Odd Cases 🤔 in the Japanese Rules

Ok, that’s enough pedantry for tonight, but I still have other topics that I would like to discuss, including at least:

  1. Special ko rules when determining life and death after passing.
  2. Moonshine life.
4 Likes

i got a big head after reading this.

just want to ask: do these kinds of situations happen in chinese rule? or they just resume game. and incidentally, does the “other side play first” rule apply in chinese?

4 Likes

I believe the above situations are unique to the Japanese rules, and maybe also the Korean rules, which are a similar territory scoring rules set. However, I’m completely uncertain about the technical details of the Korean rules, since I have not looked into those.

Under the various area scoring rules, if the game has to be resumed after both sides have passed, I believe they all simply continue with whichever player’s turn it would naturally be. So if white passed second, black gets to play first at resumption, and vice versa.

Special Ko Rules when Determining Life and Death After Passing

During the normal course of play, the basic ko rule (which is universal across all go rules) is in effect:

Article 6. Ko

A shape in which the players can alternately capture and recapture one opposing stone is called a “ko.” A player whose stone has been captured in a ko cannot recapture in that ko on the next move.

When a ko arises that both players want to win (e.g., the life/death of a large group may depend on ko), they must fight for it by making ko threats elsewhere in order to retake it. Thus, the existence, size, and balance of ko threats across the whole board will determine who wins a ko fight and ultimately how key positions will be settled.

A classic example to illustrate this point is the “Bent Four in the Corner”. In particular, let’s consider the below shape, which is a temporary seki, where white could play at A in order to force black to capture 4 stones, but be left with the bent four in the corner shape with white to play.

7-seki

From the bent four in the corner position, white can start a ko fight in an attempt to capture the black stones:

3-ko

Whether this would be successful depends on the status of ko threats on the rest of board. Black may even have enough large unremovable ko threats, such that white would never be able to favorably win the ko threat. This would make black’s stones (in the first figure above) seemingly invincible, and white should not collapse the seki to start the ko fight.

However, in spite of this, black’s stones are still dead under Japanese rules.

White should never actually play at A in (the first figure), since the position will be ruled to be dead, regardless of what ko threats black has. This is the case, since the life and death confirmation phase introduces a special ko rule different than that used during normal play:

Article 7. Life and death


2. In the confirmation of life and death after the game stops in Article 9, recapturing in the same ko is prohibited. A player whose stone has been captured in a ko may, however, capture in that ko again after passing once for that particular ko capture.

This rule has the effect of nullifying normal ko threats. Whether or not black has other ko threats becomes essentially irrelevant, since these threats would not allow black to retake the ko.

Note: the life and death confirmation phase are not actual additional plays, but rather the consideration of hypothetical, ideal play, just for the purpose of resolving life and death. For scoring, the board position, from immediately after the two passes, is restored, but with dead stones removed as prisoners (except for the case of anti-seki as discussed earlier).

Article 9. End of the game


2. After stopping, the game ends through confirmation and agreement by the two players about the life and death of stones and territory. This is called “the end of the game.

Article 10. Determining the result

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

6 Likes

Compulsory Dame Filling to Prevent Technical Sekis

In Japanese rules, eyes of groups in seki do not count as territory. This quirk of the Japanese rules could possibly be a holdover from the group tax effect of stone scoring rules, a historical forebear of the Japanese rules, (see this post in Pondering Upon The Rules Of Go which also links to an article discussing the history of the rules of go), although certainty of its origin is likely lost to history.

However, this rule is loaded with complexity since one must define what is seki in order to distinguish whether eyes should count as territory. It can be very difficult to distinguish between seki and living groups. The concept is difficult to pin down with a definition that attempts to count eyes, false eyes, etc., since there all sorts of contrived positions like non-seki living groups the only appear to have false eyes, or complicated hanezekis and hanezeki-like positons. Ultimately, the Japanese rules simply define seki as living stones that are adjacent to dame:

Article 8. Territory

Empty points surrounded by the live stones of just one player are called “eye points.” Other empty points are called “dame.” Stones which are alive but possess dame are said to be in “seki.” Eye points surrounded by stones that are alive but not in seki are called “territory,” each eye point counting as one point of territory.

This rule has the side effect of compelling players to fill nearly all dame (except those that must be left unfilled to preserve actual sekis), since otherwise any non-seki living stones that are adjacent those dame points would be considered seki and have their territory nullified.

