation and agreement by the
> two players about the life and death of stones and territory. This is called
> 'the end of the game'.
>
> NEW PROVISIONAL LITERAL TRANSLATION (jf)
> Article 9 (ending of the game)
> 2. After suspension of the game, both players validate [kakunin] the life
> and death of stones and territory in accordance with what they have agreed,
> and the game terminates. This is referred to as 'ending of the game'.
>

I have just a couple of things I wonder about.

First, maybe "mutual consent" is a good translation for "goui" (GO-EE)
in this context. I'm thinking about cases where the players find a
position after suspending the game where one player could capture some
stones but the other could save them. If he would still win the game,
the second player might consent to considering them to be dead, even if
he did not agree that they were.

Second, maybe there is some grammatical ambiguity. The first clause
("validate" or "verify") seems naturally to go with the second ("agree"
or "consent"), which modifies "koto" (abstract thing). The natural
translation would then be to nominalize both verbs. OTOH, the first
clause might go with the third ("end" or "terminate"). There is also
ambiguity about whether the phrase, "After the suspension of the game",
goes with the first clause or the third. (It does not particularly
matter, except for readability and style.)

Instead of "confirm", I prefer "validate" or "verify". I also like "by
means of" for "ni yori". I hardly think that I can add precision to the
discussion, but I can offer some alternative wordings.

Suggestion 1:

After the suspension of the game, the game ends by means of the
verification and agreement by both players about territory and the life
and death of stones.

Suggestion 2:

After the suspension of the game, both players verify territory and the
life and death of stones, and the game ends by their agreement about
those things.


I think that it is pretty clear that the agreement is not about ending
the game, but about territory and life and death. But, just in case,
here it is. ;-)

Suggestion 3:

After the suspension of the game, both players verify territory and the
life and death of stones, and the game ends by mutual consent.


Anyway, I think that "kakunin" strongly suggests that there is a right
answer and that the players are supposed to find it. OC, these rules
were written for pros. :-)

Best,

Bill

Robert Jasiek

ungelesen,
28.08.2003, 02:07:4328.08.03
an
Dear Bill,

> > NEW PROVISIONAL LITERAL TRANSLATION (jf)
> > Article 9 (ending of the game)
> > 2. After suspension of the game, both players validate [kakunin] the life
> > and death of stones and territory in accordance with what they have agreed,
> > and the game terminates. This is referred to as 'ending of the game'.

> Suggestion 1:


>
> After the suspension of the game, the game ends by means of the
> verification and agreement by both players about territory and the life
> and death of stones.
>
> Suggestion 2:
>
> After the suspension of the game, both players verify territory and the
> life and death of stones, and the game ends by their agreement about
> those things.

> Suggestion 3:


>
> After the suspension of the game, both players verify territory and the
> life and death of stones, and the game ends by mutual consent.

Interesting!

So I think that we should conclude that - given only the rules
text - it is
- ambiguous whether the players are supposed to state and
agree on life and death of each [string] [verbally] and explicitly
between game stop and the game end,
- clear that reaching the game end requires the players to agree
[verbally] to reach it.

Given professional practice, it seems that hardly ever or never
the players do state and agree on life and death of each [string]
[verbally] and explicitly between game stop and the game end.
Instead they would rather agree to reach the game end and then
possibly notice strange consequences of the rules for the position
that then cannot be changed by alternate moves or dame / teire
placing stones any longer.

Just imagine all the professionals doing formally what (similarly)
should be done under Agreement Area Scoring: For each string or
intersection one player points to it and says "dead / alive /
territory / not territory" and his opponent says "agree / disagree".
A professional doing this would feel like being in a ridiculous
position, I guess. Instead normally the professionals seem to
agree on the game end and assume / hope that both have exactly the
same belief about all status questions.

Best,
--
robert jasiek

Bill Spight

ungelesen,
28.08.2003, 02:08:1528.08.03
an
Dear Robert,

The light dawns! I wrote:

> BTW, the "as is" (sono mama) refers to the dead stones, *not* the
> territory. A more precise translation might be: "each player removes any
> opposing dead stones as such from within his territory".

It hit me what "sono mama" is talking about. How stupid of me not to see
it before! (It's a common phrase. Somehow it got away from Davies, too.)
Better translation:

"each player removes any opposing dead stones as they are from within
his territory."

