Odd Cases 🤔 in the Japanese Rules

We are basically in agreement then, and I think we have explored all the point of views that there are about these positions. The question at the crux of the matter is whether E18 and/or F19 count as enabled by the capture, or not, and depending on that, the life determination will follow :slight_smile: This discussion has seen different criteria reaching different conclusions. In fact a few weeks ago my own understanding and personal criterion was just like yours, that the enable rule works here and these do not need defense (if infinite ko threats are on the board), but I have “changed sides” so to speak :stuck_out_tongue:

As you mentioned before, we won’t really get a formal precise definition of what counts as “enabled”, which is a pretty vague term. We can instead get the different arguments and find different interesting situations to make an idea of what options we have, as we have done.

If the goal is to understand what the Japanese pros and referees would do, then we can look for an actual professional precedent, which might be even harder to find because regarding these direct kos, any match that might serve as precedent but where there is not at the same time an infinite amount of ko threats can probably be waived as “meh, actually they just solved it based on the principle of resumption, without having to care for the enable rule”.

It would be a different world if the Japan professional go association had a public database of “precedents”, with official rulings for particular positions, and this was updated with answers to questions by experts, instead of creating new precedents only every once in a blue moon, when a hard situation comes up in a real game. But that would be a huge extra work that is of very little value to the Japan association, and since rules ambiguities are really really really rare, the pros are fine in practice with the precedent-based rulings and just accepting that maybe once every 50 years or so an extremely weird rule monster will appear in an actual pro game and a new precedent will be established :slight_smile:


There seems to be a lot of misunderstandings about the Japanese Rules in so-called “J2003” and on the L19x19 forum and I think most of it is a result of the English Translation. When looking at the Japanese Rules, you can see where the same terms are used again and again, building up an understanding of that game term. The misunderstandings of the Japanese Rules are also caused by a misunderstanding of game design to a lesser extent.

For the source of my understanding, see the preamble to the Japanese Rules (日本囲碁規約 序 | 棋戦 | 囲碁の日本棋院) the Japanese Rules explanations (Ⅱ日本囲碁規約(ルール)逐条解説 | 棋戦 | 囲碁の日本棋院), the Life and Death examples (Ⅲ 死活確認例 | 棋戦 | 囲碁の日本棋院), and the Summary of the revisions (Ⅳ 日本囲碁規約改定の概要(ご参考) | 棋戦 | 囲碁の日本棋院).

  1. Misunderstanding 1: Uncapturable stones played to show life in Life & Death Confirmation (after the game is stopped) must be played after the initial capture of stones. This is a misunderstanding because there is no such requirement in Ⅱ日本囲碁規約(ルール)逐条解説 | 棋戦 | 囲碁の日本棋院. The stones under consideration can have new uncapturable stones played before some other portion of those stones are captured. And if there was any question, L&D example 4 shows a situation where the uncapturable stone is played before any portion of the stones are captured. Ⅲ 死活確認例 | 棋戦 | 囲碁の日本棋院.

  2. Misunderstanding 2: So-called “hypothetical play” (with passing before retaking a ko) is actually used to determine Life & Death. People can find situations where L&D confirmation with ko passing does not achieve the “correct” result, either because the player could create a never ending loop (which is not how L&D confirmation even works) or by pretending that the moves in so called “hypothetical play” would actually lead to capturing. This misunderstanding arises from the fact that the English translation does not include the Preamble or the Summary of the Revision. The preamble clearly states that the entire point of the revision is to find a rationale for the Japanese way of playing Go. The Committee found a rationale (the pass before retaking a ko) and applied this rationale to various precedential rulings, dismissing the inconsistent rulings (like torazu sanmoku or “3 points without capturing”) and keeping the consistent rulings. The “hypothetical play” in the L&D examples of the Japanese Rules is not showing a procedure to be followed by the players, it is showing how the rationale is consistent. The players will instead rely on the rulings as they are: a group with bent-4 is dead, even with other potential ko threats, etc. The L&D examples build up a series of definitions and once those definitions are known, they are just applied without providing the rationale. In fact, the definitions are applied in situations where so-called “hypothetical play” could be shown not to work due to never ending ko taking and retaking between passes. Example 18 is one situation. Ⅲ 死活確認例 | 棋戦 | 囲碁の日本棋院 Before Example 18, bent-4 has already been defined and seki collapse is already defined. So Example 18 just points out the bent 4 and how that leads to a double ko seki collapse. If hypothetical play were used, white might begin to start the bent-4 but black can take in the double ko, causing white to do the same endlessly. But this is not how L&D confirmation actually works. The examples are definitions/rulings, not some procedure to be followed. The Japanese Rules never state that the players will conduct hypothetical play. The Japanese Rules expect Japanese Professionals to be able to recognize the rulings and apply them.