It’s common practice for players to leave dame unfilled when playing with Japanese rules. This is a result of the common misconception that dame do not have to be filled under Japanese rules, however, as we see from the discussion above, the opposite is technically the case! I’ve previously discussed this discrepancy between this common practice and the actual Japanese rules here: Japanese Rules Popularity

7 Likes

Compulsory Capture of Dead Stones

In most situations, the Japanese rules compel players to pass before capturing all strategically dead stones, since playing additional moves to actually remove dead stones would reduce one’s points under territory scoring. For example, consider the following position.

dead%20stone

In this corner, black has 4 points (3 from territory plus 1 for the white prisoner). However, if black were to play an additional move here (at the end of the game, when they should pass) to capture the white stone, black would only get 3 points (2 from territory plus 1 for the prisoner).

It is the following rule that allows black to simply remove the single white stone without capturing it, since it is a dead stone within black’s territory.

Article 10. Determining the result

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

However, the fine print of this rule is that dead stones are only removed from territory, as we just discussed in the previous post, eyes from groups in seki do not count as territory.

Consider this example from How does this get scored (Japanese)?

The position is a seki, and the single white is dead, but it is not removed and not counted as a prisoner, if black does not play another move to actually capture it. Hence, black should play another move (which should be delayed until the very end of the game, since it is not urgent) to actually capture that white stone, in order to a get net gain of 1 point. Note that there is already zero territory to lose (from self-filling), since it is an eye from a group in seki.

The Japanese rules commentary actually gives a slightly more complicated example to illustrate this concept:

Here, not only should white capture the two black prisoners, black is then compelled to throw in another stone (to prevent the seki from collapsing), which white should then capture again before ending the game. Thus, white should extract 3 points (all from prisoners) from this position.

10 Likes

Tankyou yebellz fo both of you replies. They were very helpfull
.

1 Like

Double Ko Seki is a Terminal Anti-Seki

When we discussed anti-seki earlier above (see Odd Cases in the Japanese Rules and Odd Cases in the Japanese Rules), we considered some cases that were not “terminal” positions, i.e., they were unsettled in the sense that at least one of the players (if not both) should have played another move to maximize their score.

It might seem that all potential anti-seki positions would be of this unsettled/non-terminal nature. However, the double ko seki is an example of a terminal anti-seki position. Specifically, the single black stone and single white stone that are both in atari will be considered dead, but not removed (according to article 8) since they do not reside inside territory.

Neither player can force capture of the other. If Black plays at A, White can play at B to avoid capture, but then the cycle cannot locally continue due to ko. Likewise, if White plays at B, Black can play at A to avoid capture.

Unless some other ko arises on the board, this position will persist until the end of the game. After passing, during the confirmation phase, both players can demonstrate that the single stones in atari are dead, since they can be taken, and although it would allow the player to again play another stone, that new stone would not be uncapturable:

Article 7. Life and death

1. Stones are said to be “alive” if they cannot be captured by the opponent, or if capturing them would enable a new stone to be played that the opponent could not capture. Stones which are not alive are said to be “dead.”

The confirmation phase would actually get stuck in an infinite loop:

  1. Black captures a ko stone by playing at A.
  2. White captures a ko stone by playing at B.
  3. Black passes in order to retake the ko at B.
  4. White passes in order to retake the ko at A.
  5. Black retakes the ko at B.
  6. White retakes the ko at A.
  7. Black passes in order to retake the ko at A.
  8. White passes in order to retake the ko at B.
    … return to step 1.

After step 8, we are back at the original position before step 1, where they can continue repeating the sequence. By symmetry, a similar sequence would occur with White starting. This confirms that neither player can capture the bulk of the other player’s stones. However, the single stones being traded in ko captures are dead, since although their capture permits another stone to eventually be played in it’s place, it would ultimately be captured as well.

Note that the passes during the confirmation sequence above are necessary in order to allow ko recapture and continue the sequence according to this rule:

Article 7. Life and death


2. In the confirmation of life and death after the game stops in Article 9, recapturing in the same ko is prohibited. A player whose stone has been captured in a ko may, however, capture in that ko again after passing once for that particular ko capture.

One might find it peculiar that there could be an infinite sequence during the life and death confirmation. The “no result” rule (which I will talk much more about in another post) would seem to be relevant.

