Is white alive under Japanese rules?

I don’t think this works for a large number of life and death situations. Any group with only one eye can be killed when the defender only passes, and any group with two or more eyes can be killed when all but one of those eyes are large enough to be filled with at least a one internal eye (or false eye along the edge) group. Also any seki can be killed when the opponent only passes.

Determining life and death correctly depends on both players playing correctly in the vast majority of situations.

1 Like

You seem to have misunderstood the definition. “Can take unconditional control of” doesn’t mean he already has to have unconditional control in the stopped position. It means he can play on in a way that eventually his control of the surrounded region becomes unconditional, so that he can stop and always pass afterwards.

In practice, this can be done for independently alive groups by fixing all weaknesses, filling the region with 1 point eyes etc (as a demonstration, in hypothetical play only oc).

As for seki, no player can take unconditional control there with continued play, so neither has territory.

I would not say that this is merely a theoretical problem, nor that it only affects beginners. I’ve seen dan-level players not understand how to properly handle things like bent-four-in-the-corner. While some trickier cases are very rare, they are bound to happen eventually. Just earlier this week, we saw some dan-level players dealing with a moonshine life situation.

I think that the major territory rules (note: I’m specifically talking about the official Japanese and Korean rules, in particular) introduce so much complexity into how life and death is precisely determined, that nearly everyone (including even myself) winds up working with an incomplete approximation of these rules. I put myself in this category, since, despite spending a lot of time studying and thinking about peculiar situations, I have come to realize that the Japanese rules are even more complex than I have ever imagined, that there likely more rules beasts that may escape my understanding and challenge gaps in my knowledge, and hence I can only assert confidence in maybe only 99.9% of situations that may arise in typical OGS games.

Probably most people, especially those that reach dan-level, have a good enough approximation that they can properly handle >99% of situations that occur in their games. These approximations do go pretty far, as the tricky situations are quite rare, and the peculiarities don’t really affect strategy in the vast majority of typical games.

However, when something strange does happen, it can occur that neither player really knows how to properly handle it within the rules. It’s not necessarily just a breakdown in understanding how to best strategically handle a situation. Rather there are positions where both players will (based on their own, likely flawed, understanding) assert that the position is settled (and that the game is over and should be scored), but they are in disagreement about the status of things, based on the fundamental differences between their respective approximations of the Japanese rules. It’s like almost everyone has their own different version of the rules, and in >99% of cases, these ugly differences never have a chance to be laid bare, but occasionally there is such a case, where the players each find out that they are essentially playing with a different set of rules than their opponent. Almost no one can really claim that they are playing by the official Japanese rules, since nearly everyone is working with just an approximation.


I was just saying this definition of unconditional control is clearly insufficient.

I think territory can be defined based on unconditional control instead (ie. control that cannot be changed even if the player only passes).

If that was something like, territory is a region where the owner can eventually establish a state of unconditional control of that region against any continuation of alternating game play (strictly where the owner plays first, I think). Unconditional control being control that cannot be changed even if the player only passes.

But even with this wording there seem to be issues with cases like unclosed areas incomplete, teire weaknesses.

1 Like

What you quoted is a definition of unconditional control (an auxiliary concept). For a definition of territory, please see my original post above:

Unclosed areas are not territory as they are not surrounded. Teire weakness seem no problem either since (if the player tries to omit them) the opponent can force their fix during play for the points due.

I don’t dispute this. And even with better rules, territory scoring itself - forcing a static evaluation of a stopped position - will always be a bit harder to follow than area. But this disadvantage needs to be weighted and multipied by its practical effect - and compared to the weighted practical effects of things like more accurate scores, dame omission, easier counting, better komi, simpler ko.

I think territory is easily +EV compared to area, but oc personal preferences remain, and I don’t reject area scoring either. As mentioned above, I believe in universal rules with true (unlike AGA) area<>territory duality, so beginners can play area for example. The ideal solution seem to be minimalistic and uniform rules of play, where the same legal moves, ko and long cycle rules, two pass stops etc. could work correctly with all scoring methods and game phases.

