GT territory rule

Cassandra · Post by **Cassandra** » Sun Sep 05, 2021 9:09 pm

jann wrote:I think J89 revised the example exactly for this reason: the old version is buggy since the double ko is useless there regardless of rules.

The former version of this example had NOT been buggy at all, as the intended result can be reached without any difficulties.

But they must have revised the text of "Article 7, 7.2" BEFORE!!!

Otherwise, with the apparent initial version, the desired result would not have been achieved!
It is NOT the nested ko shape of Black's, but of White's, DOUBLE-ko group that matters. It would have been more than sufficient to just interchange the kind of DOUBLE-ko group.

jann · Post by **jann** » Sun Sep 05, 2021 10:27 pm

Cassandra wrote:The former version of this example had NOT been buggy at all, as the intended result can be reached without any difficulties.

As I wrote this is exactly the problem. The result is the same regardless of rules, even just resumed playout with basic rules (since in that version the double ko doesn't even provide a single threat - ko pass abuse aside).

But they must have revised the text of "Article 7, 7.2" BEFORE!!!

I doubt the text changed, current 7.2 still seem to match Davies translation as I quoted earlier.

Otherwise, with the apparent initial version, the desired result would not have been achieved!

The ko-pass abuse in double kos is a separate problem that affects several examples. But safe to assume that such is not allowed, one way or another.

(Even if the current text overlooks it, later they may fix that by requiring a normal pass of either player before passing for the same ko a second time, or whatever.)

Cassandra · Post by **Cassandra** » Sun Sep 05, 2021 11:05 pm

jann wrote:
But they must have revised the text of "Article 7, 7.2" BEFORE!!!
I doubt the text changed, current 7.2 still seem to match Davies translation as I quoted earlier.

It does NOT!

"それぞれ" (sorezore) is "each", "every", NOT "particular"!!!
And not only in Google Translate, but also in a "real world" paper dictionary.

Let me explain at bit deeper:
"木" (hon) is "tree".
"林" (hayashi) ==> More than only ONE tree ==> One tree and another tree and another tree ... ==> is "grove", "spinney", "small forest".
"森" (mori) ==> Many (/ a lot of) trees => is "wood", "forest".
(Japanese count multitudes "ichi, ni, takusan" => "one, two, many".)

"これ" (kore) is "this one here". I.e. "nearby" (the speaker).
"それ" (sore) is "that one there". I.e. "farther away" (from the speaker, maybe nearby the dialogue partner).
"あれ" (are) is "that one over there". I.e. "in a distance" (from both dialogue partners).

"それぞれ" (sorezore) is "それそれ" (soresore), with the second "そ" adjusted in the pronunciation, as it follows a vocal (e) in the compound.
==> More than only ONE of "that one there" ==> "that there" and another "that there" and another "that there" ... ==> "each", "every".

++++++++++++++++++++++++++++++++++++++++++++

By the way:
The adjustment of J89 must have happened BEFORE 2007.
I own a Japanese book, published in this year, that contains the CURRENT version of J89.

jann · Post by **jann** » Sun Sep 05, 2021 11:38 pm

Have you ever seen a Japanese text of 7.2 that was different from current one?

Cassandra · Post by **Cassandra** » Mon Sep 06, 2021 3:25 am

jann wrote:Have you ever seen a Japanese text of 7.2 that was different from current one?

Dear jann,

I am afraid that we were disputing about nothing...

Cointidentially, I searched for older Japanese versions of J89 this morning, but failed to find any (there was always the modified L&D example 18 included). Either the memory of the internet does not go back far enough, or there had been no respective contents ever in the Nihon Kiin website. (Or I haven't looked long enough.)

After having failed with looking for the complete J89 text, I searched for pictures of the "critical" L&D examples that contained
-- either a double-ko with attached bent-four,
-- or a double-ko with attached moonshine-life.

After a while, while studying the relevant websites from which the pictures were taken, I noticed something.
-- Either the comments were a simple copy of the J89's comment (e.g. "In the corner, the seven Black stones ..."),
-- or they started with something like "(Admittedly) there is this double-ko on the right, BUT ..."

The "..." was always someting like "... Xxxxx's stones at the left ARE dead, so after the collaps of the (temporary) seki, ..."

