0000
0000a
0000b
0000c
0000d1
0000d2
0000d3
0000d4
0000d5
Commentary on the Japanese 1989 Rules
2007-03-11; (c) Robert Jasiek
Contents
Preface
Tournament Rules
Error Fetching Procedures
Why Literal Application of the Rules Creates Nonsense
Commentary on the Official Commentary
Rule by Rule Reading
Revision of the Rules of Play
Limitations
Proposals for Future Revisions
Conclusion
Preface
This document teaches the correct interpretation of the Japanese 1989 Rules to the amateur and professional players and the referees of tournaments using those rules. Associations and federations shall be motivated to reflect how it could ever be responsible to use a ruleset of such extreme difficulty while countless simple, alternative rulesets are available.
Tournament Rules
General
The Japanese 1989 Rules are a ruleset of rules of play but also contain some tournament rules. The correct way to interpret the tournament rules depends on where the Japanese 1989 Rules are applied. If they are applied by professional Japanese players, then also the contained tournament rules are valid. If they are applied in EGF tournaments, then the contained tournament rules are not valid because they are overridden by the EGF's rulesets of tournament rules. Especially for this reason, it is good to recognize which are the tournament rules in the Japanese 1989 Rules:
April 10, 1989 (Effective May 15, 1989)
The Nihon Kiin and Kansai Kiin hereby revise the Nihon Kiin's Rules of Go formulated in October 1949 and establish the Japanese Rules of Go. These rules must be applied in a spirit of good sense and mutual trust between the players.
§1 The game of go
[Go is a game in which two players] compete in skill [on a board], from the beginning of the game until the game stops according to Article 9, [to see which can take more territory.] A "game" refers to the moves played until the "end of the game."
§10 Determining the result
3. If one player lodges an objection to the result, both players must reconfirm the result by, for example, replaying the game.
4. After both players have confirmed the result, the result cannot be changed under any circumstances.
§11 Resignation
During a game, a player may end the game by admitting defeat. This is called "resigning." The opponent is said to "win by resignation."
§13 Both players lose
2. If a stone on the board has been moved during the game and the game has proceeded, the game continues with the stone returned to its original point of play. If the players cannot agree, both players lose.
§14 Forfeit
Violation of the above rules causes immediate loss of the game, provided the result has not yet been confirmed by both players.
Competition in Skill
In §1, aspects of rules of play and tournament rules are mixed wildly. The square brackets in the citation indicate this to some extent. "compete in skill on a board" is restricted to a phase of the game: "from the beginning of the game until the game stops according to Article 9".
The official commentary on the Japanese 1989 Rules, part I, §9.2 (and playing practice in the real world sometimes confirms this, namely if dame and teire are not already filled until the game stop) says:
"1. Confirmation of the life and death of stones and territory requires that the players fill the dame and add any necessary stones inside their territory, in accordance with Article 8.
2. If the players agree, they may fill the dame and add other necessary stones after stopping the game, in which case these are not moves as defined by the rules, and need not be played according to the rules."
So while, after the game stop, stones shall be put on the board to fill the dame and to add any necessary stones inside their territory, the players do not compete in skill on the board during this. They shall do something on the board but it is not a competition. What is the relevance? Since a player can request resumption of a stopped game according to §9.3, competition may be restarted. Therefore the filling of dame, etc. after the game stop is in a context of competition that surrounds the island of no competition. To some extent, the stopped competition is a deception because it can be overridden by competition by means of resumption.
Besides the filling of dame, etc., there are other things to be done after the game stop, in particular the game end, determination of life, death, and territory, removals of dead stones, counting, etc. All this requires skill, something of this requires skill exercised on the board. However, according to §1, that skill shall not be competitive. If not competitive, then it must be cooperative. I.e., the players are supposed to agree and find the objective truth about the correct determination of life, death, and territory, etc. In this context, it makes sense that §9.2, which says "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.'", does not provide the option of the players' disagreement.
Error Fetching Procedures
General
There are two basic philosophies of dealing with errors when writing a computer program: Either do not include any errors and verify this or provide as many error fetching procedures as necessary for execution. The Japanese 1989 Rules are written like a program with error fetching procedures.
One might consider also the types of life and death or the ko rules to contain error fetching aspects but this would be a by far too simplistic view of rules to be discussed in detail later.
The tournament rules among the error fetching procedures are useful. Such or similar tournament rules should be contained in every reasonable tournament ruleset. Why have the authors of the Japanese 1989 Rules considered these tournament rules to be so important to integrate them within the ruleset of rules of play? Apparently they were afraid that error fetching is important because they feared that the Japanese 1989 Rules might contain many errors.
The two error fetching procedures Resumption and Effective Move are rules of play more than tournament rules. This gives them an appearance of being essential parts of the game. However, both are superfluous at best and create further errors by themselves.
Resumption
§9.3, the rule of resumption, is flawed: It is undefined what happens if both players request resumption simultaneously or if the players make an arbitrarily large number of resumptions. Due to the ambiguity of "move" and if the last board-play before the two passes creating a game stop was a ko capture, then it is unclear whether the player starting the resumed alternation may immediately recapture in the ko.
When a referee needs to overcome the flaws, one philosophy declares the game result to be a default tie for both players. Another philosophy tries to correct the rules: Simultaneous requests might be solved by giving the turn to the opponent of the player having passed last just before the game stop. Repeated requests (by the same player) to the same game stop position might be prohibited so that a player cannot use recurring resumptions with two passes in between to prevent the game from ending.
If the referee wants to fix the ko recapture problem, then he should do so according to the predominating view explained in the §6 interpretation: recapture is prohibited only during the board-play or pass immediately following the ko capture. If the scoring margin should be so big that the winner does not depend on the outcome of the ko, then the referee has the further alternative to declare a default win. In this case, it is hard to argue which of default tie or default win makes greater sense.
Needless to say, for the game to be resumed, the position and numbers of prisoners of the previous game stop are restored. The clock, which should have been stopped at the previous game stop, is restarted.
Resumption might occur due to different reasons: a) Before the previous game stop, one or both players made stupid strategic mistakes, i.e. overlooked some board-plays of value. b) During the filling of still unfilled dame, etc. after the previous game stop, accidental mistakes have occurred and the players do not want to correct this informally. c) A player wants to take advantage of the rule's gap that permits an arbitrary number of resumptions. d) A player wants to enable himself to recapture a ko and hopes that the referee applies the predominating view and does not declare a default result.
Since there can be several resumptions, there is not the one and only game stop but also several game stops can occur during a game.
It remains unclear why the rule of resumption violates the principle of alternation. It would have been better if the rule would always give the turn to the opponent of the player having passed last just before the previous game stop. §9.3 looks like a rule that was written too rashly after too little thought. In fact, omitting the rule entirely in a future revision seems like the best option.
Effective Move
§13.1 can apply only if one or both players have made the strategic mistake of passing too early before the game stop. E.g., the players might have overlooked some hard to see teire.
Apparently there is not a single player that likes or wants to apply this rule. However, the referee does not have a choice - he has to apply also this rule. Needless to say, a future revision of the rules ought to drop it.
§13.1 speaks of "the players [...] cannot agree" and thereby contradicts §1, which implies cooperation during the period when §13.1 may apply - after the game stop according to §9. This flaw is another reason to drop the rule in future.
Why Literal Application of the Rules Creates Nonsense
In this section, it is shown what would happen if the Japanese 1989 Rules were applied literally. This would lead to nonsense. Therefore the conclusion has to be: The Japanese 1989 Rules ought not to be applied literally. Later sections will then explain interpretations of the rules that make more sense.
Enable
General
"enable" in §7.1 is undefined and ambiguous. In Robert Jasiek's Pure Text Correction of the Japanese 1989 Rules, which - in terms both of precise wording and of success of approaching tradition - is the best of all made attempts of pure text corrections, the concept of enable is defined as follows:
[...] A player's final-string is "uncapturable-alive" if the opponent as the starting-player and the choosing-player cannot force capture of its stones.
A "permanent-stone" is a stone that is played during a hypothetical-sequence and then not removed during the rest of the hypothetical-sequence.
In the final-position, an intersection is "enabling" for a player as the starting-player if his opponent as the choosing-player can prevent the player's play of a permanent-stone on the intersection.
A string that is of a player, in the final-position, and not uncapturable-alive is "capturable-alive" if the opponent as the starting-player and the choosing-player cannot, with the same hypothetical-strategy, both force capture of the string and prevent the string player's play of a permanent-stone on an intersection that is enabling for the string's player in the final-position.
A final-string is "alive" if it is either uncapturable-alive or capturable-alive. [...]
However, such a definition of enabling is not good enough because it violates traditional scoring in positions like, e.g., example 0000.
Example 0000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Analysis question: Is the single-stone black string alive?
Substep 1: Is the single-stone black string uncapturable-alive?
The single-stone black string is not uncapturable-alive because hypothetical-sequences starting as in diagram 0000a force its capture.
Substep 2: Is the single-stone black string capturable-alive?
To answer this, firstly we need to determine Black's enabling intersections as in diagram 0000b (detailed reasons by means of move-sequences for the enabling status of each intersection are omitted here, except that White would rather pass than help Black to play a permanent-stone on the intersection of the single-stone black string while White can do nothing to prevent Black, who might start as in diagram 0000c, from playing a permanent-stone on either of the intersections marked by a triangle or 1.
Secondly, for the purpose of testing the application of the definition of capturable-alive, trivially we observe that White, after he captures the single-stone black string, can pass to prevent Black from playing a permanent-stone on any intersection that is enabling for Black and not the intersection of single-stone black string.
Thirdly, what about testing the application of the definition of capturable-alive, letting White capture the single-stone black string, and verifying whether White can prevent Black from playing a permanent-stone on the intersection of the single-stone black string, which is an enabling intersection for Black? The hypothetical-sequence (it continues a bit further than shown) in diagrams 0000d1-5 (Black plays 18 in an attempt of getting a permanent-stone on the enabling intersection of the single-stone black string) and similar hypothetical-sequences prove that White can prevent Black from playing a permanent-stone on the intersection of the single-stone black string.
Hence the single-stone black string is not capturable-alive, so it is not alive.
Example 0001
(This example presumes an understanding of the pass for a particular ko capture rule. If the reader is not confident, he might skip this example for the moment.)
The behaviour of this example is caused by the combination of enable and the pass for a particular ko capture rule in the Japanese 1989 Rules. [The topic was discovered by James Davies.]
|
|
|
|
3 = pass for the lower ko capture, 4 = pass for the upper ko capture |
7 = pass for the upper ko capture, 8 = pass for the lower ko capture |
|
4 = pass for the lower ko capture |
7 = pass for the upper ko capture, 9, 10 = pass |
|
4 = pass for the lower ko capture, 5 = pass for the upper ko capture |
8 = pass for the upper ko capture, 9 = pass for the lower ko capture |
The right side's white string is not uncapturable-alive.
Diagram 0001a shows the enabling intersections for White: Each intersection of the board is enabling for White. Most detailed explanations for this are omitted. The intersection 1 in diagram 0001b1 is enabling for White as is shown by diagrams 0001b1-2 and their beginning of the most interesting hypothetical-sequence, which recurs the hypothetical-moves 1 to 8 infinitely so that White does not get to play a permanent-stone on the intersection 1.
Now, for the purpose of testing for the right side's white string whether it is capturable-alive, Black needs to do both of the following: capture that particular string and prevent White from getting a permanent-stone anywhere on the board. In the variation shown in diagrams 0001c1-2, Black captures the right side's white string but White gets a permanent-stone on the intersection 6. With 7, Black may not recapture the upper ko yet because, due to the pass for a particular ko capture rule, Black first needs to make a pass for the upper ko capture. For another variation, its beginning is shown in diagrams 0001d1-2. Then the hypothetical-moves 2 to 9 recur forever. Black never gets a chance to capture the right side's white string while he is being busy with preventing White from getting a permanent-stone as a consequence of possibly winning the double ko.
As a consequence of both and similar variations, the right side's white string is capturable-alive, i.e., it is alive.
Conclusion
Since the examples contradict tradition severely, the concept of enable is a failure in itself. A solution with the concepts "capturable-1" and "capturable-2" of the Japanese 2003 Rules, version 35a is discussed later.
Dead Stones in Territory Do Not Exist
Citation
Here is an extract of the Japanese 1989 Rules:
§8: 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.
§10.1: [...] each player removes any opposing dead stones from his territory as is, and adds them to his prisoners.
Example 0002
|
|
|
|
|
|
Diagram 0002a shows the alive stones A and the dead stone D.
Diagram 0002b shows the eye point E and the dame D. According to the definition of "eye points", the empty point D is not an eye point because it is not (!) surrounded by the live stones of just one player - instead it is surrounded by a combination of three black live stones and of one dead white stone. Since D is not an eye point, it is a dame.
Diagram 0002c shows the stones which are alive but possess dame. The possessed dame is the empty point D in diagram 0002b. By definition of "[in] seki", the stones marked S are in seki.
Since "territory" is defined as the eye points surrounded by stones that are alive but not in seki, there is no territory in diagram 0002. The eye point E in diagram 0002b is surrounded by stones that are alive in seki.
Because the intersection D in diagram 0002a is not territory (not even an eye point because only empty points can be eye points), Black may not remove it due to §10.1 because dead stones might be removed only from territory.
Example 0003
This example is an even more obvious failure for the rules.
|
|
|
|
|
|
|
|
Diagram 0003a shows the alive stones A and the dead stones D.
Diagram 0003b shows the eye points E and the dame D. According to the definition of "eye points", the empty point D is not an eye point because it is not (!) surrounded by the live stones of just one player - instead it is surrounded by dead white stones. Since D is not an eye point, it is a dame.
Diagram 0003c shows the stones A which are alive and do not possess dame. Obviously, the dame D in diagram 0003b is not possessed by these stones. By definition of "[in] seki", the stones marked A are not in seki.
Since "territory" is defined as the eye points surrounded by stones that are alive but not in seki, the intersections marked B in diagram 0003d are Black's territory. The intersections occupied by white stones are not territory for two reasons: 1) They are not empty. 2) They are not eye points. The empty point in the middle is not territory because it is not an eye point but a dame.
Because the intersections occupied by white stones are not territory (not even eye points because only empty points can be eye points), Black may not remove the white stones due to §10.1 because dead stones might be removed only from territory.
Conclusion
Since the examples contradict tradition the most severely, the Japanese 1989 Rules' definition of "territory" is a failure in over 99% of all practically occurring scoring positions. To interpret the Japanese 1989 Rules correctly, we need concepts that model scoring tradition correctly: For a definition of "eye points", empty points and the intersections of a player's dead stones must be considered together to see if they are surrounded by the opponent's alive stones. This is done by the Japanese 2003 Rules and will be discussed later.