1 Like

Aren’t these closely related in territory rules? If you can kill a group like a bent four, but it requires you to fill a number of ko threats, which in Japanese and territory rules loses points, while in area it wouldn’t, then doesn’t it follow that there should be some special variant of the ko rule that is necessarily?

At least the alternative in this specific example would be that in hypothetical play, you get to play an unspecified number of moves while the opponent presumably is passing before starting a ko. That seems a bit cumbersome, while the idea of needing to pass for a ko (once only or more times) does kind of eliminate that need.

Is changing the ko rule not intended to be some shortcut to these procedures of fixing all weaknesses?

1 Like

The point about teire is that when such a situation occurs at the end then the rules need to resolve how this is settled for the purposes of the score. If scoring is not possible then this weakness needs to be described nearly fully by the rules so that its clear if the game can be scored or not.

The official Japanese rules seem to resolve this by fairly simple hypothetical playouts with a modified ko rule, resulting in life and death status for all groups. I don’t think this is so easily replaced by a different description.

My understanding of scoring is that most un-played teire points will still result in a living shape when exploited by the opponent, and as a result the player gains additional territory when these are not played out and the game is instead scored as it stands. I think that the rules need to operate in a consistent manner in all the situations where these points exist in the game state, if they are discovered by the players at some point or not.

Though unclosed areas are well enough defined (but depends on the life a death status being correct) that you can insist these are resolved before scoring commences.

Not necessarily. Hypothetical play itself is there exactly to neutralize points lost on capturing. Filling in a number of ko threats is not really different to playing 3 unanswered moves to capture 1 lone stone and prove it is not alive (a common beginner question). Both are free in hypothetical play - that’s why it is used instead of resumption like area.

With bent4, the complications only start if there is an unremovable threat also - which is exactly when Japanese rulings become doubtful as well (and diverge from Chinese). There are harder questions like moonshine life, but even in that case hypothetical play can match real play (either making it capturable in both by some kind of moonshine ko rule, or accepting as uncapturable like it was until 1949).

1 Like

Yes, but the opponent will not allow scoring like that, but force the fix by approaching on dame in the game - or in resumption if overlooked. This is exactly how it works on IGS and other servers: either you play your teire voluntarily, or the opponent starts to approach instead of passing.

Ok but if I have forty to play, like every tigers mouth on the board, and a number of other ataris from groups I haven’t removed, then it’s really not a matter of it being free but just a matter of practicality, not to mention having to either then remember to remove all of those moves to score properly or setup an exact duplicate board in order to do hypothetical play - which if available might be preferable for local situations anyway.

So hopefully it’s appreciable that some special ko rule might be practical even if you think it’s too far removed from area scoring procedures.

Except when they do.

The most common example of unfilled teire would be when neither player notices of course.

I’m not saying the rules should score this the same way as optimal play, but they should be able to score this situation in some consistent manner.

In Lasker-Maas if the game resumes you play new stones from your opponent’s prisoners instead of your bowl (adding a pair of prisoners to the lids when necessary), which means that while you lose a point by playing in your territory, you gain it back by removing an opponent’s prisoner

I think that from the perspective of speaking informally and/or talking about the strategy of the game, one could define terms like “alive”, “dead”, and “seki” in somewhat different ways, and in a manner that deviates than how a particular rules set might define it (or related nuanced classes) in its formal language.

For example, speaking informally in strategic discussions of the game, it seems natural that someone might classify settled groups into three disjoint classes of “alive”, “dead”, and “seki”, where “alive” specifically excludes “seki” positions, which are cast into their own special third class. In fact, this was how I was originally taught to think about the game from a strategic point of view. Sometimes this is useful, since 1) seki positions don’t score points for territory (in Japanese rules), so it’s useful to keep track of them separately (if there are an unbalanced number of eyes), 2) sometimes seki positions are tenuous depending on the life/death situation of surrounding groups (i.e., one can potentially exploit them as aji, ko threats, or possible even collapse them outright, depending on how the global situation evolves).

Thus, from a strategic and an informal perspective, I think it is reasonable that one could handle the semantics of “alive”, “dead” and “seki” in a way different than the formal language of a rules set.

