Question: What do the Japanese rules say about this position? Assume the rest of the board is filled too.

Click Here To Show Diagram Code

[go]$$cB
$$ ----------------------
$$ | O . O . X a . . O . O
$$ | 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]

Under normal play, if neither side has external ko threats, then white is stable here and will own all of the top because if black captures, white has a local ko threat of throwing in at "a".

Under Japanese rules, does white still need to spend an extra defensive move before the game ends? My "instinct as a Go player" says I want the answer to be no (actually, maybe yes, I'm not sure), but my mechanical understanding of Japanese rules says yes.

My justification: if white does not do so, then white's stones in the upper left corner are dead because they can be captured, and black capturing them does not "enable" any new alive white stone to be placed.

Black demonstrates this by capturing here:

Click Here To Show Diagram Code

[go]$$cB
$$ ----------------------
$$ | O . O . X . . . O . O
$$ | 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]

White is prohibited from recapturing without first passing for that ko. If white throws in with it still does not help because ko threats no longer allow one to recapture in a ko. The ko remains, and presumably white is still prohibited from recapturing until passing for it.

Click Here To Show Diagram Code

[go]$$cB
$$ ----------------------
$$ | O . O . X 2 3 . O . O
$$ | O O O X . X O O O O O
$$ | X X X O B O O . O . .
$$ | X . X O O O O O . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]

So passes for ko,

Click Here To Show Diagram Code

[go]$$cB
$$ ----------------------
$$ | O 7 O 5 X b X a O . O
$$ | O O O X 6 X O O O O O
$$ | X X X O X O O . O . .
$$ | X . X O O O O O . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]

And so white's upper left group dies. So white's group would be dead by confirmation. Black's stones in the initial position would also be dead too, so my understanding is that mechanically we should say that this is an "antiseki" unless white spends an extra move (losing a point) before ending play (or perhaps both players lose since an effective move was needed before ending play).

It doesn't seem to me like the "enable a new alive stone" to be played condition helps either. In the process, white can of course finish off a few black stones by following up at b, but specifically black's final capture of the 5 white stones in the upper left did not "enable" white to play a new alive stone at or "b" or "a", white could have played those all anyways. This isn't a "snapbacky" position.

The same way that if the game ended here:

Click Here To Show Diagram Code

[go]$$cB
$$ ----------------------
$$ | O b O a X d X c O . O
$$ | O O O X X X O O O O O
$$ | X X X O e O O . O . .
$$ | X . X O O O O O . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]

Then white's upper left would not be considered alive with black having the ability to play "a" and "b", regardless of the fact that white in the meantime could be playing new alive stones into "c" and "d" and "e". In this second position, of course, precisely because white cannot sustain a claim of the upper left stones as alive-as-it-stands, we might have very shortly have "both players lose" due to realizing that they should have continued play rather than passing.

Is there something I'm missing here, or is this how it would work?

The last is pretty plainly both sides lose if they are unable to agree to resume play. Whoever asks to resume play, their opponent plays first, and this is a big swing. Black to move plays at a, White to move plays at e.

I suspect that the first is also both sides lose, because Black cannot win the ko in any event.

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | O 5 O 3 X 4 2 . O . O
$$ | O O O X W 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]

at

Besides, why an anti-seki is not both lose, I don't know.

The enabling clause is, IMO, ambiguous or at least difficult to understand. Davies did not have an easy time translating it, I expect. The sentence in Japanese does not
invoke human agency directly. It is more like a stone sprouts up on the board or is given birth to. The Nihon Kiin web site's machine translation says, "a stone that can produce a new stone that can not be taken by the opponent even if taken, is called "living stone"."

Anyway, these positions where play really should not have ended can be maddening if presented as though life and death should be determinable in them, rather than saying, play ended when it shouldn't have. The official commentary on the J89 rules would be shorter and less confusing if they didn't do that.

----

Edit: BTW, the little B appears at the bottom of your diagrams because they start off $$cB instead of $$Bc.

Edit2: As far as your instincts as a go player are concerned, I think that both Go Seigen and Honinbo Shusai Meijin would have agreed.

Edit3: FWIW, Black cannot sustain a komaster claim to the ko in the first diagram, but White can.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

It seems B can actually force W to give up that point by starting the sequence before the stop, then passing (which must always lift bans). Also, he has no problems resuming with W first.

IMO the basic principle of modern territory rules is that only clean captures are given for free. Is this a clean capture? Looks similar to an overlooked teire. The other example looks like a normal unsettled position where both sides have a killing/living ("effective") move.

It is not very likely that the game will end here.

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | P 3 P 1 X . X . O . O
$$ | P P P X X X O O O O O
$$ | X X X O 2 O O . O . .
$$ | X . X O O O O O . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]

White's group in the corner is dead, as it can be captured by Black.

Click Here To Show Diagram Code

[go]$$Wc
$$ ----------------------
$$ | O . O 7 Z 5 Z 3 O . O
$$ | O O O Z Z Z O O O O O
$$ | X X X O 1 O O . O . .
$$ | X . X O O O O O . . .
$$ | . X X 2 4 6 . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]

All of Black's stones at the top are dead, as they can be captured by White.