Even better translation:

"each player removes any opposing dead stones from within his territory,
even though they have liberties."

That's the point of the phrase. The current condition (sono mama) of the
dead stones is having one or more liberties. But they are removed,
anyway.

Translation is difficult. ;-)

Best,

Bill

Bill Spight

ungelesen,
28.08.2003, 03:08:1628.08.03
an
Dear Robert,

> Nevertheless, "surround" should be defined by the rules, especially
> to distinguish adjacency to both some alive and some dead stones of
> the same player.
>

"Surround" is the wrong word. It's the traditional and basic idea, OC,
but their strange new formulations of life, death, seki, and anti-seki
require something else. How's this?

Article 8. Territory

Empty points bordered by live stones of only one player are called "eye
points." They are points of territory for that player, each eye point
counting as one point of territory. Other empty points are called
"dame." Stones which are alive but adjacent to dame are said to be "in
seki."

(Note: An empty point does not have to be adjacent to a stone to be
bordered by it. The verb "border" indicates a region of empty points.
Mathematicians may wish to use more primitive notions.)

And this:

Article 10. Determining the result

1. After the game has ended by agreement, each player removes any
opposing dead stones adjacent to his territory, and adds them to his
prisoners. The points occupied by those stones before removal also
become the player's territory.


Best,

Bill

Bill Spight

ungelesen,
28.08.2003, 03:21:1428.08.03
an
More concisely:

Article 8. Territory

Empty points bordered by live stones of only one player are points of
territory for that player. Other empty points are called "dame." Live
stones adjacent to dame are said to be "in seki."

Robert Jasiek

ungelesen,
28.08.2003, 05:46:0528.08.03
an
Dear Bill,

> Article 10. Determining the result
>
> 1. After the game has ended by agreement, each player removes any
> opposing dead stones adjacent to his territory, and adds them to his
> prisoners. The points occupied by those stones before removal also
> become the player's territory.

Is this translation or interpretation?

For my current draft of an interpretation about determination of
the result, see below.

BTW, the current text length of my complete draft, which ignores
tournament rules, is about 5 times that of the Japanese 1989 Rules:)
One reason for that is that the verb "cannot" / "could not" contains
strategy of each player in its meaning and a sufficiently clear
definition of strategy alone consumes almost a page of rules text.

Best,
--
robert jasiek


Determination of the Result


In the final-position, a black-eye-string is an intersection that is
empty and, recursively, any adjacent intersection that is empty or
occupied by a stone of a dead white string.

In the final-position, a white-eye-string is an intersection that is
empty and, recursively, any adjacent intersection that is empty or
occupied by a stone of a dead black string.

In the final-position, an intersection of a black-eye-string is a
black-eye-point if the black-eye-string is adjacent and only adjacent to
intersections with stones of alive black strings.

In the final-position, an intersection of a white-eye-string is a
white-eye-point if the white-eye-string is adjacent and only adjacent to
intersections with stones of alive white strings.

An eye-point is either a black-eye-point or a white-eye-point.

A dame is an empty intersection that is not an eye-point.

A black-region is an intersection with a stone of a black string and,
recursively, any adjacent intersection that is a black-eye-point or has
a stone of a black string on.

A white-region is an intersection with a stone of a white string and,
recursively, any adjacent intersection that is a white-eye-point or has
a stone of a white string on.

A region is either a black-region or a white-region.

A region is in-seki if at least one of its intersections is adjacent to
a dame.

An intersection is black-territory if it is a black-eye-point that is
not part of an in-seki black-region.

An intersection is white-territory if it is a white-eye-point that is
not part of an in-seki white-region.

The score is
a) the number of empty intersections that are black-territory plus
b) twice the number of not empty intersections that are black-territory
minus
c) the number of empty intersections that are white-territory minus
d) twice the number of not empty intersections that are white-territory
plus
e) the prisoner-difference.

The result is
a) a Black win if the final-position is reached and the score is
positive,
b) a White win if the final-position is reached and the score is
negative,
c) a tie if the final-position is reached and the score is zero, or
d) given due to the rules about an exceptional end of the game.