Article 12. No result

When the same whole-board position is repeated during a game, if the players agree, the game ends without result.

However, I believe this “no result rule” is not triggered since the confirmation phase is not the “game” (alternating play phase). After all, the confirmation phase is only a consideration of hypothetical play, for the purposes of demonstrating that the stones cannot be captured, since even playing out the cycle indefinitely still never results in the stones being captured. This type of reasoning perhaps presents a major difficulty for algorithmic implementations of the Japanese rules.

Double ko seki is a source of infinite ko threats and further complications

I mentioned earlier that the double ko seki would persist to the end of the game, unless another ko situation arises. Basically, the double ko seki is two active kos, where a player could always take one to force the other player to take the other (assuming the life of that group is big enough to compel a response). Thus, the double ko seki can act as a source of infinite ko threats should another ko arise.

If another simple ko were to arise, then the game very well could end up as a “no result” by “triple ko” according to rule 12 (more about this in another post). However, it could also create another completely different kind of beast, Moonshine Life, which is not necessarily a “no result” situation (I will also post more about this in another post).

6 Likes

Fantastic stuff, I love it. I had no idea you have to capture stones in seki in Japanese rules.

2 Likes

Repeating Positions Can End the Game with “No Result”

It is possible for the game to wind up in a loop, where the same board positions repeat in a cyclical sequence. When this happens, the game might (but does not always have to) end with “No Result”. Here is the relevant rule.

Article 12. No result

When the same whole-board position is repeated during a game, if the players agree, the game ends without result.

A game that ends in “no result” is viewed as an inconclusive outcome that is different than the game ending with a tied score (jigo), which would only be possible if using an integer komi. The latter can be considered a draw, but the former should not. Traditionally, if a decisive result is needed (e.g., in a tournament), the players would have to replay an entire new game to make up for a game that ends with no result.

Further, a key provision of the rule is the phrase “if the players agree”. In this post and several of the following ones, we will explore the nuances of that provision.

The classic triple lo example for no result

A Triple Ko is perhaps the most commonly given example of a cycle that can lead to a no result. Broadly speaking, any position where there are three active kos could perhaps be called a triple ko, however the exact specifics can wildly affect what happens. Hence, let’s start with a basic “self-contained” triple ko, where all of the kos are part of the same local position:

Starting with Black’s turn to play, Black can give reprieve to their stones in atari by capturing one of the kos (move 1 in the diagram below). Now, White’s stones are in atari, and since White cannot recapture the ko, they must take another ko with move 2, which in turn forces Black to take the third ko with move 3.

Note that after 3 moves, the parity of the kos have flipped:

However, it’s White’s turn to play, and White must similarly continue by recapturing the first ko with move 4, which further compels moves 5 and 6, if both players wish to save their stones.

After move 6, the entire position has returned to where we had started. If neither player wishes to give up their stones, they could continue this loop indefinitely, and hence trigger the rules to end the game with “no result”.

Side note: a helpful way to visualize triple ko is to imagine each ko in the diagrams above as a light switch. Black is trying to flip all of the switches down, while White is trying to flip all of them up. They take turns flipping one switch at a time, but they cannot flip the switch that their opponent last touched.

Triple ko does not have to end with no result

The no result outcome for the basic triple ko discussed above requires this loop to occur and for both players to agree. If one of the players was so far ahead in score that it did not matter whether their stones were captured, then their best play would be to abandon this loop (assuming that they would rather win than end the game with no result). However, even if one player was far enough ahead, they could mistakenly persist with the cycle and have a winning game turned into a no result.

In general, three active kos does not necessary mean that the players are locked into playing the cycle indefinitely either. Moonshine Life is an interesting example, which I will talk more about in another post, where there are three active kos, but should not end in no result (unless one player makes the blunder of insisting on the cycle rather than winning).

Just a few more notes for now

  • Triple ko involves a six-move cycle, but a four-move cycle (that could force a no result) is also possible with a position called Eternal Life.
  • The are “three-move” and “five-move” cycles the do not cause a “no result” since they have an unbalanced number of captures.
  • The possibility of abusing double ko seki is perhaps another reason for requiring both players to agree to a no result.
  • The exact timing/procedures for compelling players to break a cycle (if they do not agree to a no result) is somewhat unclear, and it could have strategic implications.
  • Comparing the preference between jigo vs no result would perhaps depend on external factors, such as specific tournament procedures, rating systems, prestige, etc.
  • Moonshine Life is also very interesting, as mentioned above.
  • I plan to write more about most of these topics on this list. If you thought my earlier posts revealed some really weird quirks about the Japanese rules, then buckle-up, since kos are where things really get crazy.
  • There are many more examples of longer and more complex cycles, but I can’t hope to cover or even understand them all.
3 Likes

Is a “No Result” Better than a Tie?