Possess
"possess" in
§8: [...] Stones which are alive but possess dame are said to be in "seki." [...]
is undefined and ambiguous. In example 0004, all stones are alive. Diagram 0004a shows the eye points E and the dame D. In diagram 0004b, it is clear that the stones marked S are in seki because it is obvious that they possess dame. However, the intention of the rules is unclear for the stones marked by question marks: Do they or do they not possess dame? This is relevant for determination of territory. A literal application of the Japanese 1989 Rules is not even possible for ordinary positions.
|
|
|
|
Purely Technical Sekis
The Japanese 1989 Rules define "in seki" by means of the property of being adjacent to a dame. This property is shared by all traditionally known so called sekis. However, the property is also shared by all so called independently alive groups that are adjacent to some still unfilled dame. As a consequence, whenever the players forget or intentionally omit the filling of a dame adjacent to independently alive groups, these are, according to the rules, also "in seki". If this state still persists during the scoring, then, due to the rules' definition of "territory", there is no territory inside such independently alive groups.
Nobody likes this property and the major author of the Japanese 1989 Rules, Sakai Takeshi, admits and regrets this feature of the rules. However, regardless of the general opinion and regardless of frequent breaking by some players of this rules feature, the property is being valid and does apply. Needless to say, the rules ought to be changed as soon as possible. Until then, the referees need to apply the rules as they currently are.
|
|
Example 0005 shows a position to be scored. The centre empty intersection is a dame. Therefore the black stones are in seki, the white stones are in seki, neither player has any territory, the score is 0, and the game is a tie. Before the game end, Black has made a strategic mistake: He should have filled the dame to win the game by 1 point.
As will be discussed later, defining "seki" and "independently alive" is more sophisticated than defining "in seki". So it is not a surprise that the Japanese 1989 Rules failed to define "seki" usefully.
Excess Prisoners
Here is a citation from the Japanese 1989 Rules:
§10.2: Prisoners are then filled into the opponent's territory, and the points of territory are counted and compared. The player with more territory wins. [...]
|
There are 3 white prisoner stones. |
There is 1 excess white prisoner stone. |
In example 0006, there are more prisoners of a colour than can be filled into territory of that colour's player. Diagram 0006a shows the prisoners that can be filled in. The rule now specifies that the [then remaining] points of territory are counted and compared: Black has 4 points of territory, White has 0 points of territory, so the count is 4 minus 0, what equals 4 in favour of Black. According to the rule, Black wins by 4 points. This contradicts tradition, which assumes Black to win by 5 points. The Japanese 1989 Rules fail to take excess prisoners into account. Such can even change the game's winner.
The No Result Rule
Citation
The no result rule of the Japanese 1989 Rules is:
When the same whole-board position is repeated during a game, if the players agree, the game ends without result.
Infinite Repetition
|
|
|
|
|
Diagrams 0007a1-2 show the first cycle. Then the players continue to repeat it forever. This is legal according to the rule since it requires the players' agreement to end the game without result. If the players simply don't agree or if they prefer to disagree, then the rule is useless.
Undecidable Strategy
|
|
The player having the turn might either create a so called triple ko and opt for the result No Result or create a so called double ko seki for the result tie. Since the results "No Result" and "tie" are uncomparable, the player cannot decide which strategy might be better. Strategy has become meaningless and been reduced to the quality of a lottery.
Commentary on the Official Commentary
General
This section comments on only part II of the official commentary on the Japanese 1989 Rules, which is called "Examples of Confirmation of Life and Death". One could comment on part I as well and would have to criticise it equally harshly.
All examples in part II fail to do the following:
Furthermore, the examples 5, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24a, 24b, 24c fail to verify life and death at all.
Concluding, the official commentary is a total failure, unless its intention is to pretend simplicity where there is infinite complexity.
Examples
Detailed Commentary on Example II.24a
Since it is a torture to apply the Japanese 1989 Rules as rules to an example position in writing, here such is exemplified for only one example, 24a of part II, of the official commentary, which writes:
"In this position, the black and white groups are both alive, but in seki because of the dame at A, so neither side has any territory. A move at A is needed to make Black's eyes into territory."
|
II.24a |
II.24a01 |
|
II.24a02 |
II.24a03 |
|
II.24a04a |
II.24a04b |
|
II.24a05 |
II.24a06 |
|
II.24a07 |
For the sake of simplicity, all move-sequences are to be understood as being hypothetical, the opponent of an analysed string starts, the players are assumed to alternate until two successive passes, game procedural aspects of the rules shall be ignored, all big strings shall be considered alive without studying move-sequences, presented move-sequences are meant to be representative for many other, similar move-sequences, and obvious failures like filling one's eye or playing on the outside shall be ignored. The official commentary should have specified what it means with "group" and used "eye points" instead of "eyes", which is not a term used by the Japanese 1989 Rules.
Is the squared white stone alive? Move-sequences like II.24a01 show that it cannot be captured. Therefore it is alive.
The triangled white stone can be captured, so we need to study enable for the possibility of capturable life. Move-sequences like II.24a02 or II.24a03 show that White can play a then uncapturable stone on either A or C. For the sake simplicity, we do not analyse whether White can play a then uncapturable stone on B. For the purpose of enabling, White's only hope is to get a stone on B or triangle when Black captures triangle. Move-sequences like II.24a04a-b show that White hopes in vain. So the triangled white stone does not have capturable life, either. It is dead.
Diagram II.24a05 summarizes which stones are alive A or dead D.
From that, in diagram II.24a06, we can now distinguish which of the empty points are eye points E or dame D. In particular, neither of the empty points A and B in diagram II.24a is an eye point because neither is surrounded by the live stones of just one player. B is surrounded by a combination of one alive white stone and one dead white stone. The empty point C in diagram II.24a is an eye point because it is surrounded by the live stones of just one player - White.
For the definition of "seki", we assume "stones" to mean "string". Here we shall ignore the ambiguity of "possess", trust the official commentary about the example II.24c, and apply it analogously, i.e., Black's two upper side strings shall be considered as one object. Then diagram II.24a07 shows the stones in seki because they are the alive stones possessing dame. Since this does not leave any eye points surrounded by stones that are alive but not in seki, there is no territory in the position. Hence the score is 0 (if there are no prisoners) and the game is a tie.
This example's discussion shows what the official commentary should have done at least for each of its examples.
(The following examples require the reader's understanding of the Japanese 1989 Rules' pass for a particular ko capture rule. Readers feeling uncertain should skip them for the time being.)
Example II.10
|
|
2 = pass for the ko capture |
|
5 = pass for the ko capture |
2, 3 = pass |
|
2, 4, 5 = pass |
2, 4, 5 = pass |
|
2 = pass |
4, 5 = pass |
|
2 = pass for the ko capture |
|
|
7 = pass |
|
|
14, 16, 18, 19 = pass |
For example II.10, the official commentary says that Black's eight stones were alive because of the move-sequence in diagrams II.10a1-2. The official commentary forgets to complete the move-sequence by successive passes and confuses the reader by speaking of passes where passes for the ko capture are meant. Apart from these minor faults, the official commentary is fundamentally wrong because it overlooks the most important move-sequence and the necessity to analyse each black string separately. The indicated move-sequence of the diagrams II.10a1-2 is good enough though to show that the big black string, which consists of seven stones, cannot be captured and therefore is alive. However, for the single-stone black string, one may not overlook the possibility of capturable life.
Before we can assess or reject capturable life, we need to study "enable". The move-sequence in diagram II.10b shows that Black can play an uncapturable stone on the intersection 1. The move-sequence in diagram II.10c shows that Black can play an uncapturable stone on the intersection 3. The move-sequence in diagram II.10d shows that Black can play an uncapturable stone on the intersection 3 in that diagram. The move-sequence in diagrams II.10e1-2 shows that Black can play an uncapturable stone on the intersection 3 in diagram II.10e2. Thus, when White captures the single-stone black string, Black's only chance to enable a new and then uncapturable stone of his is to place it under one of his original stones.
The move-sequence in diagrams II.10f1-5 is Black's attempt to get a new, then uncapturable stone on the intersection of the single-stone black string. White 3 and 5 are the counter-tesujis, which force Black to fill a so called eye with 10 in his continued attempt to enable a new stone where he could not play one if the initial black stone on that intersection had not been captured. Also Black's last attempt, the hypothetical play 12, is removed. So the single-stone black string in diagram II.10 does not have capturable life; it is dead according to a literal interpretation of the Japanese 1989 Rules.
The official commentary on the Japanese 1989 Rules fails to test for capturable life at all, overlooks the move-sequence in diagrams II.10f1-5, and contradicts a literal application of the rules, which determines the status dead instead of the officially claimed status alive for the single-stone black string.
Example II.16
(To save space, the official example II.16 is compressed to a smaller board here.)
The official commentary claims: "The ten white stones in the [upper] left corner are dead. The [...] white stones to the right also die through collapse of the seki."
Apart from the facts that the Japanese 1989 Rules do not know the terms "collapse" and "[the] seki" (only "[in] seki"), let us analyse whether it assesses the life and death statuses correctly according to a literal application of the Japanese 1989 Rules.
Obviously, with "the seki" the official commentary refers to the so called double ko shape on the right.
The Japanese 1989 Rules do not have any concept "collapse of a seki". So one can only conclude that, according to the official commentary and regardless of the apparent seki-like shape on the right, the white stones on the right are meant to be dead. Since analysis of an independent double ko coexistence (not shown here, but refer to the official commentary's example II.25) shows only the involved ko stones to be dead and the other involved strings to be alive, the reason for death of all white stones on the right must be found in the upper left.
The official commentary claims that all the ten white stones in the upper left were dead. Let us concentrate on the big upper white string. Black can try to capture it: One attempt is shown in diagrams II.16a1-2, which show the beginning of a move-sequence that afterwards repeats the hypothetical moves 2 to 9 forever. Another attempt is shown in diagrams II.16b1-2. Further similar attempts are possible. Black fails to capture the big upper white string, which thus is uncapturable, i.e. alive. The official commentary, which claimed dead, contradicts a literal application of the Japanese 1989 Rules.
This mistake in the official commentary of overlooking the two move-sequences might also explain its second mistake: to conclude that, due to the falsely assumed removability of all upper left white stones, the right white stones could be dead, too.
Example II.17
(To save space, the official example II.17 is compressed to a smaller board here.)
The official commentary claims: "The three black stones in the corner are alive. The [...] white stones surrounding them are dead. The [...] white stones to the right also die through collapse of the seki."
Again let us ignore the minor problems related to "collapse" and "[the] seki" but let us analyse whether the official commentary assesses the life and death statuses correctly according to a literal application of the Japanese 1989 Rules.
|
|
4 = pass for recapturing the stone 3 5 = pass for recapturing the stone 2 |
|
8 = pass for recapturing the stone 7 9 = pass for recapturing the stone 6 |
4 = pass for recapturing the stone 3 |
|
9 = pass for recapturing the stone 2 11, 12 = pass |
4 = pass for recapturing the stone 3 |
|
11, 12 = pass |
Can Black capture the big white string? One attempt is shown in diagrams II.17a1-2, which show the beginning of a move-sequence that afterwards repeats the hypothetical moves 2 to 9 forever. Other attempts are shown in diagrams II.17b1-2 and II.17c1-2. Further similar attempts are possible. Black fails to capture the big white string, which thus is uncapturable, i.e. alive. The official commentary, which claimed dead, contradicts a literal application of the Japanese 1989 Rules.
Other Official Examples
There are severe mistakes also in other examples of the official commentary on the Japanese 1989 Rules. E.g., compare example 0001 with the official example II.11and notice that the official commentary does not discuss some interesting other move-sequences for testing capturable life. In general, usage of the official commentary cannot be recommended. It is misleading rather than helpful.
Rule by Rule Reading
Preliminary Notes
The reference text is the standard translation by James Davies and cited from The Go Player's Almanac. It is the generally available English version of the Japanese 1989 Rules.
The commentary concentrates on rules of play. Tournament rules are indicated but not discussed in detail.
Preamble - Original Text
The Japanese Rules of Go
April 10, 1989 (Effective May 15, 1989)
The Nihon Kiin and Kansai Kiin hereby revise the Nihon Kiin's Rules of Go formulated in October 1949 and establish the Japanese Rules of Go. These rules must be applied in a spirit of good sense and mutual trust between the players.
Preamble - Interpretation
The preamble consists of tournament rules.
§1 - Original Text
Article 1. The game of go
Go is a game in which two players compete in skill on a board, from the beginning of the game until the game stops according to Article 9, to see which can take more territory. A "game" refers to the moves played until the "end of the game."
§1 - Splitting Tournament Rules and Rules of Play
Tournament Rules
The players compete in skill. A "game" refers to the moves played until the "end of the game."
Rules of Play
Go is a game in which two players play on a board, from the beginning of the game until the game stops according to Article 9, to see which can take more territory.
§1 - Interpretation of the Rules of Play
The game aim "more territory" is incomplete: Prisoners of opposing colour are missing. Besides the game aim is also indicated by §10.2. It is easier to remove double contents than to prolong it by completing it twice, i.e., the game aim can as well be dropped from §1.
In the rules, "game stop" is used as a term. Using it also with a verb as in "the game stops" complicates reading of the rules unnecessarily. Since, according to §9.3, resumption is possible, normally "game stop" in the rules should be used together with the indefinite article. This suggests "until a game stop" to be better for the rule text.
After splitting tournament rules from rules of play, the restriction "from the beginning of the game until the game stops according to Article 9" of the clause "Go is a game in which two players play on a board" has become superfluous because even after a game stop actions are being done on a board.
Thus, in an interpretative rule text, both the restriction and the game aim can be omitted without losing essential information.
§1 - Interpretative Text
§1 The game of go
Go is a game in which two players play on a board.
§2 - Original Text
Article 2. Play
The players can alternately play one move at a time, one player playing the black stones, his opponent the white stones.
§2 - Interpretation
To begin with the obvious, the rule fails to specify the starting player of alternating play. As world-wide playing practice dictates for even games, this is the player playing the black stones.
The phrases "the player playing the black stones" and "the player playing the white stones" are impractical. For later usage in the rules and for general usage, they should be replaced by the terms "Black" and "White".
The verb "can" expresses a not prescriptive possibility and thereby introduces ambiguity unnecessarily. The rule should not suggest the reader that the players can but need not alternate. Clearer style would be an improvement.
A weakness of the Japanese 1989 Rules throughout its text is the ambiguity of the words "play" and "move". Does either or both of them include passes? From §3, we know about the existence of what should be called board-plays. §9.1 informs about the existence of passes and their possible occurrence in alternation. However, the relation between "board-play", "pass" and ""play", "move" remains unclear.