However, the Japanese rules do formally categorize “seki” as a subset of “alive”. In particular, the Japanese rules does this in a peculiar way, by essentially saying that a group is in seki if it possesses (contains or adjacent to) dame. This has the formal effect (in Japanese rules) of requiring all unnecessary dame (i.e., all dame, except those needed to prevent sekis from collapsing) to be filled before scoring. Of course, many people (especially in informal and most online play) accept the shortcut of not filling such dame and only pretending that they have been filled for the purposes of judging whether something is in seki or not. Professionals, on the other hand, seem to always fill the dame points, before formally scoring under Japanese rules.

Japanese rules get even weirder when it comes to how it formally defines “dead” and how that is handled during scoring. In informal, strategic discussions, players will often use the following concepts interchangeably (working with the approximation that they are equivalent):

  1. Dead stones are those that cannot avoid being captured under normal alternating play.
  2. Dead stones are removed during scoring and added to the opponent’s prisoners.
  3. Dead stones are those that are formally considered dead under the Japanese rules, if both players pass and proceed to scoring.

Under Japanese rules, these three categories are not exactly equivalent, that is, there exist weird situations that only belong to some but not all of these classes.

For example:

  • Moonshine life (which is essentially a dead ko shape + the infinite source of ko threats from a double-ko seki) is an example of a position that does not belong to the first category, since it cannot be captured under normal alternating play, but it is considered dead under Japanese rules during scoring and those dead stones are removed.
  • Bent-four-in-the-corner (specifically a temporary seki position that can be reduced to a bent-four-in-the-corner) + unremovable ko threat is another case that similarly does not belong to the first category, but does belong to the other two.
  • Cases where dead stone(s) reside within the eye of a seki position are an example that does not belong in category two, but does belong to the other two categories. In such cases, Japanese rules actually compel the player to play addition move to capture strategically dead stones, just in order to be able to formally count them as prisoners. Of course, many players may take the informal shortcut of just removing these stones without insisting that their opponent capture them.
  • Anti-seki situations are another class of cases where stones will be considered dead, but not removed during scoring. Some anti-seki situations are unsettled positions (which could also formally result in the game being declared as “both players lose”, if they are worth enough points to swing the outcome) that under normal alternating play would only result in one side dying (and the other living) depending on who plays first, but these would be surprising considered as both sides dead (and not removed), if both players passed and tried to score.

This is not an exhaustive list, and quite possibly the three categories that I gave above don’t even cover all situations either. I would not be surprised (and frankly expect) to see new rules beast the illustrate even further categories and weirdness


I would consider both will be dame in this case lol

The 2-1 and 4-1 strings, or just the 3-1 and 3-2 points?

I’ve only ever considered empty points to be dame


Everything from 2-1 to 4-3.
Dame is usually referred to stones that don’t produce any points. For example, if you just connect back your group without any points, you are considered to be playing dame. I guess you can also use it for empty points at the endgame after all the moves that are worth points have been played.

Fun fact: “Dame” also means “useless” or “no good” in Japanese and the meaning came from Go! Because playing dame in a game essentially means you are playing useless moves :slight_smile:

Yes it’s mutual life so technically both are alive. But when you are trying to kill a group but only managed to make a Seki, do you consider the stones you played as “Alive”? Just saying haha

Yeah, that actually threw me a bit at first because when I heard it, my only context for it was Go, so I underestimated how widely it can be applied outside of Go. I have no doubt that the normal word dame applies equally well to useless moves, stones, and empty points, and I have no idea whether or not the technical Japanese term applies to all three of those

It is specifically the English technical Go term “dame” (which happens to be borrowed from Japanese, but that doesn’t affect the meaning) which I am surprised to hear being used to refer to more than just empty points. Even the phrase “filling dame” is compatible with said filling making them no longer dame. More generally, I also worry about how much more difficult it might be to define dame if stones qualify

As with “filling dame”, this phrase does not preclude an interpretation which allows for the act of playing said dame causing them to no longer be dame

Interesting that you list this as an afterthought. I’ve always been under the impression that this is the core meaning of dame, and any other meaning (like, point which will become dame at the end of the game, or point which gains as much as playing dame would) is derived, secondary, and less technical. Analogously to how life and death is defined in terms of the end state of the game, which allows us to use the terms during the game in a loose and derived sense of “will probably be alive at the end of the game”