JVT

ungelesen,
28.08.2003, 09:32:0728.08.03
an
Dear Bill,

You wrote:
> Empty points bordered by live stones of only one player are points of
> territory for that player. Other empty points are called "dame." Live
> stones adjacent to dame are said to be "in seki."

As you well know, this definition is wrong; it allows territory to be
bordered by stones alive in seki.

9 |. . . . . . .
8 |. . . . . . .
7 |O O O . . . .
6 |X X O . . . .
5 |. X O . . . .
4 |X X O X X . .
3 |. X X O X . .
2 |X O O O X . .
1 |. O . O X . .
+-------------
A B C D E F G

A3 and A5 are empty points bordered by live stones on only one player, but
they are not territory.

An improved formulation needs to take into account recursive propagation of
seki along chains of false eyes.

The algorithm should be something like this:
1. remove dead stones
2. mark as dame empty points not bordered by stones of only one player
3. mark as alive in seki stones adjacent to dame
4. mark empty points as dame is they are adjacent to seki stones
Repeat 3, 4 until nothing changes any more.

But It looks too much like programming for a rule text. Hard to say this
without 'repeat until' or 'recursive', though.

--
JVT


Bill Spight

ungelesen,
28.08.2003, 10:42:2328.08.03
an
Cher Robert,

> > Article 10. Determining the result
> >
> > 1. After the game has ended by agreement, each player removes any
> > opposing dead stones adjacent to his territory, and adds them to his
> > prisoners. The points occupied by those stones before removal also
> > become the player's territory.
>
> Is this translation or interpretation?
>

Neither. It is a suggestion, and it's wrong. E. g.,

. # O . . O # Top left corner.
# # O O O O # Outer stones alive.
O O O # # # #
# # # #

The corner point would not be considered White territory, because it
is not bordered by live White stones. (My suggestion for Article 8 uses
"bordered" instead of "surrounded".)

Well, the corner point is surrounded by live White stones, so maybe
going back to "surrounded" works. But then how about this?

# O . # O Top left corner.
. # # # O Outer stones alive.


# # O O O
O O O

This is the famous "anti-seki", in which, if left on the board when the
game ends, both the White stone and the Black stones are dead, but are
not removed. However, the empty points are surrounded by live White
stones, just as much as the empty point in the previous diagram. And,
like it, the 2-1 point is bordered by dead Black stones. So why are
those points not White territory? And then why are the Black stones not
removed?


So here is my new suggestion for Article 10.1.

Article 10. Determining the result

1. Dead stones are removable, even though they have liberties, if their
nearest opposing stone in every compass direction is alive. There may be
directions with no opposing stone. After the game has ended by
agreement, each player removes any opposing removable dead stones, and


adds them to his prisoners.

Since the concept of territory no longer matters to the removal of dead
stones, we can let territory be clearly defined only for the final
position after that removal.

New suggestion for Article 8:

Article 8. Territory

Empty points bordered by live stones of only one player are called "eye

points". Other empty points are called "dame". Live stones adjacent to
dame are called "seki stones". The eye points bordered by live stones
that are not seki stones are territory.

Best,

Bill

P. S. Besides being an abomination, anti-seki is a useless concept. How
many pro games have produced anti-seki under the Japanese '89 rules?
None. Don't be ridiculous. Besides, we would have heard if there had
been one.

By definition an anti-seki is a position that should have been played
out before the end of the game. It can only remain on the board, given
witting players, if the winner could afford to lose his dead stones in
the anti-seki and still win. Otherwise, both players lose. So why not
allow somebody's stones be removed in the anti-seki? E. g., Black's
stones in the example. If it matters, Black could reopen the game to
force the stones to be captured.

I have 2 predictions. 1) There will never be a rules dispute in
professional play concerning anti-seki. 2) The concept will disappear in
the next revision of the Japanese professional rules.

Bill Spight

ungelesen,
28.08.2003, 11:06:4328.08.03
an
Cher Jean-Pierre,

JVT wrote:
>
> Dear Bill,
>
> You wrote:
> > Empty points bordered by live stones of only one player are points of
> > territory for that player. Other empty points are called "dame." Live
> > stones adjacent to dame are said to be "in seki."
>
> As you well know, this definition is wrong; it allows territory to be
> bordered by stones alive in seki.