§2 - Interpretative Text
§2 Alternation
The players alternate one move at a time. A "move" is either a board-play or a pass. A "board-play" is placing one's own stone on an empty intersection. One player, called "Black," uses the black stones - his opponent, called "White," the white stones. Black starts.
§3 - Original Text
Article 3. Point of play
The board is a grid of 19 horizontal and 19 vertical lines forming 361 intersections. A stone can be played on any unoccupied intersection (called an "empty point") on which Article 4 permits it to exist. The point on which a stone is played is called its "point of play."
§3 - Interpretation
The selective reference to the restriction of §4 is confusing because there are also other restrictions in other articles, especially the §6 ko restriction and the §9.1 game stop. To avoid confusion, the reference to restrictions should be generalized.
The Japanese 1989 Rules' usage of "empty point" is unfortunate because "point" is also used for very different meanings in "eye point" and "point of territory". Therefore it is better to speak of "empty intersection", and the rule's first sentence encourages a continued use of the word "intersection".
Although "point of play" is used in §4, it is essentially a superfluous term. Terms that can be omitted make rules interpretation easier when less text has to be studied.
One would like the interpretative text to include the sentence "In a board-play, one stone is played on any empty intersection, subject to the other paragraphs.", however, the interpretative text of §2 already includes a definition of "board-play". The phrase "subject to the other paragraphs" seems redundant; its omission does not make the other paragraphs invalid. Thus the remaining interpretative text for §3 is surprisingly short. Continuing the interpretation process beyond what is shown now, one might wish that to become sentence 2 of the interpretative text of §1.
§3 - Interpretative Text
§3 The board
The board is a grid of 19 horizontal and 19 vertical lines forming 361 intersections.
§4 - Original Text
Article 4. Stones that may exist on the board
After a move is completed, a group of one or more stones belonging to one player exists on its points of play on the board as long as it has a horizontally or vertically adjacent empty point, called a "liberty." No group of stones without a liberty can exist on the board.
§4 - Interpretation
Usage of the word "group" in the Japanese 1989 Rules or its standard translation is unfortunate because for "possess" and "in seki" in §8 one would like to define "group" with a different meaning. For §4, the most frequent words in English for the concept in question are "string" and "chain". Although they do not mean the same as in common English, it is easy enough to understand that, for the game of Go, these words get a new, independent meaning. It should also be defined explicitly in the rules.
As has already been discussed, "move", "point of play", and "empty point" should be replaced by clearer terms. Furthermore, the imprecise modal verbal "can" shall be replaced by the clearer "may".
§4 tries to specify a no suicide rule and to prepare §5. The result is awkward and hard to understand. In a sense, §4 and §5 contradict each other because the stones that §5 wants to capture were already prohibited to exist in §4, which prohibits stones of either colour to exist on the board if they are without liberty. The interpretative text separates definitions of "string" and "liberty" from the no suicide rule and moves the latter to where it belongs - §5A.
§4 - Interpretative Text
§4 String and liberty
A "string" is a stone and all stones of the same colour continuously connected to it via grid lines. A "liberty" of a string is an empty intersection horizontally or vertically adjacent to it.
§5A No suicide
A player may not make a board-play if, after it and its captures, the player's string would be without liberty.
§5 - Original Text
Article 5. Capture
If, due to a player's move, one or more of his opponent's stones cannot exist on the board according to the preceding article, the player must remove all these opposing stones, which are called "prisoners." In this case, the move is completed when the stones have been removed.
§5 - Interpretation
There is little to say beyond what is said about §4, except that "liberty" should be applied. The "completed" is applied only in the original §4, so it can be omitted. It might also be useful for the §12 no result rule's specification of whole-board position, but this does not refer to "completed".
§5 - Interpretative Text
§5 Capture
The player making a board-play removes all stones of the opponent's strings without liberty. They are called "prisoners."
§6 - Original Text
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.
§6 - Interpretation
There is a general agreement that the translation is imprecise for the modal verb "can". It should be "could" because capture and recapture are not actually being done but only considered as an imagined, principle possibility for defining what a ko is.
For the basic ko rule, a definition of the shape "ko" would not be needed, but maybe the rules authors considered it helpful for the reader. The interpretative text introduces yet another term that is not needed for the §6 ko rule, however, it will be useful for the §7.2 ko rule. Since it can make also the §6 ko rule's wording clearer, it is already defined in the interpretative text of §6.
In the Japanese 1989 Rules, "move" is ambiguous: One cannot be absolutely sure that passes are not included. This is a flaw of the original rules: After a player's capture in a ko, two passes, and the player's request of resumption according to §9.3, it is unclear whether the opponent may then immediately resume alternation by recapturing in that ko. The predominating opinion considers it more likely that the Japanese 1989 Rules want the §6 ko rule to be a prohibition of an immediate recapture during the opponent's next turn only. (The Japanese 2003 Rules, version 35a's basic ko rule has a different wording but means the same as the interpretative text's §6 ko rule. For the Japanese 2003 Rules, testing has been extensive.) Therefore intervening passes lift the prohibition to recapture. In informal terms, passes serve as so called ko threats. When the interpretative text speaks of "on the next move", then this is unambiguous because the interpretative text has already defined "move" earlier.
The scope of application of the §6 ko rule is ambiguous because the §7.2 ko suggests that ko rules can have a restricted validity. In theory, it is not necessary to specify a restricted scope of application for the §6 ko rule because its application after a game stop does not alter the §7.2 ko rule's consequences at all. However, this also means that it is superfluous to apply both ko rules after a game stop. Therefore the intended scope shall be specified explicitly. This makes application of the ko rules after a game stop more practical in the interpretative text.
§6 - Interpretative Text
§6 Ko
A shape in which the players could alternately capture and recapture one opposing stone is called a "ko." A "ko-capture" is a play that is in a ko and removes exactly one stone. If, in a ko, a player's stone has been removed in a ko-capture, then he may not make a ko-capture in that ko on the next move. This rule applies only until the first game stop or in between a resumption and the next game stop following the resumption.
§§7-10 Preface
Go is advertised as a game with simple rules. §§1-6 have been simple enough. The continued moderate text length of the §§7-10 creates the first impression that also they would be simple. - Hardly anything could be a greater pretence. Over a decade of Japanese 1989 Rules research has been necessary before this commentary could be written.
The reader should take a deep breath. The real study of the Japanese 1989 Rules starts only now. Their core problem is the definition of life, death, and territory for the purpose of scoring.
Area Scoring allows, if the players disagree, to resolve all questions about removals by actual play. Then only two types of stones remain on the board: black stones and white stones. The types equal the stones' colours and hence are visible. Rules theorists have suggested types of Modern Territory Scoring that also achieve this.
However, Japanese, Korean, or Japanese style go server rules still use Traditional Territory Scoring. There questions about removals cannot be resolved by actual play because that would affect the score. Instead abstract definitions define six types of stones on the board: black stones alive not in seki, black stones alive in seki, dead black stones, white stones alive not in seki, white stones alive in seki, dead white stones. These types cannot be distinguished only visually from the stones' physical properties. Additional hypothetical analysis is required to identify each stone's correct type.
§7.1 - Original Text
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."
§7.1 - Where are the Two Eyes?
Ask a go player: "What is alive?" The most frequent answer is: "Something is alive if it has two eyes!" "But what about sekis?" "Ah, sorry. Forgot about that! Of course, there are also sekis!"
Why then do the Japanese 1989 Rules not define "independent life" via two eyes? Because the Japanese 1949 Rules and the World Amateur Go Championship 1979 Rules had made attempts in that direction and failed terribly. Only recently rules theorists have defined two-eye-alive usefully and shown its relation to the Japanese 2003 Rules, version 35a approach via uncapturable, capturable-1, and capturable-2, which completes the Japanese 1989 Rules' rough approach with uncapturable and capturable life.
Neither the rules theorists' research nor strategic wisdom replace application of the Japanese 1989 Rules as rules. In other words, for the purpose of applying them, it is immaterial whether something has two eyes.
§7.1 - Hypothetical Nature
Since playing out status questions on the board would cost points when one fills one's own so called territory to approach the liberties of the opposing stones to be removed, doing so is essentially impossible (or has to be accompanied by additional provisions about restoring the position and numbers of prisoners). In practice, it is not done, except by some beginners or their teachers. Predominating practice suggests that status questions are clarified by hypothetical means.
The official rule suggests the hypothetical nature implicitly, but one must read it carefully to notice that: In the first part of the first sentence, the nature is indicated by the combination of the conjunction "if" and the modal verb "can" in its negated form. In the second part of the first sentence, the nature is indicated by the combination of the conjunction "if" and the modal verb "could" in its negated context. This second part exhibits the hypothetical nature more clearly because the verb form expresses conditional, speculative possibility.
However, rules ought not to be a quiz for their readers. The hypothetical nature of the analysis as a whole and of the moves in particular should be spelled out explicitly. This is done by the interpretative text, especially when it defines "hypothetical-move" and "hypothetical-sequence".
§7.1 - Alternation in Hypothetical-Sequences
Since it is not clear that §2 applies also to hypothetical-sequences due to §7, there should be an explicit statement in the rules. The interpretative text includes such.
For hypothetical-sequences due to §7, the available types of hypothetical-moves are not specified. Thereby in particular the difference between pass and ko-pass is ambiguous. Obviously there must be imagined board-plays with only virtually removed stones, which do not physically become prisoners. What about imagined passes? They are also required a) to allow alternation when a player does not want to and cannot make a board-play and b) to know when a finite hypothetical-sequence has ended. Unlike the regular alternation, which allows board-plays and passes, hypothetical-sequences also need to allow a third type of hypothetical-moves: imagined ko-passes. They are necessary since §7.2 describes their existence and usage. While the Japanese 1989 Rules fail to specify all that, the interpretative text includes it.
The official rules also fail to specify the starting player of a hypothetical-sequence. Suppose that stones (as will be seen later, it should be more specifically "a string") of a particular player are to be analysed for their life and death status. Should then hypothetical-sequences analysing this start with the player, his opponent, or may they start with either of them? For questions like "Can it be captured?", the worst case is that the opponent may do the first hypothetical-move. So it is sufficient to assume and require exactly this. All examples of the official commentary agree implicitly on this feature. For clarifying whether, for the purpose of "enable", some uncapturable stone can be played in the first place, the starting player requirement is less clear; luckily we have seen that the concept "enable" should not be used at all in a good interpretation of the Japanese 1989 Rules, so we need not worry about the starting player for that purpose any longer. The interpretative text specifies the starting player of a hypothetical-sequence: It is the opponent of a string being currently analysed.
Where there are imagined passes, it must also be specified what the effect of successive imagined passes is. The original rules fail to specify that - the interpretative text specifies it.
Then it is necessary to consider whether ko-passes behave like passes for the purpose of ending a hypothetical-sequence. The answer is easy: no. This is so because otherwise in shapes like a so called double ko seki the players' ko-passes would not make any sense; the hypothetical-sequence would be ended before the players could make use of their ko-passes by then performing the threatened ko recaptures. Likewise it must be clear from the rules what the effect of mixed successions of ko-passes and passes is. The natural solution is: only two regular (imagined) passes in succession end a hypothetical-sequence. The original rules fail to specify these aspects - the interpretative text specifies them.
Apart from the extra question of whether the §12 ko rule applies to hypothetical-sequences, the ko rule(s) still allow hypothetical-sequences of infinite length. The Japanese 1989 Rules do not explain how they are to be interpreted. This is explained by the interpretative text.
To summarize, a lot of basic but necessary aspects are missing in the Japanese 1989 Rules. Also their specification in the interpretative text make it significantly longer than the Japanese 1989 Rules. Their §7 consists much more of gaps than of contents.
The interpretative text about aspects in hypothetical-sequences is derived from the Japanese 2003 Rules, version 35a because that ruleset has been successfully tested to be the ultimate model of current professional Japanese rules tradition.
§7.1 - Recursion?
The Japanese 1989 Rules' definition text of "alive" is ambiguous as to whether it is meant recursively or non-recursively. In the interpretative text's approach of using capturable-1 and capturable-2, this is not a problem any longer because neither of these terms refers to itself in its definition.
§7.1 - Multiple Threats
The "stones" in the Japanese 1989 Rules' text "Stones are said to be 'alive' if they cannot be captured by the opponent [...]" is ambiguous. E.g., if a string A is uncapturable and if a string B is uncapturable, then this does not always imply that the strings A and B are uncapturable at the same time. In other words, the official rules fail to specify whether "stones" refers to one particular string's stones or to some maximal set of stones. Since example II.25 of the official commentary on the Japanese 1989 Rules (with a so called double ko seki) says the big strings to be alive but the single ko stones to be dead, the intention of the rules authors appears to have been a string by string assessment of life and death. The interpretative text speaks of "string". - By the way, for Japanese style rules one should not consider different life and death statuses for different stones of the same string, although the "stones" of the original rules text is so ambiguous that it would not even exclude such an approach.
Example 0009
|
|
|
After Black 1, White can defend at either A or B but may not play on both intersections simultaneously. If uncapturable life is defined on a string by string basis, then White can defend the left string when this is being analysed and attacked by Black and White defend the right string when this is being analysed and attacked by Black. If uncapturable life is defined by a maximal property that requires White to defend both strings at the same time, then all White stones in the example are not uncapturable. Under the interpretative text, each string is analysed for itself and so each white string has life of the type uncapturable.
§7.1 - Uncapturable Life
Forgoing the details of "force", the relatively easy parts of the interpretative rules about life are:
A player's string is "uncapturable" if the opponent cannot force capture of its stones. [...] A string is "alive" if it is either uncapturable, capturable-1, or capturable-2.
Uncapturable is not explained here again. Examples can be found elsewhere. Also the definition of "alive" is straightforward, however, one should notice that there are three types of life.
§7.1 - Overview on Locality and Capturable Life
Since some examples for "enable" contradict tradition severely, the concept of enable is a failure in itself. To interpret the Japanese 1989 Rules correctly, we need concepts that do model scoring tradition correctly. Such are the concepts "capturable-1" and "capturable-2" of the Japanese 2003 Rules, version 35a. Japanese rules tradition claims that, during the scoring, life and death were by their nature local but has lacked the ability to define the locality. Now the Japanese 2003 Rules, version 35a do define it. The relevant definitions are integrated into the interpretative rules, except that here "force" is not defined and precise references to "hypothetical-strategy" are omitted. The interested reader might read the Japanese 2003 Rules and its commentary to learn more about such details. He should read that commentary also for more explanation and examples of "local-1", "capturable-1", "local-2", and "capturable-2".
local-1 means "under the string". (More precisely, "on one of the string's intersections in its appearance in the analysed scoring position", i.e. before the string might possibly grow during a hypothetical-sequence. For this reason, calling "local-1" "under the string" in the interpretative rules would be suggestive but also somewhat careless.) Capturable-1 strings occur mostly in nakade shapes or snapbacks. local-2 defines the local environment of a player's string and is surrounded by stones of uncapturable or capturable-1 strings of that player. Capturable-2 strings are scarce and occur especially in some minimal life shapes, where playing under the stones would fill one's second eye.