  3. Misunderstanding 3: Pretending that the opponent’s dead stones cause intersections to be dame (in J2003, etc.). Consider a white group of stones with two eyes: one big eye having one empty intersection and one dead black stone in it, and a separate eye. Some people argue that since the empty intersection is surrounded by is a dead stone, then it is not surrounded by only live stones, and so the point is dame. And these people argue that since it is dame and not territory, then the black stone cannot be taken as a prisoner after the game is stopped. First of all, this argument is just silly. Second, there is a different between “adjacent” points and being “surrounded.” Ⅱ日本囲碁規約(ルール)逐条解説 | 棋戦 | 囲碁の日本棋院. The rule on eyes vs dame use “surround.” Recognize that the definition of dame only considers lives stones, not dead stones, and it is considering “surrounding” stones, not just stones on “adjacent points.” For the definition of dame, it does not matter if there is an adjacent point with a dead stone on it. What matters is looking to the live stones surrounding the empty intersection (looking past the dead stone), whether those live stones are all from one player or if there are lives stones of both players “surrounding” the intersection.

  4. Misunderstanding 4: Pseudoseki causes a player to not get territory under Japanese Rules. There is one misunderstanding that thinks that the players must agree to play dame at all. Ⅱ日本囲碁規約(ルール)逐条解説 | 棋戦 | 囲碁の日本棋院. That is a misunderstanding because Article 9-2 Explanation 1 clearly states that dame filling is required in L&D confirmation – that is, dame must be filled to show living stones and which living stones are seki stones. The misunderstanding relies on Explanation 2 of Article 9-2 requiring “agreement.” However, Explanation 2 states that dame filling is not defined in the rules. Because it is not defined, there is agreement that the dame filling is proper. There is no need for any agreement to fill dame, just agreement that a filled dame is proper. In the case of an alleged pseudoseki, the dame inbetween the groups can be filled, giving one player territory by removing dame. There is no opportunity to resume the game with the dame filled because dame filling moves are not moves of the game. It should not be surprising that the end of game scoring gives a different result compared to playing the moves out during the game. Examples of this difference are given several times in the Japanese Rules.

  5. Misunderstanding 5: Uncapturable dame can be played across the board to prove life. This is a misunderstanding of L&D confirmation. If there is a question about L&D status, the dame would be filled to settle the question. Dame filling is a requirement of L&D confirmation. If a dame is being used to prove life, that dame can be filled and and then L&D can be considered again. Dame cannot be used to prove life.

6. Misunderstanding 6: “Enabled by the capture.” This is something taking from the English translation. The question is not “enabled by the capture” but whether stones that can be captured (at least partially) can produce an uncapturable stone. There is nothing being “enabled” by capture. Just that some stones can be captured and that a new uncapturable stone is placed.

I could go on and on. But basically anytime the J2003 comments mention a supposed issue in the Japanese Rules, it is actually a misunderstanding of the author.


Particularly about your point 6, have you seen the previous discussion about this position?

Screenshot from 2021-12-03 16-15-20

I seem to understand, but maybe I am totally misunderstanding you, that you imply that here white does not need to defend, which strikes me somewhat inconsistent with the rest of the quite “traditional feeling” things you mention and also goes against what a few Japanese pros concluded for the position. The reason that white would not need to defend is that all his stones can be considered alive because of new permanent stones appearing: for the E18 stone it is easy, like snapback, new stones appears right there. For the corner 5 stone white group, after the sequence black takes E17, white passes for ko, black D19, white takes E18, black takes B19, white takes F19, the new permanent F19 stone appears, possibly marking A19 stones as alive, thus white not needing to add a defensive move.