Yes, thank you. Both of my suggestions last night were wrong. I have
corrected them, I believe, in my recent note to Robert.

>
> 9 |. . . . . . .
> 8 |. . . . . . .
> 7 |O O O . . . .
> 6 |X X O . . . .
> 5 |. X O . . . .
> 4 |X X O X X . .
> 3 |. X X O X . .
> 2 |X O O O X . .
> 1 |. O . O X . .
> +-------------
> A B C D E F G
>
> A3 and A5 are empty points bordered by live stones on only one player, but
> they are not territory.
>

My "correction" still identifies A5 as territory, I am afraid. I simply
replaced "having" with "adjacency". But this is a problem for the
Japanese rules, anyway.

A3, A5, and C1 are all eye points. A1 is a dame.

The White group on B1 - D3 and the Black stone on A2 are seki stones, as
they share the dame on A1. I fail to see how the Black group on A6 - C3
"has" the dame on A1.

> An improved formulation needs to take into account recursive propagation of
> seki along chains of false eyes.
>
> The algorithm should be something like this:
> 1. remove dead stones
> 2. mark as dame empty points not bordered by stones of only one player
> 3. mark as alive in seki stones adjacent to dame
> 4. mark empty points as dame is they are adjacent to seki stones
> Repeat 3, 4 until nothing changes any more.
>

Sounds good, but it invalidates the definition of "eye point" in J89.
;-)

> But It looks too much like programming for a rule text. Hard to say this
> without 'repeat until' or 'recursive', though.
>

Not all rule texts are programmable, I'm afraid. ;-)

Merci,

Bill

Simon Goss

ungelesen,
28.08.2003, 11:52:0628.08.03
an
JVT writes

>The algorithm should be something like this:
>1. remove dead stones
>2. mark as dame empty points not bordered by stones of only one player
>3. mark as alive in seki stones adjacent to dame
>4. mark empty points as dame is they are adjacent to seki stones Repeat
>3, 4 until nothing changes any more.

Not quite right, because you can't remove dead stones that are not
inside territory. For example (top left corner)

+-------------
| . O # O # O
| O . # . # O
| # # # # # O


| O O O O O O

The white stone at 4-1 is dead but cannot be removed after the game stop
because it isn't inside black territory. So the algorithm needs to
establish whether or not there is seki before any stones can be removed.
--
Simon

JVT

ungelesen,
29.08.2003, 04:22:3129.08.03
an
"Simon Goss" wrote:
> JVT writes
>
> >The algorithm should be something like this:
> >1. remove dead stones
> >2. mark as dame empty points not bordered by stones of only one player
> >3. mark as alive in seki stones adjacent to dame
> >4. mark empty points as dame is they are adjacent to seki stones.

> >Repeat 3, 4 until nothing changes any more.
>
> Not quite right, because you can't remove dead stones that are not
> inside territory. For example (top left corner)
>
> +-------------
> | . O # O # O
> | O . # . # O
> | # # # # # O
> | O O O O O O
>
> The white stone at 4-1 is dead but cannot be removed after the game stop
> because it isn't inside black territory. So the algorithm needs to
> establish whether or not there is seki before any stones can be removed.
> --
> Simon

Yes. Thank you. I should not have said "remove dead stones", but remove
_temporarily_ (or ignore) dead stones for the purpose of applying the
algorithm. Then put them back in for scoring.

Most SGF editors have a simple flood-fill algorithm that counts as eyes some
points that are not territory. I have seen only one SGF editor that does it
correctly: CGoban 2.

--
JVT


JVT

ungelesen,
29.08.2003, 04:59:4829.08.03
an
"JVT" wrote:
> 4. mark empty points as dame is they are adjacent to seki stones.
Should be:
4. mark empty points as dame >if< they are adjacent to seki stones.


Robert Jasiek

ungelesen,
29.08.2003, 05:46:4729.08.03
an

JVT wrote:
>>> The algorithm should be something like this:
>>> 1. remove dead stones
>>> 2. mark as dame empty points not bordered by stones of only one player
>>> 3. mark as alive in seki stones adjacent to dame

>>> 4. mark empty points as dame is they are adjacent to seki stones

>>> Repeat 3, 4 until nothing changes any more.