Some readers might have a prejudice against the unfamiliar terms. They should overcome it. It cannot be helped that the Japanese 1989 Rules' approach is a distinction of uncapturable life and capturable life instead of an identification of two-eye life. It should not be hard for players 8 kyu or stronger to understand capturable-1 quickly in practice, although the formal definition may be more difficult. (Beginners will have much greater problems because they might have seen only a few snapbacks and advanced nakade shapes at all. Despite persistent claims of some of their proponents, the Japanese 1989 Rules or their careful interpretations are not suitable for beginners at all.) Understanding local-2 and capturable-2 requires a much greater effort because the local environment is a rather unintuitive object and because the few existing shape classes with capturable-2 strings are so scarce that most players would not have encountered them yet. The reader is asked to learn such shapes while reading on. Only the basic types of patterns with capturable-2 strings are shown here.
If you are unfamiliar with specialized English terms that consist of a word and a number, here is a guideline of pronunciation: Speak "local one", "local two", "capturable one", and "capturable two". A term like "local-1" is not more peculiar than a specific colour like "blue-green". Since there is not just one type of capturable but there are two types of capturable, it is necessary to distinguish them explicitly as "capturable-1" and "capturable-2". This is more convenient than to say "type 1 of capturable" and "type 2 of capturable" all the time.
All the terms are applied to the scoring position. Before, in particular during the alternation before the first game stop, the life and death status of every string can change, e.g., when a string is sacrificed. One must not confuse informal or strategic talk about life and death with rules terms about life and death at the game end.
Capturable-1 and capturable-2 refer to the term permanent-stone:
A "permanent-stone" is a stone that is played during a hypothetical-sequence and then not removed during the rest of the hypothetical-sequence.
Therefore, if a player needs to get a permanent-stone, then it is not enough to play that stone but it has to remain on the board until the end of that hypothetical-sequence during which it has been played. (If the hypothetical-sequence is infinite and for that reason without an end, then the stone has to remain on the board for the infinite rest of the hypothetical-sequence.) Sometimes it happens that a stone is played during a hypothetical-sequence, then removed, and that even later another stone is played on the same intersection. Then it matters what happens to latter: Does it remain on the board until the end of the hypothetical-sequence?
§7.1 - Local-1
The interpretative text defines:
For a string, "local-1" is all the string's intersections.
Although this text is not as precise as in the Japanese 2003 Rules, it should be understood that the string's intersections are those in the scoring position and not those of a possibly grown string during a currently considered hypothetical-sequence. In the following examples 0010 to 0012, a particular string's local-1 intersections are marked by "1".
|
|
The local-1 of the small white string. |
|
|
The local-1 of the upper black string. |
|
|
The local-1 of the upper black string. |
§7.1 - Capturable-1
The interpretative text defines:
A player's string is "capturable-1" if the opponent cannot force both capture of the string's stones and no local-1 permanent-stone of the player. [...] An uncapturable string is neither capturable-1 nor capturable-2.
Before one studies whether a string is capturable-1, one first studies whether it is uncapturable. Since, by the rule, an uncapturable string is not capturable-1, it has become superfluous to study capturable-1 after one has already determined a string to be uncapturable. In the examples, it is too easy to see that the studied strings are not uncapturable; therefore the analysis for this is not also shown here. Furthermore, each of the other strings in the examples should be studied for their life and death status as well; to keep this document's length reasonable, also that is not discussed here. We concentrate only on likely capturable-1 strings.
For the purpose of testing if the definition of capturable-1 applies to a particular string, the opponent has to capture the string or at least make every serious attempt to capture it. The string's player wants the opposite: avoid the string's capture, if possible. Furthermore, if the opponent succeeds in capturing the string, then he has to achieve or seriously attempt to achieve more: If anyhow possible, he has to prevent the player from getting a permanent-stone on one of the original string's intersections, i.e., in its local-1. Contrarily, the string's player has to opposite objective: to establish a permanent-stone in the string's local-1.
Example 0010 (continued)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This is a typical so called nakade shape. Is the small white string capturable-1? In the hypothetical-sequence of diagrams 0010b1-4, Black 3 captures the string but White 10 plays a permanent-stone on one of the original string's local-1 intersections. In the hypothetical-sequence of diagrams 0010c1-5, Black's attempt is a bit more ferocious. Black 3 captures White's original small string, White 10 plays a permanent-stone on one of the original string's local-1 intersections, and after the hypothetical-move 16, Black finally gives up his attempt when the stone 10 is already part of a - what some other rulesets call - pass-alive string. Similar hypothetical-sequences are possible. Whatever Black tries, he cannot prevent White from getting a local-1 permanent-stone after the original small white string's capture. Hence that string is capturable-1, i.e., alive.
Example 0011 (continued)
|
|
|
|
|
|
|
|
|
This is a typical so called snapback. Is the upper black string capturable-1? In the hypothetical-sequence of diagrams 0011b1-3, White 1 captures the string but Black 2 plays a permanent-stone on the original string's local-1 intersection. White's exotic attacks in the hypothetical-sequences of diagrams 0011c1-2 and 0011d fail even to capture the upper black string at all; in an application of the definition of capturable-1, such weak attacks favour the defender of the string's currently claimed capturable-1 status. Other hypothetical-sequences are possible but not shown here. To summarize, the upper black string is capturable-1 and therefore alive.
Example 0012 (continued)
|
|
|
|
|
|
|
|
|
This is another so called snapback. Is the upper black string capturable-1? In the hypothetical-sequence of diagrams 0012b1-4, White 1 captures the string but Black 4 plays a permanent-stone on the original string's local-1 intersection. If, as in the hypothetical-sequence of diagrams 0012c1-2, White makes a hypothetical pass earlier, then also this is in vain: Now Black 2 plays the local-1 permanent-stone. The upper black string is capturable-1 and alive.
§7.1 - Local-2
The interpretative text defines:
For a player's string, "local-2" is local-1 and, recursively, any adjacent intersection without a stone of a string that is of the player and either uncapturable or capturable-1.
A string's local-2 is a local environment that includes the string's intersections and some (or even many) further intersections in its neighbourhood. The local environment reaches as far as and is surrounded by such stones of the string's player that are either uncapturable or capturable-1. Therefore first one has to determine all the uncapturable and capturable-1 strings before one can determine a string's local-2 (or then test whether it is capturable-2).
The term local-2 has not been among the traditional go terms because until recently nobody has understood the Japanese style rules' concept of locality. Now go theory is advancing and this concept of locality has been discovered and first defined in the newest versions of the Japanese 2003 Rules.
In the following examples, a particular string's local-2 intersections are marked by "2". Uncapturable strings are marked by "U", capturable-1 strings are marked by "1", if there are any (actually, in these examples there are none). The "U" or "1" marks are shown only for the stones of the particular string's player because only uncapturable or capturable-1 stones of his restrict the local-2.
0000f
The local-2 of the small black string.
II.10
(continued)
II.10g
That the big black string is uncapturable has already been seen earlier in diagrams II.10a1-2. That the small black string is not capturable-1 follows from hypothetical-sequences like in diagrams II.10f1-5.
II.10h
The local-2 of the small black string.
II.2
II.2a
Neither of the small black strings is capturable-1: If White captures either, Black cannot get a permanent-stone on its intersection.
II.2b
This is the local-2 of the upper small black string. It is also the local-2 of the left small black string. By coincidence, the two small black strings' local-2 environments are identical.
|
(continued) |
The small black string is not capturable-1 due to hypothetical-sequences like in diagrams 0000d1-5. See the discussion of "enable". |
|
The local-2 of the small black string. |
(continued) |
|
That the big black string is uncapturable has already been seen earlier in diagrams II.10a1-2. That the small black string is not capturable-1 follows from hypothetical-sequences like in diagrams II.10f1-5. |
The local-2 of the small black string. |
|
|
Neither of the small black strings is capturable-1: If White captures either, Black cannot get a permanent-stone on its intersection. |
|
This is the local-2 of the upper small black string. It is also the local-2 of the left small black string. By coincidence, the two small black strings' local-2 environments are identical. |
§7.1 - Capturable-2
The interpretative text defines:
A player's string is "capturable-2" if the opponent cannot force both capture of the string's stones and no local-2 permanent-stone of the player. An uncapturable string is neither capturable-1 nor capturable-2. A capturable-1 string is not capturable-2.
Before one studies whether a string is capturable-2, one first studies whether it is uncapturable, then whether it is capturable-1. Since, by the rules, neither an uncapturable string nor a capturable-1 string is capturable-2 (each string has only one type), it has become superfluous to study capturable-2 after one has already determined a string to be either uncapturable or capturable-1. In the examples, we concentrate only on likely capturable-2 strings.
For the purpose of testing if the definition of capturable-2 applies to a particular string, the opponent has to capture the string or at least make every serious attempt to capture it. The string's player wants the opposite: avoid the string's capture, if possible. Furthermore, if the opponent succeeds in capturing the string, then he has to achieve or seriously attempt to achieve more: If anyhow possible, he has to prevent the player from getting a permanent-stone on an intersection of the original string's local-2 environment. Contrarily, the string's player has to opposite objective: to establish a permanent-stone in the string's local-2. Note: During the same hypothetical-sequence, it does not make a difference for the definition whether a permanent-stone is played before or after the original string's capture.
Example 0000 (continued)
|
|
|
|
|
Is the small black string capturable-2? In the hypothetical-sequence of diagrams 0000g1-3, White 1 captures the string but Black 2 plays a permanent-stone on one of the original string's local-2 intersections. White can attack the stone 2 harshly but all Black needs to do is prevent its capture. It does not hurt that Black 8 and 10 are more local-2 permanent-stones. He needs to get at least one. Since he does (and in similar hypothetical-sequences also would), the small black string is capturable-2, i.e. alive.
Maybe the reader wonders why Black is not in a hurry to play on the original small black string's intersection. He does not have to. Playing in the local-2 for the purpose of applying the capturable-2 definition is easier than playing in the local-1 for the purpose of applying the capturable-1 definition because the local-2 comprises more intersections than the local-1. For clarifying whether the small black string is capturable-2, Black may establish his new permanent-stone on any of the local-2 intersections. With many nearby black uncapturable stones, this is a very easy task. The more difficult thing has been to verify exactly which intersections are in the local-2.
Example II.10 (continued)
|
2 = ko-pass |
5 = ko-pass |
|
9, 10 = pass |
2 = ko-pass |
|
|
7, 8 = pass |
Is the small black string capturable-2? In the hypothetical-sequence of diagrams II.10i1-3, White 1 captures the string but Black 4, 6, and 8 play permanent-stones on the intersections of original string's local-2. In the hypothetical-sequence of diagrams II.10j1-3, White 1 captures the string but Black 4 and 6 play permanent-stones on the intersections of the original string's local-2. Although White has different hypothetical strategies available for attacking, Black can refute each of them: for each, Black gets at least one local-2 permanent-stone. Hence the small black string is capturable-2 and alive.
Example II.2 (continued)
|
|
|
Is the upper black string capturable-2? In the hypothetical-sequence of diagrams II.2c1-2, White 3 captures the string but Black 2, 4, and 6 play permanent-stones on the intersections of original string's local-2. Therefore the upper black string is capturable-2 and alive.
Is the left black string capturable-2? In the hypothetical-sequence of diagrams II.2c1-2, White 3 captures the string but Black 2, 4, and 6 play permanent-stones on the intersections of original string's local-2. Therefore the left black string is capturable-2 and alive.
By coincidence, the considered hypothetical-sequences are the same for assessing the capturable-2 status of the upper black string and for assessing the capturable-2 status of the left black string. However, according to the definition of "capturable-2", each string has to be analysed for itself.
§7.1 - Dead
Once "alive" has been defined, a definition of "dead" can be trivial:
A string is "dead" unless it is alive.
Players and referees can fall into the trap of thinking: The definition is so simplistic that one does not need to study its application in practice carefully. This, however, would be a severe mistake. To know that something is dead, one has to exclude the possibilities of uncapturable, capturable-1, and capturable-2 life. I.e., one actually has to test each of the three terms' definitions. Especially capturable-2 is not excluded easily; we have seen how relatively easy it can be to establish a local-2 permanent-stone. Where one would want to see death, one first needs to work hard and prove in particular that capturable-2 life is indeed not possible. Things are not quite as tough though because capturable-2 has undergone extensive testing under the Japanese 2003 Rules, version 35a and agreement to current professional Japanese tradition has been verified. So unexpected surprises are not to be expected from the derived interpretative rules themselves - only from possibly their too careless application.
Example 0001 (continued)
|
The uncapturable strings. |
The local-2 of the small black string. |
|
The local-2 of the white ko stone string. |
The local-2 of the right white string. |
|
|
Summary: the dead strings D. |
By using infinite hypothetical-sequences like those in diagrams 0001k1 to 0001n4 and 0001p1 to 0001s4 studied further below, we can conclude the uncapturable strings (diagram 0001e) and that neither of the ko stones is capturable-1.
Is the small black string capturable-2? Its local-2 (diagram 0001f) is such that the infinite hypothetical-sequences in diagrams 0001k1 to 0001n4 can be used. The small black string is not capturable-2; it is dead. (To be more precise: Using one of the hypothetical-sequences consists of ko-captures and ko-passes continuing forever. Black never captures the big white string. Each ko stone Black plays is soon captured again; none of them remains as a permanent-stone. Playing on the right side would be meaningless because that is outside the local-2 and Black does not threaten the big white string by playing such tenukis.)
Is the white ko stone string capturable-2? Its local-2 (diagram 0001g) is such that the infinite hypothetical-sequences (now starting by Black) in diagrams 0001p1 to 0001s4 can be used. The white ko stone string is not capturable-2; it is dead.
The maybe most exciting question is whether the right white string is capturable-2. Its local-2 is shown in diagram 0001h. The particularly noteworthy aspect here is that none of the so called double ko seki's intersections is in the local-2. This has consequences for the hypothetical-sequences. Already the hypothetical-sequence in diagram 0001i is convincing enough: Black 3 captures the right white string. White 2 and 4 play permanent-stones but they are not in the local-2 of the right white string. So they do not suffice the definition of capturable-2. White does not get any chance to play a permanent-stone in the local-2 of the right white string. Hence it is not capturable-2 but dead.
There is a myth that Japanese rules would have an independence of life and death by virtue of where the hypothetical-moves of a hypothetical-sequence may be played. This is not the case. The hypothetical-moves of a hypothetical-sequence may be played everywhere on the board (subject to the ko-pass rule, which delays the right to play on some particular intersections), i.e., globally. There is not an independence of the hypothetical-moves but a locality of intersections of where a new permanent-stone may provide life for a particular, currently analysed string.