Also, this is just a language issue, but I do not get what you mean when emphasizing “it is not enabled by the capture”, just “can produce an uncapturable stone”. Both are similar intuitive vague English terms to me, which must be interpreted “correctly” (say, adding a new stone inside your previous safe territory somewhere else obviously does not count, examples such as the above might count or not count… etc)

1 Like

I did not mean to imply anything about the specific positions discussed above, just general misunderstandings I see.

As for the difference between “it is not enabled by the capture” and “can produce an uncapturable stone,” some people argue that “enabled by the capture” means that the capture must first take place and then the new uncapturable stone is played after the capture, but that is not my understanding of the Japanese Rules. My understanding is that living stones are stones that cannot be captured or, even if they can be captured, they can produce a new uncapturable stone (before or after the capture).

As for this position, is the question whether the 5 white stones can be determined as alive by playing a new stone at F19 which is uncapturable (or even E18)? If this is the question, then I think there is a misunderstanding.

The purpose of Life & Death Confirmation in the Japanese Rules is to determine the status of stones. If the question is what is the status of the 5 White stones in the upper left then F19 (and E18) is irrelevant. F19 just shows that the 17 White stones are alive. F19 being uncapturable says nothing about the L&D of the 5 White stones because the Life & Death of the 17 stones is not intertwined with the 5 White stones. Considering the 22 White stones together is not how Life & Death confirmation works in the Japanese Rules. The examples in the Japanese Rules only consider separate groups of stones when their Life & Death is intertwined.

Since the Life of the 17 White stones has already been confirmed, they are not part of the consideration of whether the 5 White stones are alive.

Example 4 shows a position where the L&D of two seperate White groups is intertwined. Ⅲ 死活確認例 | 棋戦 | 囲碁の日本棋院

Here, the 17 White stones are alive. The 9 Black stones are alive. The 5 White stones are dead. The 3 Black stones are dead. There are now two dead groups together, but they cannot be taken off the board as prisoners at the end of the game. The players could have played a move during the game, or they can resume the game. If neither player wants to resume the game then they both lose. There are examples of all of these situations in the Japanese Rules.

The 1989 Japanese Rules do have differences from tradition. This is why they are very careful to point out situations where the players should play moves in the game to ensure stones are not seki (no territory), or to get additional prisoners, or to ensure that they do not end up in a position where both players lose. This position being discussed is one of those positions.

Yes, that is the question. This is all in the INDEPENDENT analysis for the 5 stones, this is ignoring the analysis for the white stones on the right. One might argue that if the five corner stones were considered dead, then those 5 and the 3 black stones are all dead, creating all a neutral zone there. Then by playing, when considering hypothetical play for those 5 stones, white can force a permanent stone in F19. This feels at least somewhat similar (but there are important differences, like hexahedron’s analysis explains) to the conclusion in the Life-and-Death Example 2: black to start can kill the white corner stones going first, but white manages to place new permanent stones anyway in what would not be their territory if those white stones were counted dead. Since all of these definitions are quite informal, it is hard to be sure of which interpretation is the correct, unless you have a clear precedent of this position or a very similar one.

There are a few similar (but not exactly the same situation though) precedents that I have seen, which rule that the ko must be defended. Indeed, I think that “a normal, direct ko (by normal play) must always be defended” is a very strong precedent in Japanese rules, and basically “kills” any loophole you might try to find in the “can put permanent stones” rule (if there were any).

1 Like

But that’s the thing – it’s NOT an independent analysis for the 5 White stones because the uncapturability of a stone F19 depends on the life of the 17 White stones which has already been determined. The position as presented is not a proper position for consideration of Life & Death in the Japanese rules because the stones that can already been determined to be alive are being used in the consideration of the status of other stones. If this position were prepared as an Example by the Japanese Rules Committee, they would have excluded the 17 white stones or they would have presented a case where the L&D was intertwined (e.g., the 17 white stones do not have 2 eyes and they are surrounded by black stones).

Yes, that’s what I said. That is the proper analysis. It is possible for both groups to be deemed dead in Life & Death confirmation even if that group could not be killed during the game as explained in the Japanese Rules: Ⅲ 死活確認例 | 棋戦 | 囲碁の日本棋院

This board position presented above is NOT similar to Example 2: Ⅳ 日本囲碁規約改定の概要 | 棋戦 | 囲碁の日本棋院

In Example 2, the status of the 2 White stones depends on the status of the 6 White stones. That is, the 6 White stones cannot be deemed alive on their own, and the 2 white stones cannot be deemed alive on their own. Therefore, the L&D considers the 8 White stones 白の八子 together.