That would confirm what you wrote about the traditional Japanese understanding of double-ko in such combined positions.
(Which, by the way, agrees with the result of my analyzes regarding the "INTENDED" result of the status comfirmations.)

It seems to me now that Japanese tradition examines the "attached" formation INDEPENDENTLY (that is my "There in no double-ko during status confirmation.")
This implies that Japanese tradition would NEVER use the double-ko captures as a part of a sequence that frustrates the capture of the attached part of the combined formation forever. This is what you called an "abuse" of the double-ko cycle (and matches with my "A capture into a double-ko must not be used as a ko-threat.")

The remaining problem seems to be (ONLY) that the authors of J89 (apparently) did NOT want this "traditional understanding" becoming an EXPLIXITE part of the rules.
But have come up with some kind of "replacement mechanism". Which, in the understanding of James Davies' translation of the (supposed) initial version of that "replacement mechanism" ALONE (i.e. WITHOUT any additional knowledge of that Japanese tradition) did not really work.

Gérard TAILLE · Post by **Gérard TAILLE** » Mon Sep 06, 2021 6:41 am

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . O . X O X O . . . . $$ | O X X X O X O O . . . $$ | . X O O O X X O O . . $$ | X X O O . O X X O , . $$ | O O O . O X . X O . . $$ | X X O O X . X X O . . $$ | . X X O O X X O O . . $$ | . . X X X O O O . . . $$ | . . . . . . . . . . . $$ | . . . , . . . . . , .[/go]

This position is very interesting indeed and ask a lot of questions concerning the logic of japanese go.

I do not know what could be the expected result and I guess very few people can give this expected result.
When reading the exchanges between Thomas and Jann I discover that this position was the original position for example 18 but this position has been replaced by a simpler one. Why this replacement? Two possible answers
1) this position corresponds to a buggy analyse
2) No agreement for the expected result has been reached

I have not the answer but I perfectly know what guided me when building the "GT territory rule".
First of all I have been very influenced by Jann post viewtopic.php?p=266694#p266694.
Unless you have a good reason to do differently you must trust "normal play" to decide the status of the different groups of stones.

In area oriented rules the things are simple : you use normal play till the end of the game.
In territory oriented rules it is not as simple : normal play stops after two passes but it is not the end of the game. The confirmation phase begins and the particularity of territory oriented rules is that normal play cannot simply be used in this confirmation phase.

My own analyse tells me that it exists only two cases in which other considerations are used instead of simple normal play.

1) The most important case is to recognise that regions on the board may be considered independant (remember how Thomas drawed separations on the board by means of two-eye alive groups). This point is absolutly crucial in the logic of territory rules. The two basic examples are the following:

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . O . X O . . . . . . $$ | O X X X O . . . . . . $$ | . X O O O . . . . . . $$ | X X O . . . . . . . . $$ | O O O . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | X X X X X X X . . . . $$ | O O O O O O X . . . . $$ | . X X X . O X . . . . $$ +----------------------[/go]

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . X . X O . . . . . . $$ | X X X O O . . . . . . $$ | O O O O . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | X X X . . . . . . . . $$ | O O X X X X . . . . . $$ | . O O O O X . . . . . $$ | O X O . O X . . $$ | X X X O O X . . . . $$ | . X . X O X . . . . $$ +----------------------[/go]

This independancy is handled by J89 by the wording "enable". Yes OC "enable" is not well defined and it addresses basically you intuition as a go player. But you cannot ignore the idea of independancy.
Remember Jann's post viewtopic.php?p=266962#p266962. "enable" only allows to resolve such positions, without the help of pass-for-ko.

2) the second case is to allow breaking loops. When using only normal play loops cannot be breaked because no superko is used. Pass-for-ko is the tool chosen by J89 to break these loops and this tool is quite efficient in examples 8, 16, 17, 18 (I do not consider example 11 which is reolved by "enable" tool).

In all other situations my view is to trust normal play for finding the status of the groups (see Jann post viewtopic.php?p=266694#p266694 I mentionned above)

With this in mind I consider that the pass-for-ko may be used abusively, I mean in case with no loop where normal play can be used very easily.