Example 0013
|
[Discovered by: Robert Pauli.] |
U = uncapturable white string 1 = capturable-1 white string |
|
The local-2 of the right white string. |
|
|
Summary: the dead strings D. |
Is the right white string capturable-2? When, in diagram 0013c, Black 1 captures it, White 2 and 4 try to play a permanent-stone on an intersection in the local-2 of the right white string. Playing outside the local-2 with White 6 does not play a permanent-stone on an intersection in the local-2 of the right white string, either. So the right white string is not capturable-2 but dead.
§7.1 - Number of Hypothetical-Sequences
|
|
|
|
|
White 4 - 28 = pass |
How many hypothetical sequences starting with Black do you think are possible in the position of diagram 0014? Just the one of diagram 0014a?
More hypothetical-sequences are possible. Here is a selection: 1) The hypothetical-moves of diagrams 0014b1-2 followed by a Black pass. 2) The hypothetical-moves of diagrams 0014b1-2 repeated twice, then followed by a Black pass. 3) The hypothetical-moves of diagrams 0014b1-2 repeated thrice, then followed by a Black pass. 4) Etc. up to an infinite repetition of the hypothetical-moves of diagrams 0014b1-2.
The incredible number of all legally possible hypothetical-sequences is infinite. Most (in fact, almost all) of the hypothetical-sequences are strategic nonsense. However, the Japanese 1989 Rules do not prohibit strategic nonsense. In particular, there is no rule that would prohibit filling of one's eye of a two-eye-formation.
What is the relevance for interpretation of the rules? Somehow, during hypothetical-sequences, strategic choices must distinguish good from bad. The cardinal mistake of the Japanese 1989 Rules is to hide this key aspect behind a word of grammar: "cannot", which appears in "Stones are said to be 'alive' if they cannot be captured by the opponent [...]". During the interpretation process, we have already separated grammar from contents as in "cannot force". "force" is already a concept familiar to go players, when it appears, e.g., in "forcing move".
For the rules, a definition is required, and the Japanese 2003 Rules provide a definition via the terms "left-part", "hypothetical-strategy", and "compatible". Unfortunately, these definitions please only the mathematician and are inapplicable in practice due to the infinite numbers of hypothetical-sequences and hypothetical-strategies. Players and referees need a compromise in between the theoretically complete solution and the official commentary of the Japanese 1989 Rules' pretence that only one hypothetical-sequence would be sufficient for assessing the status of a particular string. The interpretative text uses a compromise:
"force" is left undefined but shall abide by the following guidelines: One status question of one string is analysed at a time. If a player forces something, then his opponent pursues the opposite objective. Detailed analysis is done only where the players disagree. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting hypothetical-sequences.
§7.2 - Original Text
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.
§7.2 - Interpretation
General
During study of the aspects of hypothetical-sequences, already a lot of gaps of the original pass for a particular ko capture rule have been mentioned. Earlier it has been demonstrated that the original §7.2 ko rule contradicts the official commentary on the Japanese 1989 Rules. Therefore both a much more precise rules text and a different concept of ko-pass rules are required for the interpretative rules.
Furthermore, ko-pass rules must avoid the following unclarities: What is the meaning and purpose of three or more successive ko passes? Answer: This should be prohibited. What is the meaning and purpose of a ko pass made before the opponent has captured in the ko? Answer: This should be prohibited. What is the meaning and purpose of a player's second ko pass for the same ko until he recaptures in it? Answer: This should be prohibited. When a ko temporarily disappears (like in a so called multi-stage ko, a so called perpetual ko, a shape including so called triple-ko-stones, or in a so called molasses ko), does one have to make a ko-pass for the ko on the same intersections again before one's recapture? What about a ko that disappears and later reappears on the earlier ko's intersections but with swapped placement of the black or white stone, respectively? Answer: The rules should not create ambiguity related to temporarily disappearing kos.
As discussed earlier, the Japanese 1989 Rules cannot be interpreted literally because that would lead to nonsense in the professional Japanese players' view. At the same time, it makes sense to assume that those professional Japanese players that wrote the rules and its official commentary were able to know which scoring behaviour each particular example of the official commentary should have. This assumption makes sense since professional Japanese players are confronted with countless problems of similar reading difficulty in their own games. Hence we need ko-pass rules that model the desired scoring behaviour of all the official commentary's examples correctly.
Such are the ko-pass rules of the Japanese 2003 Rules, version 35a. They also dissolve the flaws mentioned above. The interpretative text becomes:
The following applies after a game stop in §9 but not in between a resumption and the next game stop following the resumption. A player may make a ko-pass only if a ko-capture by the player in some ko is currently available but prohibited. In an imagined sequence of hypothetical-moves, between a player's ko-capture on one intersection of a ko and the next ko-pass by him or his opponent, the opponent may not make a ko-capture on the other intersection.
In practice, the interpretative ko-pass rules and the original §7.2 ko-pass rule differ only if during a hypothetical-sequence the players make ko-captures in two or more kos. The most important new aspect of the interpretative ko-pass rules is that a ko-pass applies to all kos' previously made ko-captures on the board and then allows both players to recapture in those kos.
Application of the Interpretative Ko-pass Rules: Example 0001 (continued)
|
3 = ko-pass |
6 = ko-pass |
|
3 = ko-pass 4 = pass |
7 = ko-pass 8 = pass |
|
3 = pass 4 = ko-pass |
7 = pass 8 = ko-pass |
|
3 = ko-pass |
6 = ko-pass 7 = pass |
|
10 = ko-pass |
13 = ko-pass 14 = pass |
The hypothetical-sequences in diagrams 0001k1 to 0001n4 indicate a representative selection of possible, legal infinite hypothetical-sequences starting by White. A ko-pass may not succeed a ko-pass because that would violate the rule "A player may make a ko-pass only if a ko-capture by the player in some ko is currently available but prohibited.". However, a pass may succeed a ko-pass (like in the hypothetical-sequence indicated by diagrams 0001l1-2 and then recurring forever), or a ko-pass may succeed a pass (like in the hypothetical-sequence indicated by diagrams 0001m1-2 and then recurring forever). The opponent of a player having made a ko-pass may also be the next to make a ko-capture (like in the hypothetical-sequence indicated by diagrams 0001k1-2 and then recurring forever). The beginning of mixed type is shown in the hypothetical-sequence of diagrams 0001n1-4.
Such a great variety of hypothetical-sequences is legal that one might wonder what is illegal.
|
3 = pass |
White 4 is illegal. |
White 4 is illegal because neither player has made a ko-pass (Do not confuse ko-pass with pass!) since the ko-capture 1 yet. If White wants a chance to recapture, he has to create the right to do so by using his hypothetical-move 4 for a ko-pass. If Black 3 were a ko-pass, then already the recapture White 4 would be legal.
|
3 = ko-pass |
6 = ko-pass |
|
3 = ko-pass 4 = pass |
7 = ko-pass 8 = pass |
|
3 = pass 4 = ko-pass |
7 = pass 8 = ko-pass |
|
3 = ko-pass |
6 = ko-pass 7 = pass |
|
10 = ko-pass |
13 = ko-pass 14 = pass |
These are the beginnings of some infinite hypothetical-sequences starting by Black. As undoubtably the reader has already noticed, the variety is great because a succession of either pass and ko-pass or ko-pass and pass does not end a hypothetical-sequence. To end a hypothetical-sequence and let it be finite, there have to be two successive hypothetical passes.
Application of the Interpretative Ko-pass Rules: Example II.16 (continued)
|
(repeated) |
U = obviously uncapturable strings ? = strings with still unclear status |
|
- = obviously not uncapturable strings |
Is this string uncapturable? |
|
4, 5 = pass |
2 = ko-pass 4, 5 = pass Due to II.16g and II.16h, the string in II.16f is not uncapturable. |
|
Is this string uncapturable? |
2 = ko-pass |
|
6 = pass |
10, 11 = pass |
|
II.16k1 2 = ko-pass |
II.16k2 8 = ko-pass |
|
II.16k3 12, 13 = pass 9 is legal because 4 was not a ko-capture. 10 is legal because, after the ko-capture 7, there was the ko-pass 8. |
4 = ko-pass |
|
10 = ko-pass |
14, 15 = pass 11 is legal because 6 was not a ko-capture. 12 is legal because, after the ko-capture 9, there was the ko-pass 10. Due to II.16j1-3, II.16k1-3, and II.16l1-3, the string in II.16i is not uncapturable. |
It is becoming clearer and clearer that a complete application of the rules to the position II.16 is an endless story. We need some analysis tesujis for a semi-formal shortcut. Let us prove that Black, if he starts, can force capture of all white strings on the upper and right sides and fill all the intersections there with black stones or ponnukis so that White does not get any permanent-stone there. This is shown by hypothetical-sequences like in diagrams II.16m1-6 and II.16n1-6.
|
II.16m1 2 = ko-pass |
II.16m2 8 = ko-pass |
|
II.16m3 9 is legal because 4 was not a ko-capture. 10 is legal because, after the ko-capture 7, there was the ko-pass 8. Variations: It does not help White to make a ko-pass or pass with 10. |
II.16m4 16, 18, 20 = pass 22 = ko-pass 21 is legal because, after the ko-capture 6, there was the ko-pass 8. 23 is legal because it is not a ko-capture. |
|
II.16m5 30, 32, 34, 36, 38, 40 = pass |
II.16m6 42, 43 = pass |
|
II.16n1 4 = ko-pass |
II.16n2 10 = ko-pass |
|
II.16n3 |
II.16n4 18, 20, 22 = pass 24 = ko-pass 23 is legal because, after the ko-capture 8, there was the ko-pass 10. 25 is legal because it is not a ko-capture. |
|
II.16n5 34, 36, 38 = pass |
II.16n6 40, 42, 43 = pass |
Now we can apply these (and similar) hypothetical-sequences to answer each of the following status questions:
|
II.16o Is this string uncapturable? No. |
II.16p Is this string capturable-1? No. |
|
II.16q Is this string capturable-1? No. |
II.16r Is this string capturable-1? No. |
|
II.16s Is this string capturable-1? No. |
II.16t Is this string capturable-1? No. |
|
II.16u For each white string on the upper or right side, this is its local-2. It coincides for all those white strings because none of them is uncapturable or capturable-1; the local-2 is restricted only by white's uncapturable string in the lower left. |
II.16v Is this string capturable-2? No. |
|
II.16w Is this string capturable-2? No. |
II.16x Is this string capturable-2? No. |
|
II.16y Is this string capturable-2? No. |
II.16z Is this string capturable-2? No. |
So each of the white strings on the upper and right sides is dead. Now we analyse each of the black strings on the right side. White starts hypothetical-sequences for that.
|
II.16aa The marked black string is uncapturable. For typical hypothetical-sequences, see example 0001. |
II.16ab Is this string capturable-1? |
|
II.16ac1 3 = ko-pass |
II.16ac2 7 = ko-pass With 5 or 7, White may not recapture 4 yet because, after Black 4, there has not been a ko-pass yet. |
|
II.16ac3 13 = ko-pass |
II.16ac4 19 = pass 14 is legal because 9 was not a ko-capture. 15 is legal because, after the ko-capture 12, there was the ko-pass 13. |
|
II.16ac5 |
II.16ac6 25 = pass 27 = ko-pass |
|
II.16ac7 31, 32 = pass Black 28 is a permanent-stone in the original string's local-1. Because of this and similar hypothetical-sequences, the string in II.16ab is capturable-1, i.e., alive. |
II.16ad Summary: alive strings A and dead strings D. |
Application of the Interpretative Ko-pass Rules: Example II.17 (continued)
|
(repeated) |
U = uncapturable strings ? = strings whose status is still unclear |
|
4 = ko-pass This hypothetical-sequence will be used to simplify further rules application to the example. |
8 = ko-pass |
|
14 = ko-pass |
18 = ko-pass |
|
24 = ko-pass |
|
|
|
46, 48, 50, 52, 54, 56, 58, 60, 62, 64 = pass |
|
66, 68, 70, 71 = pass |
4 = ko-pass Also this hypothetical-sequence will be used to simplify further rules application to the example. |
|
8 = ko-pass |
16, 18 = pass. 20 = ko-pass. 13 is legal because 12 was not a ko-capture. 14 may not recapture 13 because, since Black 13, there has not been any ko-pass yet. White 14 is legal because, after the ko-capture 7, there has been the ko-pass 8. 19 is legal because, after the ko-capture 6, there has been the ko-pass 8. |
|
22 = pass |
24, 26, 28, 30, 32, 34, 36, 38, 40, 41 = pass |
|
- = neither uncapturable nor capturable-1 white strings For each of the marked white strings, this can be concluded from hypothetical-sequences like in diagrams II.17e1-9 and II.17f1-5. |
This is each marked white string's local-2. |
|
D = dead white strings For each of the marked white strings, this can be concluded from both their local-2 and hypothetical-sequences like in diagrams II.17e1-9 and II.17f1-5. |
- = not uncapturable black strings |
|
Is this string capturable-1? |
2 = pass |
|
6 = pass If White did not attack the stone 4 but passed, then it would end up as a local-1 permanent-stone. So Black 6 can relax and watch White's actions for a moment. |
|
|
19, 20 = pass Black 10 and 12 play local-1 permanent-stones. Therefore and because of similar possible hypothetical-sequences, the string of II.17k is capturable-1, i.e. alive. |
Is this string capturable-1? |
|
3 = ko-pass |
7 = ko-pass With 5 or 7, White may not recapture 4 yet because, since Black 4, there has not been any ko-pass yet. |
|
11 = ko-pass |
16 is legal because 15 was not a ko-capture. With 17, White could not recapture 16; a ko-pass would have been necessary first. |
|
22 is legal because it is not a ko-capture. |
27 = pass |
|
33, 34 = pass Black 30 plays a local-1 permanent-stone. Therefore the string in II.17m is capturable-1 and alive. |
Summary: alive strings A and dead strings D. |
Further Study
(The following remarks are not essential for an understanding of how the original rules or the interpretative rules work, so the reader might skip the following examples of this section. However, what follows is interesting with respect to the general relevance and historical context of ko-pass rules per se. Note that those ko-pass rules work differently from the interpretative rules' ko-pass rule; do not use the following examples to learn the latter.)