Screenshot from 2021-12-03 16-15-20
Why bother arguing that playing a White stone at F19 proves the life of the 5 White stones? Why not just play a move at J16? Or better yet, a move at M11 certainly cannot be captured because it is off the board.

All of these statements are misunderstanding what Life and Death Confirmation is and what it does.

A White stone at F19 can only prove the life of the 17 White stones – it has no bearing on the status of the 5 White stones.

Well, the argument is that J16 “obviously has no relation to the white corner group. If you change all the white corner group to be black stones instead, you still can play J16 in identical way.”. While, if you for example change all the five white stones to be black stones instead, now white has not way to force a permanent stone at F19, so that is a big obvious difference between things like J16 or K19, and F19.

I think that you are saying that, “if you have already proved before that one of your groups is alive, then adding an extra stone immediately adjacent to that group of yours can never count as enabled by another group under analysis”. This is not obvious from the rules text at all, of course, which is vague, but I agree that it makes “intuitive sense” to a go player, and more or less aligns with the hexahedron analysis. And would rule all these kind of precedents as needing defense.

That’s not what I’m saying, though it is similar. I did not use the term “enable” and I already explained why using the term “enable” is misleading. I’ll copy what I previously said:

From this definition of Living Stone, I am saying: “A stone that is uncapturable by means of confirmed-as-alive stones does not ‘confirm’ anything about the status of a different set of stones.” The point of L&D Confirmation is not to just place some stone that cannot be captured with already alive stones. The point of L&D Confirmation is to determine whether a particular set of stones having an unknown status is alive.

I think the Japanese Rules might be vague if taken out of context. But words have meaning and context is important to understanding. “Confirmation” refers to a determination of something unknown or merely suspected. A White stone at F19 does not help determine anything that is not already known about Life and Death status. Therefore, it is no longer part of “Life and Death Confirmation.” As for context, it is clear from the L&D Examples in the Japanese Rules that Life & Death is confirmed one set of stones at a time. It is also that the sets of stones being analyzed are only only analyzed together when their L&D status is intertwined.

Given the context of the Rules and the meaning of “confirmation,” it’s clear that a White stone at F19 does not “confirm” that the 5 White stones are alive.

As mentioned earlier the real enabled stone is at E18, after B captures and W replaces it. B can prevent a new stone there originally (by not taking the old one), but cannot prevent it if he wants to capture the corner, thus it is enabled by the corner capture.

This is very similar to LD example 5, which shows that even inside captures (of the 3 W stones there) that depend on gaining temporary liberties from the outside are not valid captures, if the opponent can replace those temporarily taken outer stones.

E18 is not a so-called “enabled” stone for the 5 White stones. E18 is a so-called “enabled” stone for the 6 white stones (5 in the corner and 1 capturable in the ko), proving that the 6 white stones are alive. However, E18 is not a so-called “enabled” stone for the 5 white stones because those stones must also be considered on their own and they are dead when considered on their own. The Japanese Rules often deem a larger group of stones to be alive while a subset are dead. Since a portion of the stones are dead, this means that the position is seki and does not have territory without White teire inside what would be territory.

The Japanese Rules often consider multiple groups of stones both separately and together when determining their status. This is shown in examples 24 and 25: Ⅲ 死活確認例 | 棋戦 | 囲碁の日本棋院

In Example 24 (above) Black and White considered as a whole are living stones, BUT they are seki because a subset of White’s stones are dead. 白△は「死に石」白□は「活き石」White-triangle is a dead stone while White-square is a living stone. If Black captures White-triangle and White plays a new uncapturable stone at c, that does not mean that White-triangle is alive. All it means is that White-square is alive, which can already be confirmed. A new uncapturable stone that is uncapturable due to stones that can already be deemed alive does not determine the status of any other stones. Because White-triangle is dead, Point B is dame, and so these stones are seki without territory even if Point A is filled.

In Example 25 (above), the single stones capturable in a ko are dead but the groups of stones as a whole are alive. Stones must be considered both separately and together.