Consider games that are played with integer komi, which makes the possibility of a tie by jigo (equal scores) a possibility. Suppose that a cyclic position arises (like triple ko or eternal life) and the following becomes apparent to one of the players:

  1. Their opponent is committed to continuing the cycle is willing to agree to ending the game as “no result”.
  2. They can abandon the cycle (by playing elsewhere), but that would lead to the game ending as a tie (jigo).
  3. They can continue the cycle and accept the game ending with “no result”.

What should the player do? It is not clear from just the rules whether a tie or “no result” is preferable. While one could assume that both are preferable to losing and that winning is better than both, these two are not directly comparable.

In fact, I think that depending on other external factors, either could be preferable, or it could remain indeterminable.

Ratings Systems

From a ratings system perspective, a tie should be treated as a data point that suggests that the two players are evenly match and close in strength. Hence, although different systems will vary in the exact formulas and implementation, the general expectation would be that a tie would bring the two players’ ratings closer together (by decreasing the higher-rated player’s rating and increasing lower-rated player’s rating).

On the other hand, a “no result” should be treated a game that did not produce an outcome, and hence most rating systems would ignore such games, until they have been replayed to produce a settled result.

Hence, the higher-rated player might prefer the “no result”, while the lower-rated player might prefer the tie".

Likelihood of Winning a Rematch (and Risk Tolerance)

Suppose the agreed contingency is to play a rematch in the case of a “no result”. If a player believes that they have a very high chance of winning a rematch, then they would likely prefer the “no result” rather than accepting a tie. On the other hand, if they feel that they have a high chance of losing the rematch, they would probably prefer the tie instead.

If a player believes that they have a roughly 50-50 chance of winning vs losing the rematch game, then whether they prefer to take a certain tie versus forcing another game by taking a “no result” would depend on their risk tolerance, and how strongly they might prefer a tie over a loss, or a win over a tie. This risk tolerance and outcome preference could also depend on further additional factors, such as tournament considerations.

Tournament Procedures and Position

Suppose the choice to accept a tie versus a “no result” occurred in a tournament setting. Consider a tournament structure that involves accumulating points in order to move onto the next round, and that they give a full point for a win, half a point for a tie, and no points for a loss, while “no result” games are replayed.

A player might find themselves in a situation where they must win their next game (i.e., the tie is as bad as a loss since the full point is needed in order to advance). In such a case, they would prefer the “no result” over a tie, since that still gives them a chance to advance. Alternatively, a player might find themselves in a situation where either a win or tie is sufficient to advance, but a loss would lead to elimination before the next round. In this case, the player would prefer the certain tie versus the uncertainty of replaying the game after a “no result”.

Alternatively, a tournament might decide not to replay any “no result” games, and simply treat them as ties for scoring purposes. Then a player might not care whether they finish the game with a tie or a “no result”, unless they are also concerned about potential rating implications.

Subjective Concerns about Image and Prestige

Even in a one off game, where the players understand that they will not play a rematch in the case of a “no result”, but instead let that stand as the outcome, the players might still have different preferences regarding the possibility of the tie versus “no result” endings. One of them might subjectively view reaching a jigo with their opponent as more or less prestigious/desirable than reaching a “no result”.

4 Likes

Do you know of any rating system that does something different from the obvious solution of:taking, for each player; the average of the points he would have gotten for a victory and the (negative) points we would have gotten for a loss?

1 Like

Since many rating systems aren’t about “this game is worth X points”, but use your performance in multiple games at once to estimate your rating, the answer is yes. (Simply because they don’t assign a point value to the game, they cannot average them)

3 Likes

(I accidentally bumped with a stupid question :expressionless: )

eyes of groups in seki [in Japanese rules] do not count as territory

I really should have known this, huh?

1 Like

In another post, I linked to and briefly discuss a study of a the history of the rules. There’s speculation that the peculiar “eyes in seki don’t count” aspect of Japanese rules may be a historical remnant of the group tax effect from the stone scoring rules.