The interpretative text's ko-pass rules always model current professional Japanese rules tradition correctly, where that tradition has been clear enough to be known to the public. As far as it is known so far (but study of examples has not been exhaustive yet), the §6 ko rule (which is the ordinary ko rule) applied to hypothetical-sequences instead of the §7.2 ko-pass rule models tradition about as well or badly as the Japanese 1989 Rules' §7.2 ko-pass rule does. Both are about equally far away from the interpretative text ko-pass rules' success. Of course, there are strategic differences of applying either the §6 ko rule or the §7.2 ko-pass rule, but those are hard to find. The most "frequent" known differences are already about as rare as a position with a so called moonshine life at the end of the alternation, which has changed its life and death status a few times during the professional Japanese rules history. The following examples show them:
|
[Discovered by: Winfried Borchardt.] Is the lower white ko stone alive? |
Study for the §6 ko rule + enable. |
|
White 4 enables a new, uncapturable stone. Therefore the white ko stone is alive. |
Study for the §7.2 ko rule + enable. |
|
4 = pass for recapturing stone 1 6 = pass for recapturing stone 3 8, 9 = pass White does not get a newly enabled, uncapturable stone. (Playing next to the lower ko was possible anyway and would not enable anything.) Therefore the white ko stone is dead. |
Does the big white string have uncapturable life? (There are similar positions where some of kos are open for the opponent. Each such position would have to be studied afresh and independently.) |
|
Study for the §6 ko rule + enable. |
The hypothetical-moves 1 to 6 can recur forever. The big white string is not captured. |
|
Study for the §6 ko rule + enable. (continued) |
The big white string is not captured. Both hypothetical-sequences do not capture it. So it is alive. |
|
Study for the §7.2 ko rule + enable. 4 = pass for recapturing 1 |
6, 7 = pass The big white string is captured. Afterwards it would be impossible for White to enable a new, then uncapturable stone. Hence the big white string is dead. |
In view of the rarity of status differences between the §6 ko rule and the §7.2 ko-pass rule, one wonders why a ruleset like the Japanese 1989 Rules has invented a ko-pass rule that very likely still over 99% of all go players do not know at all. Have the authors of the Japanese 1989 Rules put greater emphasis on their personal preference about the score of a very few, particular rare positions than on the needs of the ordinary go players? One might guess that those authors wanted to enforce some vague aims like "Two eye groups partition the board into localities." and "One exclusive, internal non-ko eye beats one exclusive, external ko.". The authors' preference set those aims at a high priority while the aims were not included in the Japanese 1989 Rules because the aims were not well defined and would have made the rules even longer without confidence as to their success in general. With a ko-pass rule included in the rules instead, for a few years the authors could pretend generalized insight until counter-examples that contradict tradition were exhibited by critics of the Japanese 1989 Rules. Ironically, their authors supplied some counter-examples themselves and all that had to be done was proving this carefully. Now it is those authors' turn to learn from the critics.
§7 - Interpretative Text
§7 Life and death
0. A "ko-pass" announces the word ko-pass. A "hypothetical-move" is either an imagined board-play, an imagined pass, or an imagined ko-pass. A "hypothetical-sequence" is an imagined sequence of hypothetical-moves that a) starts with a particular player from the position in that life and death are to be determined, b) lets the players alternate hypothetical-moves, and c) either has a finite number of hypothetical-moves and ends with the pass succeeding a pass or has an infinite number of hypothetical-moves and does not have a pass succeeding a pass. A "permanent-stone" is a stone that is played during a hypothetical-sequence and then not removed during the rest of the hypothetical-sequence.
1. A player's string is "uncapturable" if the opponent cannot force capture of its stones. For a string, "local-1" is all the string's intersections. A player's string is "capturable-1" if the opponent cannot force both capture of the string's stones and no local-1 permanent-stone of the player. For a player's string, "local-2" is local-1 and, recursively, any adjacent intersection without a stone of a string that is of the player and either uncapturable or capturable-1. A player's string is "capturable-2" if the opponent cannot force both capture of the string's stones and no local-2 permanent-stone of the player. An uncapturable string is neither capturable-1 nor capturable-2. A capturable-1 string is not capturable-2. A string is "alive" if it is either uncapturable, capturable-1, or capturable-2. A string is "dead" unless it is alive.
2. The following applies after a game stop in §9 but not in between a resumption and the next game stop following the resumption. A player may make a ko-pass only if a ko-capture by the player in some ko is currently available but prohibited. In an imagined sequence of hypothetical-moves, between a player's ko-capture on one intersection of a ko and the next ko-pass by him or his opponent, the opponent may not make a ko-capture on the other intersection.
3. In every hypothetical-sequence, the opponent of an analysed string starts, the players alternate hypothetical-moves until a player makes a pass and his opponent makes a pass in succession or else for an infinite number of hypothetical-moves, and analysis is done hypothetically, i.e. without physically executing moves or captures. Whether something is achieved in case of an infinite number of hypothetical-moves depends on whether it is achieved already until including the first repetition of a sequence of hypothetical-moves that then infinitely often recurs.
4. "force" is left undefined but shall abide by the following guidelines: One status question of one string is analysed at a time. If a player forces something, then his opponent pursues the opposite objective. Detailed analysis is done only where the players disagree. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting hypothetical-sequences.
§7 - Remarks on the Interpretative Text
So far the numbering of the subparagraphs is chosen to agree to the original's subparagraphs 1 and 2. In a revision of the interpretative text, the numbers of every paragraph's subparagraphs shall start counting from 1.
Compared to the high standard of the interpretative text's §§1-6, §7 lacks a bit in clarity. This is a consequence of searching a compromise in between text length, readability, and precision of the Japanese 2003 Rules source. Now the interpretative §7 is dense - every clause has new contents. The reader is asked to go back once more to the interpretation sections for §7 further above to see how the interpretative text as such applies.
§8 - Original Text
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.
§8 - Interpretation
General
The Japanese 1989 Rules' §8 contains two major flaws. One concerns a wrong definition of "eye point", the other concerns a missing definition of "possess". Both major flaws let any naive attempt of a literal application of the Japanese 1989 Rules be futile.
"eye point" is defined wrongly as follows: Dead strings in a player's territory in the official §10.1 do not exist before their stones' supposed removal because eye points fail to include the dead strings' intersections. The interpretative text corrects this by first defining "black-eye-string" and "white-eye-string" as objects that can consist of empty intersections and intersections of opposing dead strings. Like strings, each such object is a set of intersections connected via the grid lines. "black-eye-point", "white-eye-point", and "eye-point" are then derived from those objects. Where the original rules text lets a player surround only sets of empty intersections, simply speaking, the interpretative text lets a player surround sets of intersections that are empty or have dead opposing stones on.
The missing definition of "possess" lets it be ambiguous whether "[in] seki" applies only to strings directly adjacent to dame or also to strings of certain sets of strings that, as a set, are adjacent to dame. Therefore "possess" in §8 has to be clarified. As can be seen from example II.24.c of the official commentary on the Japanese 1989 Rules, it refers not only to strings adjacent to dame but more generally to, let us call it, groups of strings of that at least one string is adjacent to dame. For the clarification, the interpretative text does not define "possess" but instead becomes even clearer by using the new terms "black-region" and "white-region" and then defining "in-seki" as their property of being adjacent to dame.
In the original text, the term "surrounded" is ambiguous. It might be defined as follows: "A set of intersections that is connected via grid lines is said to be 'surrounded' by stones with a particular feature if the set is adjacent to stones with that feature but not adjacent to empty íntersections outside the set nor to stones without that feature." However, in the interpretative text, "surrounded" becomes superfluous again because the more precise definitions use the phrase "adjacent and only adjacent to". Similarly, "possess" is instead expressed by means of "adjacent". The new definitions are taken from the Japanese 2003 Rules.
Doubtful usage of language is replaced by greater precision: "empty point" becomes "empty intersection", "live" becomes "alive", and "in 'seki'" becomes "in-seki" to indicate that it is a property and not to be confused with the object called "[a / the] 'seki'".
The original text does not clarify when territory scores. The original §§8 and 10 together confuse the reader about the scoring value of territory intersections with dead stones on them and of those dead stones because they do not clearly distinguish a score definition from the physical counting mechanics for calculating the score. The interpretative text clarifies this by defining the score explicitly and without confusing it with the counting mechanics. To easy reading, this is done in a new subparagraph of §8.
Structure
Upon seeing the many terms in the original text and even more so in the interpretative text, the reader might ask for an overview of their structure. It is as follows:
Why is there the in-seki exception? Why is the following approach that would be significantly simpler not used?
Mainly the answer lies in the Japanese rules tradition. 1) It has viewed upon so called independent life, so called seki, and so called death as three types of life and death of equal importance. 2) It has considered each to be already rules objects and not only objects on a purely strategic level. 3) Already on the rules level and not only on a level of purely strategic objects, it has wanted to apply the term "alive" to both independent life and seki. (Ironically, sekis are remarkably scarce in Japanese professional games. E.g., in comparison, sekis are about 25 times as frequent in amateur go server games. Also the more aggressive Korean or Chinese playing styles favour a much greater frequency of sekis.) It is possible to design Modern Territory Scoring naturally without using concepts of "seki" in the rules at all (some such rulesets let sekis have territory, others let them have none). It is also possible to design close approximations to Traditional Territory Scoring without introducing "seki" as an exception (then sekis do not have any territory). Only the traditionalists' desire to maintain clearly more than 99.9% of the tradition requires the construction of "in-seki" in the rules at all, by its nature as an exception there, leads to side effects like rules technical sekis (due to dame in between independently alive strings of both players), and asks for a definition of the object [the] "seki" in a manner even more sophisticated than a definition of "in-seki". One wonders though why the world of go outside Japan admires also the 0.01% of the Japanese rules tradition so much that it copies the Japanese 1989 Rules 1:1. Probably the reason has just been that until recently understanding of good alternatives has not been wide-spread yet.
What it the relevance of excluding territory from sekis? Most sekis do not even have eye-points but only shared dame. By far most of the sekis with eye points are symmetrical: Each player has equally many eye-points in the seki. Therefore it does not matter whether they are territory or not - their difference is zero in both cases. Only in the scarce asymmetrical sekis the score differs when their eye points are territory. Why do some Japanese and Korean professionals fear a decline of Japanese or Korean go tradition and culture per se whenever a change to territory in sekis is suggested? Maybe it is rather a political trick to draw attention to such a minor issue in an attempt to avoid discussion of the fundamental topic why usage and definitions of "alive" and "dead" already on the level of rules terms (and not only on higher, pure strategic levels) should be justified, good, and practicable at all.
Example 0002 (continued)
|
(repeated) A = alive. D = dead. |
B = black-eye-strings |
W = white-eye-strings |
|
B = black-eye-points (There are no white-eye-points.) |
E = eye-points (There are no dame.) |
B = black-region not in-seki (There are no black-regions in-seki and no white-regions.) |
|
B = black-territory (There is no white-territory.) |
T = territory |
Black's score = 4 White's score = 0 |
The black-eye-strings and white-eye-strings may look somewhat strange. They are a sort of algorithmic test strings. The black-eye-strings consist of the empty intersections and the intersections with stones of dead white strings. The white-eye-strings consist of the empty intersections and the intersections with stones of dead black strings. Then, for every black-eye-string, it is verified whether it is adjacent to and only adjacent to stones of alive black strings. This gives the black-eye-points, which are already more familiar to a go player. Similarly, from white-eye-strings one derives the white-eye-points, except that in this example there are none because there are no alive white strings on the board that possibly could surround a white-eye-string.
Why are the diagrams for the black-eye-strings, the black-eye-points, and the black-territory identical? The first two are identical because there are no alive white strings and no dame on the board. The last two are identical because there are no in-seki black-regions.
In the examples, White does not have a score greater than zero. If White had some scoring intersections, then in diagrams the values should be negative numbers. Furthermore, the examples assume that earlier prisoners were not made; if there were some, then they would also have to be mentioned.
Example 0003 (continued)
|
(repeated) A = alive. D = dead. |
B = black-eye-strings |
W = white-eye-strings |
|
B = black-eye-points (There are no white-eye-points.) |
E = eye-points (There are no dame.) |
B = black-region not in-seki (There are no black-regions in-seki and no white-regions.) |
|
B = black-territory (There is no white-territory.) |
T = territory |
Black's score = 19 White's score = 0 |
Example 0004 (continued)
|
A = alive (There are no dead strings.) |
B = black-eye-strings |
W = white-eye-strings |
|
B = black-eye-points W = white-eye-points |
(repeated) E = eye-points. D = dame. |
Sb = black-region in-seki Sw = white-region in-seki (There are no black-regions or white-regions not in-seki.) |
|
(There is neither black-territory, white-territory, nor territory.) |
Black's score = 0 White's score = 0 |
Example 0017
|
|
A = alive. D = dead. |
B = black-eye-strings |
|
W = white-eye-strings |
B = black-eye-points W = white-eye-points |
E = eye-points. D = dame. |
|
Sb = black-region in-seki Sw = white-region in-seki (There are no black-regions or white-regions not in-seki.) |
(There is neither black-territory, white-territory, nor territory.) |
Black's score = 0 White's score = 0 |
§8 - Interpretative Text
§8 Territory
1. A "black-eye-string" is an intersection that is empty and, recursively, any adjacent intersection that is empty or occupied by a stone of a dead white string. A "white-eye-string" is an intersection that is empty and, recursively, any adjacent intersection that is empty or occupied by a stone of a dead black string. An intersection of a black-eye-string is a "black-eye-point" if the black-eye-string is adjacent and only adjacent to intersections with stones of one or more than one alive black strings. An intersection of a white-eye-string is a "white-eye-point" if the white-eye-string is adjacent and only adjacent to intersections with stones of one or more than one alive white strings. An "eye-point" is either a black-eye-point or a white-eye-point. A "dame" is an empty intersection that is not an eye-point. A "black-region" is an intersection with a stone of an alive black string and, recursively, any adjacent intersection that has a stone of an alive black string or is a black-eye-point. A "white-region" is an intersection with a stone of an alive white string and, recursively, any adjacent intersection that has a stone of an alive white string or is a white-eye-point. A black-region or a white-region is "in-seki" if at least one of its intersections is adjacent to a dame. An intersection is "black-territory" if it is a black-eye-point that is not part of an in-seki black-region. An intersection is "white-territory" if it is a white-eye-point that is not part of an in-seki white-region. "territory" are the intersections of black-territory or white-territory.
2. Just before §10 is executed, each empty intersection of black-territory scores one point for Black, each occupied intersection of black-territory scores two points for Black, each prisoner stone of white colour made earlier scores one point for Black, each empty intersection of white-territory scores one point for White, each occupied intersection of white-territory scores two points for White, each prisoner stone of black colour made earlier scores one point for White.
§9.1 - Original Text
Article 9. End of the game
1. When a player passes his move and his opponent passes in succession, the game stops.
§9.1 - Interpretation
There is little to say except that clearer terms should be used instead of verbs. Since §9.3 allows resumption, "game stop" requires the indefinite article. Presumably the most interesting change concerns the headline, which fits the contents of §9.2 better, as will be seen later when the "game end's" aspects are studied.