Let's take an example to illustrate independancy:

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . O . X W . . . . . . $$ | O X X X W . . . . . . $$ | . X W W W . . . . . . $$ | X X W . . . . . . . . $$ | W W W . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | X X X X X X X . . . . $$ | O O O O O O X . . . . $$ | . X X X . O X . . . . $$ +----------------------[/go]

Is the upper left corner independant of the other parts of the board? As it is drawn it is not independant because the status of white marked stone is not clear.

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . O . X W . W . W . . $$ | O X X X W W W W W . . $$ | . X W W W . . . . . . $$ | X X W . . . . . . . . $$ | W W W . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | X X X X X X X . . . . $$ | O O O O O O X . . . . $$ | . X X X . O X . . . . $$ +----------------------[/go]

In this above position the upper left corner can really be considered independant.

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . O . X W . O X . . . $$ | O X X X W X O X . . . $$ | . X W W W X O X . . . $$ | X X W O . X O X . . . $$ | W W W O O O O X . . . $$ | X X X X X X X X . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ +----------------------[/go]

Clearly in this other position there are no independancy. The best proof of dependency is the following. Depending of the status of seven black stones in the corner, the three black stones at the top may be either dead or alive.

Now let's show you a new position in which the pass-for-ko is used (abusively) in a no loop situation:

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | . O . X W . O X . . . $$ | O X X X W X O X . . . $$ | . X W W W X O X . . . $$ | X X W O . X O X . . . $$ | W W W X O O O X . . . $$ | X X X X X X X X . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ +----------------------[/go]

Strict J89 application says that the two black groups are dead. Isn't it an unexpected result which goes against the go logic?

jann · Post by **jann** » Mon Sep 06, 2021 7:46 am

Cassandra wrote:This implies that Japanese tradition would NEVER use the double-ko captures as a part of a sequence that frustrates the capture of the attached part of the combined formation forever. This is what you called an "abuse" of the double-ko cycle

I'm not sure were are on the same page. I never talked about double ko abuse, only "pass-for-ko abuse" in double kos. By that I meant passing for a ko in a double ko (which forces the opponent to pass for the other to be safe), then take the ko (opponent answers), then pass for the other ko (forcing again, etc.) perpetually.

This is a longstanding question in J89 which has two possible answers:

Maybe passing for each ko is only required once for each player (see here)
Or the problem was overlooked by J89 authors, and will be fixed in the future (with some hack like requiring a normal pass of either player before passing for a ko a second time)

In both cases J89 tries to treat double ko sequences logically, by actual confirmation play, showing several examples and sequences with remote double kos. And - until anything supporting the opposite - I still think the Japanese text for 7.2 is still the same as what Davies (a bit freely) translated. My weak Japanese tells me it says something like "whichever ko recapture (he) passed for, THAT KO can be recaptured afterwards".

Cassandra · Post by **Cassandra** » Mon Sep 06, 2021 7:54 am

Gérard TAILLE wrote:
$$B
$$ +----------------------
$$ | . O . X O X O . . . .
$$ | O X X X O X O O . . .
$$ | . X O O O X X O O . .
$$ | X X O O . O X X O , .
$$ | O O O . O X . X O . .
$$ | X X O O X . X X O . .
$$ | . X X O O X X O O . .
$$ | . . X X X O O O . . .
$$ | . . . . . . . . . . .
$$ | . . . , . . . . . , .
Click Here To Show Diagram Code
[go]$$B $$ +---------------------- $$ | . O . X O X O . . . . $$ | O X X X O X O O . . . $$ | . X O O O X X O O . . $$ | X X O O . O X X O , . $$ | O O O . O X . X O . . $$ | X X O O X . X X O . . $$ | . X X O O X X O O . . $$ | . . X X X O O O . . . $$ | . . . . . . . . . . . $$ | . . . , . . . . . , .[/go]
This position is very interesting indeed and ask a lot of questions concerning the logic of japanese go.

I do not know what could be the expected result and I guess very few people can give this expected result.
When reading the exchanges between Thomas and Jann I discover that this position was the original position for example 18 but this position has been replaced by a simpler one. Why this replacement? Two possible answers
1) this position corresponds to a buggy analyse
2) No agreement for the expected result has been reached

The one who can read is in advantage!

The position above IS the CURRENT L&D example 18, HAS BEEN L&D example 18 since at least 2007, and apparently replaced a former, and simpler, position that had been L&D example 18 at the time of James Davies' translation of J89 into English.