None of the dead White, and Black, groups is completely surrounded by independendly alive stones, so they cannot be taken off the board.

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | O . O . X . X C Q M Q
$$ | O O O X X X Q Q Q Q Q
$$ | X X X Q C Q Q M Q . .
$$ | X . X Q Q Q Q Q . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]

White's triangled group is connected to dame points (circled), so it cannot contain any territory.

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

lightvector, Cassandra,

1) "the Japanese rules" is an ambiguous phrase. You mean "the Japanese 1989 Rules".

2) You do not specify but mean the L+D confirmation phase.

3) A concept of "external ko threats" does not appear in the Japanese 1989 Rules.

4) A concept of "stable" does not appear in the Japanese 1989 Rules.

5) A concept of "ko threat" does not appear in the Japanese 1989 Rules.

6) A concept of "instinct as a Go player" does not appear in the Japanese 1989 Rules.

7) A concept of "I want" does not appear in the Japanese 1989 Rules.

8) I have proved that "enable" in the Japanese 1989 Rules is ambiguous and provided the disambiguation in the Japanese 2003 Rules. Instead, the concepts local-1, local-2, local-3, capturable-1, capturable-2 and capturable-3 are needed.

http://home.snafu.de/jasiek/j2003.html
http://home.snafu.de/jasiek/j1989c.html

9) I have proved that the concept of passing for a particular ko in the Japanese 1989 Rules is wrong and provided the correction by the concept of generic passing for ko in the Japanese 2003 Rules.

10) Removal does not prove death. It only proves "not uncapturable". To distinguish life from death, identify uncapturable, capturable-1, capturable-2 or else dead.

11) Apply the Japanese 2003 Rules to determine the correct statuses for the Japanese 1989 Rules.

12) If, however, you want to argue that the 2003 are not the 1989 Rules, then observe that the latter contradict themselves with respect to application to different examples so are inconclusive whenever their application is ambiguous, as in your examples.

Ah, yes! The infamous stones in seki rule. I took lightvector to be interested in positions with the outside liberties filled, such as this.

Click Here To Show Diagram Code

[go]$$Bc
$$ ---------------------------
$$ | O . O . X . X C Q M Q X .
$$ | O O O X X X Q Q Q Q Q X .
$$ | X X X Q C Q Q M Q X X X .
$$ | X . X Q Q Q Q Q X X . . .
$$ | . X X X X X X X X . . . .
$$ | X X . . . . . . . . . . .
$$ | . . . . . . . . . . . . .
$$ | . . . . . . . . . . . . .[/go]

where the outside is Black territory.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Yes, that was my intent when I said "Assume the rest of the board is filled too." in the original post.

So let's call this position A:

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | O . O . X . . . O . O
$$ | 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]

It seems like there's some agreement, that position A requires white to play an extra move and thereby lose one point, and perhaps black can try some things before ending play to force white to do so if there's any doubt. That seems okay. How about the next two positions?

B:

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | O . O . X . . O . . O
$$ | 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]

C:

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | O . O . X . X O . . O
$$ | 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]

In position B, black cannot force white to have to spend an additional move under normal play (except maybe if there's a rules-beasty thing somewhere else on the board too, suppose there is not). The position is stable under normal play for white without white having to add a move here, even if there are many ko threats for black, because unlike position A, White's throw-in ends up producing a double-ko-death.

However, my understanding of the mechanics of the confirmation phase at the end says that white will still be required to make an extra move here. Because if the game ends here, then in the confirmation phase black can claim white's upper left stones are not alive because with black first to play, black will capture the ko, then when white throws in to create the double ko death, black will capture that too, and white will be in the awkward position of having *both* ko mouths in the double-ko-death prohibited and not-passed-for yet! So black will get a free move and kill white in the corner.

In position C, however, black *already* has an extra stone that fully forms the double-ko-death shape. So now, my understanding is that white does *not* need to add a stone, Black is dead and white is alive as it stands. In the confirmation, if black plays first and captures the ko, white will play and capture the other ko, and then black has no working moves and no time to pass for ko before white captures black.

So, is my understanding correct? (again, assume the rest of the board is filled up and settled).
* Position A - this is stable under normal play for white if there no ko threats on the rest of the board and not even any temperature 0 moves. Otherwise black can fight the ko to try to force white to protect, using even temperature 0 moves as threats.
* Positions B and C - white is stable under normal play without an extra move even if black has any finite positive number of ko threats.
* Positions A and B - Japanese rules would mandate that white needs to add a stone here, costing white 1 point.
* Position C - Japanese rules would say that white does NOT need to add an extra stone here and cost themself 1 point, black is dead and white is alive already.

I think you are right.

Position A: Good point about Black playing a dame to force White to win the ko without regard to rules about hypothetical play.

Position B: I think this is a rules beast for the J89 rules. The fact that in hypothetical play Black can win the double ko death is an anomaly. However, note that in taking the ko and forcing White to throw in, Black gains 1 pt. for the captured ko stone.