§9.1 - Interpretative Text
§9 Actions after a game stop
1. When a player makes a pass and his opponent makes a pass in succession, a "game stop" is reached.
§9.2 - Original Text
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."
§9.2 - Interpretation
General
Let us start by reading the original text and its context:
Japanese professional (and amateur) playing practice tells us that in between the (last) game stop and the actions of §10 something happens that is neither an explicit "confirmation and agreement by the two players about the life and death of strings and territory" nor an explicit reaching of the game end but an either formal, alternating or non-formal, possibly not alternating filling of still unfilled so called two-sided dame and so called teire. This raises several questions:
Almost everything in the rules' process after a game stop is unclear. However, since the Japanese Rules use Traditional Territory Scoring, life, death, in seki, and territory cannot be determined by playing things out on the playing board (unless everything including numbers of prisoners but excluding filling of two-sided dame and teire and those prisoners made during that would be restored, but such does not agree to playing practice). Besides, from Japanese professional play, usage of extra boards for playing things out is not reported. Hence one has to conclude that determination of life, death, in seki, and territory, if implicit determination by the players' apparent implicit agreement is not enough, requires analysis of some hypothetical nature.
What can we conclude for the rules? The following should be done: add "filling of still unfilled so called two-sided dame and so called teire", clarify the process, and specify the hypothetical nature. However, since "two-sided" and "teire" would be undefined, the rules should not require the players to fill all dame but only allow them to fill those they want to fill and should not require the players to fill specific intersections but, less prescriptively, only such intersections that the players want to fill. Recently, in Japanese professional play, it has become mandatory to alternate when filling dame. This makes great sense because otherwise the filling (and prisoners made during it) might have to be undone for letting it then be done alternately nevertheless.
The original text is ambiguous about the purpose of confirmation phase, the determination of "end of the game", the nature of application of life-death definitions, and - as a consequence of the original §8 - the "opposing dead stones from his territory".
The interpretative text uses the clear noun phrases "game stop", "game end", and "intersection" as terms consistently. Since resumption is possible, normally game stop is used together with the indefinite article and the scope of application is restricted "until a resumption", i.e., until the next resumption following the previous game stop. Model verbs are used with greater clarity. The term "string" is used to avoid confusion with "stones" and such "groups" that may consist of one or several strings.
The interpretative text allows the players a broad range of styles of how to process their actions in between the last game stop and the actions of §10. This liberality models that of the apparently contained scope of freedom expressed by the original rules text. The interpretative text does not over-interpret by being more restrictive than the Japanese 1989 Rules suggest. However, future revisions of the interpretative text might consider to provide a straightforward standard procedure instead.
It is particularly noteworthy that the players might fill either all or some two-sided dame and teire already before the first game stop. After a game stop, the players may fill further two-sided dame and teire, however, the Japanese 1989 Rules (and so also the interpretative text) do not clarify it beyond doubt whether the players would be required to fill all two-sided dame (maybe with the exception of those in so called sekis; note that zero-sided dame would not be meant anyway) and teire until the actions of §10.
Example 0005 - Variation A
|
(repeated) Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
Black: "Do you agree that the strings are alive, Tb is the black-territory, and Tw is the white-territory?" White: "Yes." Black: "Do you agree to have the game end?" White: "Yes." |
Example 0005 - Variation B
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
Black: "Do you agree that the strings are alive, Tb is the black-territory, and Tw is the white-territory?" White: "Yes." |
Example 0005 - Variation C
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
Black: "Do you agree to have the game end?" White: "Yes." |
Example 0005 - Variation D
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
The players do everything implicitly by being about to start the actions of §10. |
Example 0005 - Variation E
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
Black: "Do you agree that the strings are alive, Tb is the black-territory, and Tw is the white-territory?" White: "No. My string is alive and Tw white-territory but your string is dead and Tb is not territory." |
|
Black: "Look! This is a hypothetical-sequence by that White tries to show that the black string is not uncapturable." |
Black: "This is another such hypothetical-sequence. White fails with both attempts. There are no other plausible attempts. So the black string is uncapturable, i.e., alive." White: "Ok, you have convinced me of that! But what about black-territory?" |
Black: "These are the black-eye-strings." White calls a referee. The referee explains how to apply §8 and confirms Black's statement. White: "Ah, now I understand what black-eye-strings are!" |
|
Black: "These are the black-eye-points." White: "And because the black string is not in-seki, these are also the black-territory?" Black: "Yes." |
Example 0005 - Variation F
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
After the game stop, the players fill the dame alternately. Black: "Do you agree to have the game end?" White: "Yes." |
Example 0005 - Variation G
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
After the game stop, the players fill the dame alternately and pass. |
Example 0005 - Variation H
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
After the game stop, the players fill the dame alternately and pass. |
|
White: "Do you agree that the strings are alive, Tb is the black-territory, and Tw is the white-territory?" Black: "Yes." |
Example 0005 - Variation I
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
After the game stop, the players fill the dame informally. Then they are about to start the actions of §10. |
Example 0005 - Variation J
|
Black to make a move during the alternation. |
The rest of the alternation until the first game stop. |
White: "Do you agree to have the game end?" Black: "Yes." |
|
White: "Do you agree that D is a dame, therefore the strings S are in-seki, and the eye-points E are not territory?" |
Black calls a referee. The referee confirms White's explanation and states that after the mutually agreed, even explicit game end a board-play on D is not possible any longer. |
Many other legal variations of §9.2 applications can be imagined but are not shown.
§9.2 - Interpretative Text
2. After a game stop, until a resumption, and before the actions of §10, some or all of the following is done without any particular order and implicitly and informally or, if necessary, explicitly and formally: a) the "game end", which is the players' agreement of prohibited board-plays afterwards, b) continued alternation until the game end or until a player makes a pass and his opponent makes a pass in succession - without this being another game stop -, c) the objective "hypothetical-analysis", which uses hypothetical means to determine, according to §7 and §8, which strings are alive, which strings are dead, and which intersections are territory, d) the players' agreement as a subjective shortcut for the hypothetical-analysis.
§9.3 - Original Text
3. If a player requests resumption of a stopped game, his opponent must oblige and has the right to play first.
§9.3 - Interpretation
Resumption is superfluous, it adds nothing to the game other than allowing the players' stupidity of forgetting to make some board-plays already until the first game stop, and has hardly ever been used. The original §9.3 contains further flaws: restoration of the board-position and numbers of prisoners, simultaneous resumptions and repetitive resumptions are not treated at all. Although it is criticised that resumptions are superfluous, the interpretative text does not over-interpret by removing that aspect of liberality. The other flaws are corrected though. The resulting text offers an awkward list of conditions, but this cannot be helped if over-interpretation shall be avoided. Future revisions of the interpretative text should provide a more useful choice of conditions (or drop resumptions entirely).
For prohibiting repeated resumptions, the interpretative text has some freedom of whether to use a positional or situational superko condition. The positional condition is chosen because Sakai Takeshi's so called World-Wide Go Rules of 1996 for some go server used it. "position" will be defined in the interpretative text of §12.
§9.3 - Interpretative Text
3. A "resumption" is requested after a game stop, restores the position and numbers of prisoners at the previous game stop, and restarts the alternation of §2, subject to the following sentence's exception. If exactly one player requests resumption, then his opponent makes the next move. If both players request resumption, then the opponent of the last passing player just before the previous game stop makes the next move. A player may not request resumption if the restored position would be the same as that one during one of either player's earlier requested resumptions.
§10 - Original Text
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.
2. Prisoners are then filled into the opponent's territory, and the points of territory are counted and compared. The player with more territory wins. If both players have the same amount the game is a draw, which is called a "jigo."
3. If one player lodges an objection to the result, both players must reconfirm the result by, for example, replaying the game.
4. After both players have confirmed the result, the result cannot be changed under any circumstances.
§10 - Splitting Tournament Rules and Rules of Play
Tournament Rules
3. If one player lodges an objection to the result, both players must reconfirm the result by, for example, replaying the game.
4. After both players have confirmed the result, the result cannot be changed under any circumstances.
Rules of Play
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.
2. Prisoners are then filled into the opponent's territory, and the points of territory are counted and compared. The player with more territory wins. If both players have the same amount the game is a draw, which is called a "jigo."
§10 - Interpretation of the Rules of Play
General
The major flaw of the original §10.2 is the unspecified treatment of excess prisoners. They have to be included in the scores. A flaw in §10.1 concerns imprecise translation: the phrase "as is" ought to be omitted.
It is superfluous for rules to prescribe during counting which player removes which stones. On the other hand, it also does not do harm. So the interpretative text does not drop this requirement. A future revision might reconsider whether to keep it. Contrarily, introduction of a superfluous term, here "jigo", is harmful because each term requires careful interpretation of when it might apply. Therefore this term is omitted. "tie" shall replace "draw" because the other English meanings of "tie" are less harmful than those of "draw".
The clause "After agreement that the game has ended" is somewhat imprecise and clarified in the interpretative text.
The structure into subparagraphs is also revised for greater clarity. (The new §10.3 has nothing to do with the original's §10.3.)
Examples (continued)
|
(repeated) |
There are 8 white prisoner stones.. Black has removed the dead white string from his territory. |
There are 8 excess white prisoner stones. They should be filled into white-territory, but there is none. |
|
There are 8 excess white prisoner stones. Black's score = 11 + 8 = 19. White's score = 0 + 0 = 0. Black wins. |
(repeated) There are 3 white prisoner stones. |
There are 3 white prisoner stones. Dead strings should be removed from territory and their stones added to the prisoners, but there are no dead strings. The 3 prisoners are those carried over from before. |
|
(repeated) There is 1 excess white prisoner stone. Two white prisoner stones are filled into white-territory. |
There is 1 excess white prisoner stone. Black's score = 4 + 1 = 5. White's score = 0 + 0 = 0. Black wins. |
(repeated) |
|
Dead strings should be removed from territory and their stones added to the prisoners, but there is no territory. |
Prisoners should be filled into territory, but there are no prisoners and there is no territory. |
Black's score = 0 + 0 = 0. White's score = 0 + 0 = 0. The game is a tie. |
§10 - Interpretative Text
§10 Determining the result
1. Due to the determination of the dead strings and the territory intersections in §9.2, the score defined in §8.2 is determined by a mechanical counting procedure as follows.
2. Each player removes all stones of opposing dead strings from his territory, and adds them to his prisoners. Prisoners are then filled into the opponent's territory as far as possible.
3. Then the player with more remaining empty intersections of territory plus the number of excess prisoners of opposing colour wins. If both players have the same amount, the game is a tie.
§11 - Original Text
Article 11. Resignation
During a game, a player may end the game by admitting defeat. This is called "resigning." The opponent is said to "win by resignation."
§11 - Interpretation
This is a tournament rule.
§12 - Original Text
Article 12. No result
When the same whole-board position is repeated during a game, if the players agree, the game ends without result.
§12 - Interpretation
General
Starting with a minor flaw, "whole-board position" is undefined. Also the scope of application of the no result rule is ambiguous. The intended scope shall be specified explicitly in the interpretative text. Which is suitable? Should the rule apply to hypothetical-sequences? Suppose it would. Then in one hypothetical-sequence an application might be necessary when a cycle might suggest a no result while in another hypothetical-sequence an application might be not necessary because it is without cycle. How could such two hypothetical-sequences be compared? It would hardly make any sense at all. So it is straightforward to not apply the rule to hypothetical-sequences.
If, under the original rule, the players continue with not agreeing, the referee would have to intervene, but the referee is a tournament rules concept. Therefore the rule is reformulated in the way that expresses how the referee would judge. This dissolves the major flaws about ambiguous strategy due to the uncomparable "no result" and about the unspecified treatment of infinite alternation. Along the way, the terms "Black" and "White" are introduced for greater convenience.
Further moves beyond recreation of the position by a board-play are superfluous. To model the Japanese 1989 Rules as precisely as possible and because quite some players do not consider it stupid to make and be allowed to make them, one is tempted to include "If, beyond recreation of the position by a board-play, accidentally the players have made further moves, then these are considered to have not occurred." in the interpretative text. However, such is a tournament rule and therefore not included in them because they are meant to contain rules of play only.
Interpreting the no result rule and the resumption rule together is not easy. Resumption creates side effects if one is careless. It is clear though that professional Japanese rules tradition wants to allow a so called double ko seki, while no other ko is on the board, to be available for regular scoring, i.e., a player should not be able to force a "no result" during the alternation. This is ensured by "After a resumption, recreation compares only those positions since the last resumption.". One might argue though whether this would already be too good interpretation, i.e. over-interpretation... (The rarity of a double ko seki at the end of the alternation in professional Japanese games does not help.)
For easier comparison, the interpretative text still carries §12. To fill the gap, the revision's rule text decreases the paragraph numbers 12 and above by 1.
Examples
|
0001 (repeated) Black to make a move during the alternation. |
0001t1 Black 1 is a strategic mistake. |
0001t2 Just after Black 5, the position just before Black 1 is recreated. §12 applies and the result is "no result" because of 2 removed black and 2 removed white stones (equally many). |
|
(repeated) White to make a move during the alternation. |
White 1 is a strategic mistake. |
Just after White 3, the position just before White 1 is recreated. §12 applies and the result is a win for Black because of 1 removed black and 2 removed white stones (fewer black than white). |
Needless to say, in a so called triple ko, normally playing a cycle and applying §12 is not a strategic mistake by both players, unless one player has such a big lead that he should sacrifice the triple ko for nothing on the board in exchange.
Under the original rule, in example 0001, the players would or should agree in the §12 sense. In example 0004, if White plays the cycle nevertheless, the players would not agree in the §12 sense but - with a single pass in between every two sequences of board-plays - after many (say 400) cycles Black will have collected so many more prisoners of opposing colour that he can sacrifice the whole board's position to White and still win the game by scoring. The interpretative rule abbreviates this process by anticipating the winner due to the nature of the removals difference per cycle. Typically, European 3 kyus or weaker players are not aware of this implicit consequence of the original rule's phrase "if the players agree". The original rule is also a quiz.
§12 - Interpretative Text
§12 Long cycle
A "position" is the distribution of black or white stones on the board's specific intersections. Recreation of the position after three or more board-plays ends the alternation prematurely with an exceptional result. The result depends on the numbers of black and white stones that have been removed from the board since first leaving the position and until including recreating the position:
This rule applies only until the first game stop or in between a resumption and the next game stop following the resumption. After a resumption, recreation compares only those positions since the last resumption.
§13 - Original Text
Article 13. Both players lose
1. After the game stops according to Article 9, if the players find an effective move, which would affect the result of the game, and therefore cannot agree to end the game, both players lose.
2. If a stone on the board has been moved during the game and the game has proceeded, the game continues with the stone returned to its original point of play. If the players cannot agree, both players lose.