J89 has White's groups in the corner "alive", and Black's groups in the corner "dead".

Let me know if YOU are unable to reproduce this result using your GT territory rules. Then I can tell you the solution (returned from several walks with Honte until now

).

jann · Post by **jann** » Mon Sep 06, 2021 8:24 am

Gérard TAILLE wrote:When reading the exchanges between Thomas and Jann I discover that this position was the original position for example 18 but this position has been replaced by a simpler one.

No the opposite. The simpler one was the original, which is defective since the double ko is useless even with just normal play.

Unless you have a good reason to do differently you must trust "normal play" to decide the status of the different groups of stones.

+1 so far.

My own analyse tells me that it exists only two cases in which other considerations are used instead of simple normal play.

1) The most important case is to recognise that regions on the board may be considered independant
...
This independancy is handled by J89 by the wording "enable".

IMO the opposite again: "pass-for-ko" is the one that is supposed to create ko independence, "enable" is there to RECREATE some dependence (which is required by cases where capture is only possible at the cost of giving up something).

2) the second case is to allow breaking loops. When using only normal play loops cannot be breaked because no superko is used. Pass-for-ko is the tool chosen by J89 to break these loops

I think I see where the problem is in your logic. Remember I talked about some redundancy in your rules?

Essentially, you use two ideas: Korean-style localization, and a new rule vs moonshine loops. But the synergy is poor (or nonexistent) between these two, they are contradicting / redundant.

Korean localization (consider remote groups independently) is a historical answer for the moonshine problem. It is logically complete, and (even if not my preference) it works ok. The side effect of this approach is combined moonshine cases can live.

Japanese rules do NOT use such localization, J89 went for the approach of separating everything at the ko level instead (and consider global sequences, watching "enable"). This is another acceptable approach.

Now your rules: If you use Korean localization, why not accept combined moonshine? If you have a new, general rule for moonshine, why do you want Korean localization for? Without moonshine there is no good reason to consider L/D locally, normal play doesn't work like that. You can just use "enable" with your ko rule. OC, the result won't match J89 - but will be a logical consequence of having a (often better) territory ruleset that handles moonshine correctly without localization.

Your rules do not seem to give the expected (Japanese-like) result in the revised J89 example in question (bent4 + combined double ko) as that is a case where your two patches can not complement each other (like they do in combined moonshine cases).

Cassandra · Post by **Cassandra** » Mon Sep 06, 2021 8:41 am

Gérard TAILLE wrote:
$$B
$$ +----------------------
$$ | . O . X W . O X . . .
$$ | O X X X W X O X . . .
$$ | . X W W W X O X . . .
$$ | X X W O . X O X . . .
$$ | W W W X O O O X . . .
$$ | X X X X X X X X . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .
$$ +----------------------
Click Here To Show Diagram Code
[go]$$B $$ +---------------------- $$ | . O . X W . O X . . . $$ | O X X X W X O X . . . $$ | . X W W W X O X . . . $$ | X X W O . X O X . . . $$ | W W W X O O O X . . . $$ | X X X X X X X X . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . . $$ +----------------------[/go]
Strict J89 application says that the two black groups are dead. Isn't it an unexpected result which goes against the go logic?

Quite apparently, traditional Japanese understanding includes the general principle that there are NO ko-threats in a (temporary) seki during status confirmation.
Which also includes double-ko formations as a special case dealt with in Examples 16-18.

See L&D example 7-2: https://www.nihonkiin.or.jp/match/kiyak ... -07-2.html

Please note that the comment declares White's seven stones in the bent-four as "dead" ONLY.

Gérard TAILLE · Post by **Gérard TAILLE** » Mon Sep 06, 2021 10:06 am

jann wrote:
Gérard TAILLE wrote:When reading the exchanges between Thomas and Jann I discover that this position was the original position for example 18 but this position has been replaced by a simpler one.
No the opposite. The simpler one was the original, which is defective since the double ko is useless even with just normal play.

Unless you have a good reason to do differently you must trust "normal play" to decide the status of the different groups of stones.
+1 so far.

My own analyse tells me that it exists only two cases in which other considerations are used instead of simple normal play.

1) The most important case is to recognise that regions on the board may be considered independant
...
This independancy is handled by J89 by the wording "enable".
IMO the opposite again: "pass-for-ko" is the one that is supposed to create ko independence, "enable" is there to RECREATE some dependence (which is required by cases where capture is only possible at the cost of giving up something).