In both cases play might stop when White takes the ko, Black has no ko threat, and there are no dame left; so Black passes and then so does White.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Suppose that you mean the Japanese 1989 Rules.

They do not have rules (not even §13.1) mandating, or not mandating, adding an extra stone.

Instead, analyse position A by assessing its statuses, analyse position B (created by playing an extra stone before the first game stop, if you like) by assessing its statuses, and draw the strategic, rules-independent conclusion which position favours a player so that he (not: the rules) prefers a move creating it.

Your problem with rules application is that you a) ask others to apply them instead of applying them by yourself, b) introduce lots of rules-unrelated concepts making it unnecessarily difficult for you to apply the rules and c) presume that the Japanese 1989 Rules would be unamabiguous allowing application without interpreting other rulesets.

A careful rules application / interpretation can take 1/4 to 4 hours per position IMX. I cannot spend that much time now to do the detailed application for you. Naive short applications do not clarify whether they are correct. Correct application has to be careful step by step and object by object. It does not work if one tries to do all analysis steps for several objects simultaneously.

Position B: I would think under Japanese rules white does not have to add a move, because white is safe locally. Black would need another double ko elsewhere to keep fighting in this position, but in the encore, black is not allowed to use ko threats in an unrelated situation elsewhere. If there is another double ko elsewhere, then black may choose to fight it before passing and the game can end up in a quadruple ko, voiding the game.

There is no such prohibition; also during confirmation, play is global. However, plays "elsewhere" are ignored and answered by "local" replies, passes or ko-passes. A player's plays elsewhere have no (noteworthy) impact on the determination of a local status.

Ok, but doesn't that come down to the same thing: a ko threat elsewhere has no effect on the status of the local position, so it's effectively a null-move?

It has "no" impact but is also not a pass. (IIRC, chess uses null-move, go uses pass.)

I used the term null-move on purpose, because I wanted to avoid any side effects that a true pass may have (such as ending the game). But perhaps there really isn't any difference in practice.

So Robert, what would be your verdict on position B? Does white need to add a stone or not, or do we need the whole position to determine that? Or is it just ambiguous under Japanese rules?

I do not guess. For a verdict on B and a decision on strategic reinforcement, I would apply the Japanese 2003 Rules carefully. As I have said, this might take up to 4 hours; time I do not have.

I agree with Bill that B sets a precedent for unexpected rules treatment of perpetual ko. Since the position is outside the rules precedents of Japanese professional player-rules makers, such does, however, not make reinterpretation of other precedents necessary. Rather B is another flag for how not to design rules - different during different phases.

Since the surrounding groups are pass-alive and so uncapturable even if part of a set of strings, the rest of the board is immaterial. This presumes that a play elsewhere to postpone pass or ko-pass and avoid self-atari of either pass-alive string is not needed in the local variations to be analysed.

I do not belong to the "just ambiguous" faction because I have created J2003 and other means of disambiguation and interpretation:)

But my understanding is that even "local" ko threats don't work. The Japanese (89) rules seem to specify that "in the encore" (i.e. during confirmation) in order to recapture a ko, you have to pass for that ko. So it doesn't matter that white is safe locally under normal play, white is NOT safe under altered ko rules where recapturing in any ko requires passing for that ko. Right?

Click Here To Show Diagram Code

[go]$$Bc
$$ ----------------------
$$ | O . O . X 2 3 O . . O
$$ | O O O X T 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]

If this sequence takes place during the confirmation phase, at this point, my understanding would be that white can NOT recapture at the triangled point. White has not passed for that ko, and it doesn't matter that W2/B3 have occurred, nor would it matter if white had any ko threats anywhere else - ko threats do not enable recapturing kos in confirmation. Similarly, white cannot recapture at the new ko shape at W2 because white has not passed for that ko either. So if you mechanically apply what the rules seem to say, white is in trouble.

Of course, I don't know what Jasiek's formal 2003 version would say, but did any Japanese professionals have a hand in crafting the 2003 version? If not, it seems to me not necessarily relevant?

If you read my J89 commentary carefully, J89 is ambiguous also WRT whether the ko is still the same ko because, after White 2, the previous ko at 1 is NOT a ko...! So if you insist on literal J89 application, you can only conclude: ambiguous!

If Japanese professionals had participated in J2003, it would have been as hopelessly ambiguous as J89. Of course, J2003 is my work. After 10 years of preparation in J rules study, then 11 months of full time work (including work while sleeping!). Call such diligence "not relevant", LOL.

The J89 definition of ko is not ambiguous. It is a position with a possible take and take back sequence of play, if there were no rule preventing it. makes the ko at 1 not a ko, but captures and restores its ko status. We can argue that it is not the same ko now, but the J89 rules plainly interpret it as the same ko, formed by stones on the same points with the same take and take back possible play.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Same intersections, ko stone and ban are insufficient to define whether it is still the same ko. You also would not say same ko stone just because it is on the same intersection after capture and recapture; the earlier ko stone is not on the board but in the lid!