3 Likes

There is a fascinating side-effect of this version of the “written Seki rule” (which I think, nobody actually applies in practice, even professionals). I’ve seen it called “pseudo-seki”, so it is the name I use for the lack of a better name. I found it on the internet here: https://lifein19x19.com/viewtopic.php?f=45&t=3666&start=60.

The position is the following:

unfillable-dame-pseudo-seki

No prisoners, 5.5 Komi.

The black group in the upper right corner is involved in the possiblity of step ko: When completely surrounded, White could capture the corner stone, then if black cannot retake the ko, white can add the J7 stone to start a direct ko for the life of the black group. In the game however, white has one and only one ko threat at A5, and black also has one and only one ko threat at b (which is large enough: this will change later!), and thus white has no hope of winning the step ko during the game (since in the resulting direct ko, black will take first).

Furthermore, the black group is deemed alive by endgame hypothetical-play, with no need to add an extra defense move (capturing the white stone). This is because the existing ko is a step ko, so in hypothetical analysis, there are no ko threats, so white will take first, black will pass for the right to take the ko, white can only make the J7 move and then when black takes the ko, white can do nothing, so black wins the hypothetical play.

This is crucial because the game is very close: if black makes an additional defensive move like capturing the white stone so that there is no ko and his group is clearly alive, he loses by 0.5 points.

So, it looks like black should win by 0.5 points. However, white will argue that the top groups touch dame, and so are in seki! If they are counted as seki, white only loses 4 points of territory, but black losses 5 points (4 territory + the stone inside), and thus, white wins. So, to avoid this, black’s only hope is to fill dame, playing both triangle-marked intersections.

Note that, if white fills those, our analysis so far has been correct, the groups would not be in seki anymore, and black rightfully wins by 0.5 according to Japanese rules.

However, when black fills those dames, the situation changes, because filling them has made the black group larger. This means that, if white captures it, he will gain more points than before (2 more territories + 2 more prisoners). This makes black’s ko threat at b) no longer large enough: that means that if black fills the triangle-marked dames, then he loses because white will now start the ko, and will not answer black threat at b), effectively exchanging both corners, but now the difference is winning.

Thus, the seki rule makes this possition winning for white, if followed as written.

To the best of my knowledge, such a situation has never been actually claimed in a real professional game. My understanding is that in a real game, even professionals would deem the black group to be alive and not claim a seki instead, considering the written rule just “flawed”. Unfortunately I have no evidence for this, other than the lack of a counterexample (that would be: an actual pro game where a pseudoseki situation occurred and in which white won by claiming seki).

8 Likes

I’m convinced that Go Seigen would be the kind of pro player that claims seki. The whole argument reminds me of his rule disputes.

2 Likes

What a position!

Interestingly, (without having explicitly checked) I think KataGo’s computer-suitable formalization of Japanese-like rules actually does agree with your guess of what pros would think, rather than the literal interpretation of the written rules. White can neither win the ko during normal play nor during post-game cleanup, and during cleanup black can safely fill the dame and capture the white stone without cost, so will win by 0.5.

3 Likes

Wow!! Thanks for sharing that “KataGo” rules formalization, I had not seen them before. Indeed, the two-phase idea is genius, the amount of “theoretical match” with actual Japanese rules is incredible for such a well-defined ruleset that solves disputes based on actual board play (and not some “hypothetical best play as defined by competent authority / exhaustive analysis”).

All territory-style rules based on actual play that I had seen before “suffer” from such non-Japanese features mentioned, such as one-sided dame or “fight for the point of territory within the last ko” (which is called a “pass fight” in the KataGo page, and is directly linked to the Go Seigen disputes, as that point of territory is awarded by Chinese rules).

Indeed, in KataGo formalization the situation would not end as seki, as black fills dame as soon as the first phase starts and the ko rules turn into “local mode”, since by then filling the dames becomes completely safe and ko-threats irrelevant.

This is quite amusing because the concept of “enabling a new permanent stone in the locality of the group”, which is a key new feature of the Japanese 1989 rules that is necessary to explain various rulings (snapback, a common corner seki not being instead an alive-group-with-prisoners, some positions where playing a new stone exactly on top of the single live-stone is no possible, but it is possible to do so “near” that), is a concept that is not present at all in KataGo formalization, yet the theoretical best play result is identical for these positions.

3 Likes