2) the second case is to allow breaking loops. When using only normal play loops cannot be breaked because no superko is used. Pass-for-ko is the tool chosen by J89 to break these loops
I think I see where the problem is in your logic. Remember I talked about some redundancy in your rules?

Essentially, you use two ideas: Korean-style localization, and a new rule vs moonshine loops. But the synergy is poor (or nonexistent) between these two, they are contradicting / redundant.

Korean localization (consider remote groups independently) is a historical answer for the moonshine problem. It is logically complete, and (even if not my preference) it works ok. The side effect of this approach is combined moonshine cases can live.

Japanese rules do NOT use such localization, J89 went for the approach of separating everything at the ko level instead (and consider global sequences, watching "enable"). This is another acceptable approach.

Now your rules: If you use Korean localization, why not accept combined moonshine? If you have a new, general rule for moonshine, why do you want Korean localization for? Without moonshine there is no good reason to consider L/D locally, normal play doesn't work like that. You can just use "enable" with your ko rule. OC, the result won't match J89 - but will be a logical consequence of having a (often better) territory ruleset that handles moonshine correctly without localization.

Your rules do not seem to give the expected (Japanese-like) result in the revised J89 example in question (bent4 + combined double ko) as that is a case where your two patches can not complement each other (like they do in combined moonshine cases).

Interesting Jann. It remains unclear for me if you consider that pass-for-ko may have bad effect in some situations like the following already identified. IOW are the results according to J89 text the expected results or are they unexpected results?
BTW, in the following example J89 and J2003 give the same results.

Click Here To Show Diagram Code: [go]$$B $$ ---------------------- $$ | O . O . X . . O O . O $$ | O O O X O X O O O O O $$ | X X X O . O O . O . . $$ | X . X O O O O O . . . $$ | . X X . . . . . . . . $$ | X X . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

Click Here To Show Diagram Code: [go]$$B $$ ----------------- $$ | O X . X . O . | $$ | . O X X X O O | $$ | O . O O X X O | $$ | O O O . O X O | $$ | X X O O X X O | $$ | . X X O O O . | $$ | X . X X X O O | $$ -----------------[/go]

jann · Post by **jann** » Mon Sep 06, 2021 10:31 am

Gérard TAILLE wrote:It remains unclear for me if you consider that pass-for-ko may have bad effect in some situations like the following already identified.

IMO these are unexpected side effects of the pass-for-ko idea (there were also a third one decades ago). Lightvector's example (the first above) is particularly bad - may or may not be considered borderline tolerable.

This also shows how dangerous it is to invent new rules for things that are not supposed to be there. If you change ko play OC many outcomes will change, and not always as desired. And again: there is no good reason for local views to begin with (besides moonshine) - normal play doesn't work like that.

Cassandra · Post by **Cassandra** » Mon Sep 06, 2021 10:56 am

jann wrote:Your rules do not seem to give the expected (Japanese-like) result in the revised J89 example in question (bent4 + combined double ko) as that is a case where your two patches can not complement each other (like they do in combined moonshine cases).

OK, I will provide you with the "solution" earlier than intended, but which Gérard will not like at all, as it will be proven that GT City would have better opened its gates to dogs, especially Labradors like Honte with his "magical DOUBLE-ko spell".

++++++++++++++++++++++++++

Click Here To Show Diagram Code: [go]$$W $$ +---------------------- $$ | . O . X O X O . . . . $$ | O X X X O X O O . . . $$ | . X O O O X X O O . . $$ | X X O O . O X X O , . $$ | O O O . O X . X O . . $$ | X X O O X . X X O . . $$ | . X X O O X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

Black claims "controlled area" in the entire upper left corner.
Black is the "defender"; White is the "attacker", moving first.

Click Here To Show Diagram Code: [go]$$W :b4: pass $$ +---------------------- $$ | 1 O . X O X O . . . . $$ | O X X X O X O O . . . $$ | . X O O O X X O O . . $$ | X X O O 2 O X X O , . $$ | O O O . O X . X O . . $$ | X X O O X 3 X X O . . $$ | . X X O O X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

The defender's pass with

creates a "critical position", as the attacker captured a ko with

just before. Black claims for a "permanently prohibited ko", i.e. White's capture into the lower ko-shape of the double-ko is prohibited next time.