>> Yes. Thank you. I should not have said "remove dead stones", but >> remove


>> _temporarily_ (or ignore) dead stones for the purpose of applying the
>> algorithm. Then put them back in for scoring.

> 4. mark empty points as dame >if< they are adjacent to seki stones.

When you will have fixed your algorithm and defined "bordered",
hopefully with respect to the score it will be equivalent to my
definitions, as I have stated them in an earlier message:)

--
robert jasiek

Matti Siivola

ungelesen,
29.08.2003, 08:52:2629.08.03
an
O O O
O # # # O
O # # . # # O
O # . O . # O
O # . O . # O
_ _ _ _ _ _ _ _ _


It is quite a problem to decide what is territory according to
japanese rules
in the this position (stones outside are undefined) inner white two
stones may be in black's territory or black stones in white's
territory or it might be seki.
See http://www.helsinki.fi/~msiivola/tilanne1.html
With japanese rules either sides loses points if he tries to make an
exchange and it is left as seki. With area rules white can make some
profit with capturing after dame have been filled in the game.

Matti Siivola

JVT

ungelesen,
29.08.2003, 09:18:2529.08.03
an
"Robert Jasiek" wrote:
> When you will have fixed your algorithm and defined "bordered",

Nothing left to fix. It's only a 'you get the idea' version. :)

--
JVT


Planar

ungelesen,
30.08.2003, 07:08:1130.08.03
an
In article <bin2a6$apvfb$1...@ID-150905.news.uni-berlin.de>,
"JVT" <j...@volcanomail.com> wrote:

> Most SGF editors have a simple flood-fill algorithm that counts as eyes some
> points that are not territory. I have seen only one SGF editor that does it
> correctly: CGoban 2.

CGoban2 does not do it correctly in some cases. As far as I can tell,
the problem involves bamboo joints.

Always play all the teire (and force your opponent to do it) before
passing. All strong players do it, and it's no coincidence.

--
Planar

JVT

ungelesen,
31.08.2003, 11:43:2131.08.03
an
"Planar" <damien....@inria.fr> wrote:
> CGoban2 does not do it correctly in some cases. As far as I can tell,
> the problem involves bamboo joints.
> Always play all the teire (and force your opponent to do it) before
> passing. All strong players do it, and it's no coincidence.

By all means. After all these moves have been played and dead stones marked,
however, I do believe a simple algorithm as the one I posted can score any
position correctly. I tested some complicated positions such as hane seki
with CGoban 2, and got the correct result.

--
JVT


Chris Schack

ungelesen,
04.09.2003, 11:40:1104.09.03
an
In article <3F4DAA7D...@pacbell.net>,
Bill Spight <Xbsp...@pacbell.net> wrote:

<snip>

>Article 8. Territory
>
>Empty points bordered by live stones of only one player are called "eye
>points." They are points of territory for that player, each eye point
>counting as one point of territory. Other empty points are called
>"dame." Stones which are alive but adjacent to dame are said to be "in
>seki."

This seems to imply that any stone adjacent to dame is in seki. It's
the inability to live without the dame, combined with the inability of
the opponent to fill the dame, which forms seki.

Chris Schack

Bill Spight

ungelesen,
04.09.2003, 18:46:2404.09.03
an
Dear Chris,

> >Article 8. Territory
> >
> >Empty points bordered by live stones of only one player are called "eye
> >points." They are points of territory for that player, each eye point
> >counting as one point of territory. Other empty points are called
> >"dame." Stones which are alive but adjacent to dame are said to be "in
> >seki."
>
> This seems to imply that any stone adjacent to dame is in seki. It's
> the inability to live without the dame, combined with the inability of
> the opponent to fill the dame, which forms seki.
>

Unfortunately, that is not the case. If independently living groups are
accidentally left adjacent to dame, their territory does not count.
(That can be remedied by restarting the game, however.) There was, in
fact, an example in professional play where best play was to leave an
independently living group in seki. It was discussed here on rgg not
long ago.

Best,

Bill

Chris Schack

ungelesen,
10.09.2003, 10:45:3610.09.03
an
In article <3F57C0E9...@pacbell.net>,

That's just weird...

Chris Schack

Allen antworten
Dem Autor antworten
Weiterleiten
0 neue Nachrichten