Example 24 is more relevant than Example 5. Example 24 shows a dead stone near already alive stones where adding a stone to the already-alive group does not mean that the dead stone is alive. Also, Example 5 does not prove your point because it is not based on so-called “enabled” stones (it is based on stones that cannot be captured at all). In fact, Example 5 actually shows my point.

Example 5 (above) has 白四子 White’s 4 stones are being considered together. If Example 5 were to consider the 1 white stone separately, it would state that it was dead. Example 5 is similar in that the stones are seki, just as they are in this example we are discussing:

You are confusing L/D (which works string-by-string), and seki/territory detection (which is for multiple strings after their L/D is determined). Enable is part of L/D, string-by-string capturability detection.

The corner string in lightvector’s example is alive, because it cannot be captured without enabling the opponent to play a new uncapturable stone at E18. The opponent (W) cannot play a new uncapturable stone at E18 originally, this is only made possible by taking his old stone from there, which is inavoidably part of the corner capture.

The outside sacrifical capture is necessary to get new temporary liberties for the other capture. This is the same as example 5, where 3 W stones cannot be captured without the outside sacrifical capture, enabling new uncapturable stones outside, some of which could have been prevented originally (if not trying to capture inside).

I am not “confusing” these two. Seki/territory determination is based on L&D, obviously. A seki occurs (A) when both players have uncapturable stones with dame or (B) when both sides have living groups where a portion of the stones are dead, such that they have dame. This diagram being discussed is case B: the 3 inner black stones are dead but there are also 5 dead White stones in the 22 living stones.

A new uncapturable stone at E18 simply shows that the existing stone at E18 is alive. It says nothing about the 5 dead stones because those are a separate group of stones that would be considered both separately and as a whole under the Japanese rules. As a whole, the 22 White stones are alive but that doesn’t change the fact that the 5 White stones considered separately are dead. A new uncapturable stone at E18 is a separate group of stones from the 5 White stones. The stone at E18 is alive when considered separately (unlike some ko stones, which are dead), whereas the 5 White stones are dead when considered separately.

This is utterly irrelevant. L/D is at string level (no groups yet), and a new enabled stone can appear anywhere (as long as it only became possible due to the disputed capture, was preventable otherwise). Example 4 shows this, but is clear from the rule anyway.

The 5 W stones here are like the 3 W stones in example 5: cannot be captured without a sacrificial capture on the other side for temporary liberties, thus without an enabled stone appearing on that other side, away from the disputed capture.

Screenshot from 2021-12-03 16-15-20

You’re misunderstanding me. When I say that 5 White stones are dead, I am talking about the 5 White stones separately, not E18. If we are talking about 6 White stones including E18 then I have always said that those stones are alive. There is no doubt to me that the stones as a whole are alive. The real question is whether a portion of the stones are dead such there is dame that would require teire to make territory. In example 4, the stones have dame and no territory.

My position does not depend on where the new uncapturable stone is placed, it depends on what original stones are being considered. I have no issue with Example 4.

Also, Example 4 is not really relevant because it involves living White and Black stones, and if I’m not mistaken, you are arguing that the White stones are alive while the black stones area dead. Also, Example 4 is seki, and I believe you are arguing that this is not seki.

The 3 White stones in Example 5 are seki stones and White does not have territory. Are you saying that the 5 white stones are seki stones? Because that is my position. The 5 White stones are seki stones because while the 6 White stones are living stones (they can create new uncapturable stones), the 5 White stones are dead when considered separately. Because the 5 white stones are dead separately, B19 is dame and White has no territory.

I think this situation is also like Example 25:
The 12 White stones are alive but they are seki stones because a portion of the stones are dead when considered separately (the 1 ko stone). Therefore, point ‘a’ is dame, not territory.

The Japanese rules states that the stones in the ko are dead stones when considered separately:

You seem to have missed almost everything I wrote. Even considered separately (this is what “L/D works string-by-string” means) the corner string is alive. B cannot show a sequence with which he could capture it without allowing W to play an otherwise/previously unplayable new stone.

This is the same for the corner string here and 3 W stones in example 5: both are uncapturable without enabling a new stone that only became possible because of the capture.

My position is that it is the capture of E18, not the 5 White stones, that so-called “enables” White to play E18 and another stone such that they are uncapturable.