Please note that this will make the double-ko disappear AFTER the next ko-capture in it (Honte is lurking around the corner...).

Click Here To Show Diagram Code: [go]$$W $$ +---------------------- $$ | O O 5 X O X O . . . . $$ | O X X X O X O O . . . $$ | 6 X O O O X X O O . . $$ | X X O O X . X X O , . $$ | O O O . O X . X O . . $$ | X X O O . O X X O . . $$ | . X X O O X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

Click Here To Show Diagram Code: [go]$$W :b10: pass $$ +---------------------- $$ | . 7 . X O X O . . . . $$ | . X X X O X O O . . . $$ | X X O O O X X O O . . $$ | X X O O X 9 X X O , . $$ | O O O . O X . X O . . $$ | X X O O 8 O X X O . . $$ | . X X O O X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

The defender's pass with

creates a "critical position", as the attacker captured a ko with

just before. Claiming for a "permanently prohibited ko" is irrelevant here.

Click Here To Show Diagram Code: [go]$$Wm11 :w13: pass $$ +---------------------- $$ | 2 O . X O X O . . . . $$ | 1 X X X O X O O . . . $$ | X X O O O X X O O . . $$ | X X O O . O X X O , . $$ | O O O . O X . X O . . $$ | X X O O X x X X O . . $$ | . X X O O X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

gives atari in the corner.
Playing at

, instead, will not lead to success either.

can only pass, as "x" is a "permanent prohibited ko" for her, as attacker.

Click Here To Show Diagram Code: [go]$$Wm11 $$ +---------------------- $$ | X O a X O X O . . . . $$ | 5 X X X O X O O . . . $$ | X X O O O X X O O . . $$ | X X O O 4 O X X O , . $$ | O O O 6 O X . X O . . $$ | X X O O X x X X O . . $$ | . X X O O X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

captures into the double-ko here, also giving atari at White's group therein.
As a matter of course, he could also succeed by capturing at "a" in the corner first.

can only re-capture in the corner, as a move at "x" is prohibited for her.

captures White's double-ko group.

Click Here To Show Diagram Code: [go]$$Wm11 $$ +---------------------- $$ | . O . X . X O . . . . $$ | O X X X . X O O . . . $$ | X X . . . X X O O . . $$ | X X . . . . X X O , . $$ | . . . . . X . X O . . $$ | X X . . X . X X O . . $$ | . X X . . X X O O . . $$ | X . X X X O O O . . . $$ | X X X O O O . . . . . $$ | O O O O . . . . . , . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

The claimed "controlled area" is Black territory!!!
You will easily realise that this does NOT match J89's expectations!

++++++++++++++++++++++++++

What are the reasons for this striking result?

Well, you will have realised that the DEFENDER (sufficient liberties of their attachment to the double-ko provided) will be able to kill the attacker's double-ko group.

L&D Example 18 contains a sandwich with FOUR layers (i.e. even). The DEFENDER owns the OUTER double-ko shape (but which is the one expected to die).

The ATTACKER's layer at the other end of the sandwhich (here in the corner) is NOT CONNECTED to the OUTSIDE. This implies that BOTH layers of the ATTACKER will die in the process.

What is the difference to L&D Example 16 (double-ko with attached moonshine-life).

The attacker's double-ko group cannot be killed, due to the liberty-count of the attachment.
(In the case of nested ko-shapes for the inner double-ko group, the defender's double-ko group could survive. However, this would not help her, as she would be unable to find any suitable "controlled area".)

L&D Example 16 contains a sandwich with THREE layers (i.e. odd).

The DEFENDER's layer at the other end of the sandwich is CONNECTED to the OUTSIDE (i.e. Black's encirclement). The DEFENDER will be unable to prevent the ENTIRE attachment from becoming captured. Thus, there will be NO territory for the DEFENDER in any claimed "controlled area".
(This would be true even in the case she captured the attacker's double-ko group.)

++++++++++++++++++++++++++

By the way:
The same principle would apply here, in a corner position WITHOUT ko, which is probably easier to understand.

Click Here To Show Diagram Code: [go]$$B $$ +---------------------- $$ | O . X O O O X X O . . $$ | O O X O O . O X O . . $$ | . . X O . O X X O . . $$ | X X X O O X . X O , . $$ | O O O O X . X X O . . $$ | X X X O O X X O O . . $$ | . X . X X O O O . . . $$ | X X X X O O . . . . . $$ | O O O O O . . . . . . $$ | . . . , . . . . . , . $$ | . . . . . . . . . . .[/go]

As a matter of course, White could also claim the corner being her "controlled area", and succeed.
White has 19 stones in the corner, Black has 18 stones. Thus, the corner is worth a surplus of two points of territory for Black.

Gérard TAILLE · Post by **Gérard TAILLE** » Mon Sep 06, 2021 11:05 am

jann wrote:
Gérard TAILLE wrote:It remains unclear for me if you consider that pass-for-ko may have bad effect in some situations like the following already identified.
IMO these are unexpected side effects of the pass-for-ko idea (there also were a third one from decades ago). Lightvector's example (the first above) is particularly bad - may or may not be considered borderline tolerable.

This also shows how dangerous it is to invent new rules for things that are not supposed to be there - if you change ko play, OC many outcomes will change, and not always as desired. And again: there is no good reason for local views to begin with (besides moonshine).

I agree with you Jann. Let me try a deeper analysis.
First of all you said in your previous post:
"pass-for-ko" is the one that is supposed to create ko independence, "enable" is there to RECREATE some dependence (which is required by cases where capture is only possible at the cost of giving up something).
That's a very good point Jann. We all know that any apparently independant regions may be linked by a ko fight. Using a pass-for-ko to avoid this influence between regions is good point. Agreed Jann.

What about the unexpected side effects of the pass-for-ko?

Click Here To Show Diagram Code: [go]$$B $$ ---------------------- $$ | O . O . X 2 3 O O . O $$ | O O O X O X O O O O O $$ | X X X O 1 O O . O . . $$ | X . X O O O O O . . . $$ | . X X . . . . . . . . $$ | X X . . . . . . . . . $$ | . . . . . . . . . . . $$ | . . . . . . . . . . .[/go]

In normal play when Black takes the ko by

, white find a local ko threat at

, black should answer by

and white is able to retake the ko at

.
In confirmation phase it is not allowed to use a ko threat in order to retake a ko. Fine but the point is that the ko threat is here purely local (it is not a ko threat in completly different area on the board). Does that mean that local ko threat may be acceptable in a local ko fight?

Click Here To Show Diagram Code: [go]$$B $$ ----------------- $$ | O X . X . O . | $$ | 1 O X X X O O | $$ | O 2 O O X X O | $$ | O O O 3 O X O | $$ | X X O O X X O | $$ | . X X O O O . | $$ | X . X X X O O | $$ -----------------[/go]

In normal play Black takes the ko by one, white answers locally by

and on

white is able to retake the ko at

In confirmation phase it is not allowed to use a play on the board (to put a stone on the board) in order to retake a ko. Here again the point is that the move (here

) is purely local and we may wonder if it is acceptable that local move can be use in ko fight to retake a ko.

What is your view?

jann · Post by **jann** » Mon Sep 06, 2021 12:39 pm

As I wrote earlier, I think you should think in terms of logical concepts.

For the Korean approach (regions considered separately), it is natural that local ko fights work the same way as in normal play (hence no problems here but there are combined moonshine cases).

For the Japanese approach these examples are again natural consequences. If ko fight is not allowed at all, even combined moonshine is dead, but that - in some rare cases - may also help the dying side. There is nothing wrong with that (1 pt diff for extra reinforcing move) - IF you believe the original theory/idea (absolute ko independence) was correct.

Which IMO is not really the case: there is nothing in go that suggests ko-independent L/D with no threats, no reason besides moonshine (and tradition afterwards). So this unnecessary difference to normal play shows me the Japanese concept itself is slightly (tolerably) wrong. Same with the Korean concept with bent4 + remote seki for example.

The reason to look at normal playout is both to stay close to Chinese/area, and also not to give any undeserved gifts to either player. You must have a tool for free captures from territory (for correct scoring), but there is no need for differences beyond the saved cost of pure cleanup moves. If you have a rule vs moonshine kos, "enable" without any locality seems the most correct.

Life In 19x19

GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule

Re: GT territory rule