The stones in Example 5 have dame. My position here is that the 5 White stones have dame. I agree that the example is similar.

I posted at the original thread over on L19x19.

The main issue here is whether White owes teire or not in a position where White might win by half a point if this were territory. I think the confusion arises because this is not a proper end-of-game position. Black will not end with this position. Black WILL take the ko and either force White to throw-in and eventually play teire, or if there is a double ko on the board the game can end in No Result.

White cannot leave this position because Black threatens to capture the 5 White stones and save his stones. White must retake. If White retakes and tries to end the game without playing an additional stone and claim territory, black can dispute and retake the ko. White teire is mandatory to achieve territory.

This kind of logic doesn’t work. Taking the ko is a necessary preparatory move, just like in example 5 the capture of W 3 stones starts by capturing in the corner for temporary liberties. In example 4, it is also not the capture of a W string that literally enables the save of the other, but the necessary approach move before it (which loses a liberty).

But all this is irrelevant, what matters is can the string be captured without enabling a new stone or not (always starting from the stopped position, for each string).

(And you keep posting the wrong diagram, the example had 3 variants but A and C was only for comparison, the interesting one was B, where the reinforcement question was real.)

This is not how it works. Consider this example modified from the same diagram in the Japanese Rules (also posted to L19x19):

If W7 passes for the Double Ko (instead of passing for the Upper ko as in the actual Example) then some might think that B8 also needs to pass for the Double Ko, causing a never-ending passes and retakes in the Double Ko without ever resolving the upper corner. However, consider if W7 passes for the Double Ko and B8 captures White in the upper corner. This lets W9 retake the Double Ko leading to capture of Black in the lower corner. Even in this case, the fact that White was able to create new uncapturable stones in the lower left corner says NOTHING about the L&D status of the White stones in the upper left because the stones in the lower left were already deemed to be alive by virtue of the double ko. This is the same as adding 2 new uncapturable stones to the already living stones with 2 eyes in Variation A and B.

Context is important. I am discussing Variation “A” with you because our conversation is a continuation of the discussion where Elsantodel90 asked a question about Variation A and I responded discussing Variation A.


A position similar to Variation B was just mentioned on L19x19 (where there is 1 empty intersection that would create a ko shape after White throws-in). In that situation, White still has dame in the 5 stones unless they play an additional teire by the same reasoning I’ve given here. But in Variation B, Black actually would NOT capture and induce White to throw in because that would lead to a position where White’s 5 stones were actually uncapturable in L&D confirmation.

In Variation B, White wants a half point win by claiming everything as territory, but they will lose because the 5 White stones have dame and they will only have territory if White plays another move.

Even if we rely on the “enable” interpretation that you like, you should recognize that it’s not simply the case that “taking the ko is a necessary preparatory move,” it is the case that the taking of the ko stone is the only move that “enables” new uncapturable stones to be placed. This should be clear because the 5 stones do not even have to be captured at all.

If Black attempts to prove that the 1 White ko-stone is dead, this diagram can show why the 1 White ko-stone is alive, separate from any capture of the 5 White stones in the corner. The capture of the 5 White does not “enable” any new uncapturable stones to be played.

At this point in Life & Death Confirmation, Black is not concerned with whether the 5 black stones are capturable or not. Black is only considering the L&D status of the 1 White ko-stone.

There is no such thing as other stones “already deemed to be alive” because L/D works string-by-string, checking capturability without enable for each string, always starting from the original stopped position, and regardless of status of other strings (no circular reference or order of analysis).

It is possible to invent different interpretations for enable, and as someone wrote a Japanese referee might even ignore the written rule and apply his own informal judgement, unconstrained by the need of logical consistency (like you). But J89 has no room for changing the string-by-string nature of L/D, with a fresh analysis for each string independently of the status of others. You need a new ruleset.

Just like in example 4, where it is the approach move that enables the save (and new stones) already, the actual capture itself is not even needed for that. But this is irrelevant - W is alive there because the capture cannot be performed without enabling the save.

This would only matter if you would be interested in L/D for the position AFTER the ko was already taken. But for scoring the stopped position you need L/D for that position, not for a different position a few moves later. And from the scoring position capturing the corner string would necessarily enable a new stone.