§13 - Splitting Tournament Rules and Rules of Play
Tournament Rules
2. If a stone on the board has been moved during the game and the game has proceeded, the game continues with the stone returned to its original point of play. If the players cannot agree, both players lose.
Rules of Play
Article 13. Both players lose
1. After the game stops according to Article 9, if the players find an effective move, which would affect the result of the game, and therefore cannot agree to end the game, both players lose.
§13 - Interpretation of the Rules of Play
General
"effective move" and "affect" are undefined. "therefore cannot agree" is ambiguous. "to end the game" refers to the §9.2 "game end". Of course, one better uses the noun "game stop" instead of the verb "[the game] stops".
Now what does the rule mean? It is not decrypted easily but, after some thinking, one can come up with the interpretative text.
Examples
|
(repeated) After the first game stop. |
If White requests resumption, the result will be a win of Black. |
If Black requests resumption, the result will be a win of Black. Thus either or both players might as well request resumption rather than apply §13. |
|
Black made the last pass. |
If either Black or both players request resumption, the result will be a tie. |
If White requests resumption, the result will be a win of Black. |
|
The current-result is a win of Black. |
Suppose that Black does not want to request resumption because he prefers the current-result. White does not want to request resumption because he dislikes the win of Black. Then §13 would apply: Since the current-result and the white-result are unequal, the result would be a loss of both players. Therefore Black's best option is to request resumption and get at least the result tie. |
|
|
(repeated) komi = 0.5 Black made the last pass. |
If either Black or both players request resumption, the score will be 0.5 for White and the result a win of White. |
If White requests resumption, the result will be a win of Black. |
|
The current-result is a win of Black. |
Neither player wants to request resumption because the result would be a win of the opponent. Therefore §13 is applied. Since the current-result and the white-result are unequal, the result is a loss of both players. |
|
§13 - Interpretative Text
§13 Loss of both players
1. §13 applies only if neither player requests resumption according to §9.3. §13 applies only after a game stop, after the players do not put any further stones on the board according to §9.2, until the next following resumption, until the game end, and - if the game end is not made clearly - until the players start the actions of §10. A "no result" in §12 is supposed to equal a tie also in perfect-play for the purpose of §13. If tournament rules set a komi, then this is considered for the purpose of §13.
2. The "current-result" is the result if §§7, 8, and 10 were applied to the current position and numbers of prisoners. After imagined resumption with Black making the next move and imagined perfect-play since that resumption to the imagined next game stop, the "black-result" is the result if §§7, 8, and 10 were applied at that game stop. After imagined resumption with White making the next move and imagined perfect-play since that resumption to the next imagined game stop, the "white-result" is the result if §§7, 8, and 10 were applied at that game stop.
3. "perfect-play" is left undefined but shall abide by the following guidelines: Black tries to maximize the difference of Black's score and White's score according to §8.2 applied at the imagined next game stop. White tries to minimize that difference. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting imagined sequences of moves to continue the alternation.
4. If the current-result, the black-result, and the white-result are equal, then the game end is reached and the players continue with the actions of §10. If the current-result, the black-result, and the white-result are not all the same of either win of Black, tie, or win of White, then the result is a "loss of both players".
§14 - Original Text
Article 14. Forfeit
Violation of the above rules causes immediate loss of the game, provided the result has not yet been confirmed by both players.
§14 - Interpretation
This is a tournament rule.
Revision of the Rules of Play
Preliminary Notes
These rules are the clarification and interpretation of the rules of play in the Japanese 1989 Rules. Intentionally, aspects discussed later in the chapter Limitations are not treated by these rules. The only remaining missing clarification might be a reference to the Japanese 2003 Rules, version 35a as a final clarification of the term "force", as a conceptual hint of how to understand the term "perfect-play", and for greater precision of the wording.
Rules
The Japanese Rules of Go
§1 The game of go
Go is a game in which two players play on a board, which is a grid of 19 horizontal and 19 vertical lines forming 361 intersections.
§2 Alternation
The players alternate one move at a time. A "move" is either a board-play or a pass. A "board-play" is placing one's own stone on an empty intersection. One player, called "Black," uses the black stones - his opponent, called "White," the white stones. Black starts.
§3 String and liberty
A "string" is a stone and all stones of the same colour continuously connected to it via grid lines. A "liberty" of a string is an empty intersection horizontally or vertically adjacent to it.
§4 Capture
The player making a board-play removes all stones of the opponent's strings without liberty. They are called "prisoners."
§5 No suicide
A player may not make a board-play if, after it and its captures, the player's string would be without liberty.
§6 Ko
A shape in which the players could alternately capture and recapture one opposing stone is called a "ko." A "ko-capture" is a play that is in a ko and removes exactly one stone. If, in a ko, a player's stone has been removed in a ko-capture, then he may not make a ko-capture in that ko on the next move. This rule applies only until the first game stop or in between a resumption and the next game stop following the resumption.
§7 Life and death
1. A "ko-pass" announces the word ko-pass. A "hypothetical-move" is either an imagined board-play, an imagined pass, or an imagined ko-pass. A "hypothetical-sequence" is an imagined sequence of hypothetical-moves that a) starts with a particular player from the position in that life and death are to be determined, b) lets the players alternate hypothetical-moves, and c) either has a finite number of hypothetical-moves and ends with the pass succeeding a pass or has an infinite number of hypothetical-moves and does not have a pass succeeding a pass. A "permanent-stone" is a stone that is played during a hypothetical-sequence and then not removed during the rest of the hypothetical-sequence.
2. A player's string is "uncapturable" if the opponent cannot force capture of its stones. For a string, "local-1" is all the string's intersections. A player's string is "capturable-1" if the opponent cannot force both capture of the string's stones and no local-1 permanent-stone of the player. For a player's string, "local-2" is local-1 and, recursively, any adjacent intersection without a stone of a string that is of the player and either uncapturable or capturable-1. A player's string is "capturable-2" if the opponent cannot force both capture of the string's stones and no local-2 permanent-stone of the player. An uncapturable string is neither capturable-1 nor capturable-2. A capturable-1 string is not capturable-2. A string is "alive" if it is either uncapturable, capturable-1, or capturable-2. A string is "dead" unless it is alive.
3. The following applies after a game stop in §9 but not in between a resumption and the next game stop following the resumption. A player may make a ko-pass only if a ko-capture by the player in some ko is currently available but prohibited. In an imagined sequence of hypothetical-moves, between a player's ko-capture on one intersection of a ko and the next ko-pass by him or his opponent, the opponent may not make a ko-capture on the other intersection.
4. In every hypothetical-sequence, the opponent of an analysed string starts, the players alternate hypothetical-moves until a player makes a pass and his opponent makes a pass in succession or else for an infinite number of hypothetical-moves, and analysis is done hypothetically, i.e. without physically executing moves or captures. Whether something is achieved in case of an infinite number of hypothetical-moves depends on whether it is achieved already until including the first repetition of a sequence of hypothetical-moves that then infinitely often recurs.
5. "force" is left undefined but shall abide by the following guidelines: One status question of one string is analysed at a time. If a player forces something, then his opponent pursues the opposite objective. Detailed analysis is done only where the players disagree. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting hypothetical-sequences.
§8 Territory
1. A "black-eye-string" is an intersection that is empty and, recursively, any adjacent intersection that is empty or occupied by a stone of a dead white string. A "white-eye-string" is an intersection that is empty and, recursively, any adjacent intersection that is empty or occupied by a stone of a dead black string. An intersection of a black-eye-string is a "black-eye-point" if the black-eye-string is adjacent and only adjacent to intersections with stones of one or more than one alive black strings. An intersection of a white-eye-string is a "white-eye-point" if the white-eye-string is adjacent and only adjacent to intersections with stones of one or more than one alive white strings. An "eye-point" is either a black-eye-point or a white-eye-point. A "dame" is an empty intersection that is not an eye-point. A "black-region" is an intersection with a stone of an alive black string and, recursively, any adjacent intersection that has a stone of an alive black string or is a black-eye-point. A "white-region" is an intersection with a stone of an alive white string and, recursively, any adjacent intersection that has a stone of an alive white string or is a white-eye-point. A black-region or a white-region is "in-seki" if at least one of its intersections is adjacent to a dame. An intersection is "black-territory" if it is a black-eye-point that is not part of an in-seki black-region. An intersection is "white-territory" if it is a white-eye-point that is not part of an in-seki white-region. "territory" are the intersections of black-territory or white-territory.
2. Just before §10 is executed, each empty intersection of black-territory scores one point for Black, each occupied intersection of black-territory scores two points for Black, each prisoner stone of white colour made earlier scores one point for Black, each empty intersection of white-territory scores one point for White, each occupied intersection of white-territory scores two points for White, each prisoner stone of black colour made earlier scores one point for White.
§9 Actions after a game stop
1. When a player makes a pass and his opponent makes a pass in succession, a "game stop" is reached.
2. After a game stop, until a resumption, and before the actions of §10, some or all of the following is done without any particular order and implicitly and informally or, if necessary, explicitly and formally: a) the "game end", which is the players' agreement of prohibited board-plays afterwards, b) continued alternation until the game end or until a player makes a pass and his opponent makes a pass in succession - without this being another game stop -, c) the objective "hypothetical-analysis", which uses hypothetical means to determine, according to §7 and §8, which strings are alive, which strings are dead, and which intersections are territory, d) the players' agreement as a subjective shortcut for the hypothetical-analysis.
3. A "resumption" is requested after a game stop, restores the position and numbers of prisoners at the previous game stop, and restarts the alternation of §2, subject to the following sentence's exception. If exactly one player requests resumption, then his opponent makes the next move. If both players request resumption, then the opponent of the last passing player just before the previous game stop makes the next move. A player may not request resumption if the restored position would be the same as that one during one of either player's earlier requested resumptions.
§10 Determining the result
1. Due to the determination of the dead strings and the territory intersections in §9.2, the score defined in §8.2 is determined by a mechanical counting procedure as follows.
2. Each player removes all stones of opposing dead strings from his territory, and adds them to his prisoners. Prisoners are then filled into the opponent's territory as far as possible.
3. Then the player with more remaining empty intersections of territory plus the number of excess prisoners of opposing colour wins. If both players have the same amount, the game is a tie.
§11 Long cycle
A "position" is the distribution of black or white stones on the board's specific intersections. Recreation of the position after three or more board-plays ends the alternation prematurely with an exceptional result. The result depends on the numbers of black and white stones that have been removed from the board since first leaving the position and until including recreating the position:
This rule applies only until the first game stop or in between a resumption and the next game stop following the resumption. After a resumption, recreation compares only those positions since the last resumption.
§12 Loss of both players
1. §12 applies only if neither player requests resumption according to §9.3. §12 applies only after a game stop, after the players do not put any further stones on the board according to §9.2, until the next following resumption, until the game end, and - if the game end is not made clearly - until the players start the actions of §10. A "no result" in §11 is supposed to equal a tie also in perfect-play for the purpose of §12. If tournament rules set a komi, then this is considered for the purpose of §12.
2. The "current-result" is the result if §§7, 8, and 10 were applied to the current position and numbers of prisoners. After imagined resumption with Black making the next move and imagined perfect-play since that resumption to the imagined next game stop, the "black-result" is the result if §§7, 8, and 10 were applied at that game stop. After imagined resumption with White making the next move and imagined perfect-play since that resumption to the next imagined game stop, the "white-result" is the result if §§7, 8, and 10 were applied at that game stop.
3. "perfect-play" is left undefined but shall abide by the following guidelines: Black tries to maximize the difference of Black's score and White's score according to §8.2 applied at the imagined next game stop. White tries to minimize that difference. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting imagined sequences of moves to continue the alternation.
4. If the current-result, the black-result, and the white-result are equal, then the game end is reached and the players continue with the actions of §10. If the current-result, the black-result, and the white-result are not all the same of either win of Black, tie, or win of White, then the result is a "loss of both players".
Limitations
Purely Technical Sekis
The revision of the rules of play does not treat the purely technical sekis discussed earlier because the non-treatment is a valid feature of the Japanese 1989 Rules. One wants to treat the nonsense in a future revision of the rules, however, this is not possible yet. So far "seki" and "independently-alive group" could be defined only for positions in that there are no two-sided dame outside sekis when they have already been filled. Defining this before requires further research. Until then and unless one uses more fundamental changes for definitions of alive, one might only consider a tournament rule, which might prescribe (alternate) filling without defining the details.
Must all two-sided dame outside sekis be filled until the actions of §10? The Japanese 1989 Rules leave it unclear. The official commentary on §9.2 suggests it while the same official commentary on the examples I.15, II.24a, and II.24c suggest or might suggest the opposite for it is explicitly said that there is no territory because of the dame. The absence of definitions for "two-sided", "outside", "seki", and "teire" in the Japanese 1989 Rules does not encourage the theory that filling would be mandatory. Contrarily, recently - at least in theory - an amendment (an English translation is not known so far) to the Japanese 1989 Rules has made filling in alternation mandatory.
Purely Technically Alive Basic Endgame Ko Stones
The concepts of local-2 and capturable-2 have a side effect if the players have not filled all dame in between independently-alive groups: The string of an unfilled so called basic endgame ko might be alive instead of dead. This will be treated by the concepts of local-3 and capturable-3 together with a direct ko rule.
Direct Kos
General
The Japanese 2003 Rules, version 35a agree to professional Japanese rules tradition even beyond the Japanese 1989 Rules' conceptual attempt in that direction. As a consequence, the Japanese 2003 Rules include a direct ko rule for capturable-3 ko stones. Since the exact degree of the Japanese 1989 Rules' principle intended ability to describe current professional Japanese rules tradition is unclear in this respect, one cannot know how the Japanese 1989 Rules should be interpreted in positions like the following. It is more likely though that the suggested interpretative direct ko rule should be applied.
In the definition of "local-3", stones of both players' uncapturable or capturable-1 strings restrict the environment. For comparison, in the definition of the "local-2" of a player's string, only that player's uncapturable or capturable-1 strings restrict the environment.
Example 0015 (continued)
|
(repeated) |
Is the marked ko-stone capturable-2? (Obviously, it is not capturable-1.) |
|
The local-2 of the lower white ko-stone. |
Hypothetical-sequences like this show that the white ko-stone is capturable-2 because White 2 plays a local-2 permanent-stone. (The upper right corner is immaterial.) |
|
The local-3 of the lower white ko-stone. |
The local-2 and the local-3 for the capturable-2 white ko-stone are unequal. Therefore it is capturable-3. Due to the direct-ko rule, the alternation may not stop as in diagram 0015. |
Example 0019 (continued)
|
The local-3 of the black ko-stone. |
The local-2 and the local-3 for the capturable-2 ko-stone are unequal. Therefore it is capturable-3. Due to the direct-ko rule, the alternation may not stop as in diagram 0019. |
Example 0020
|
[Discovered by: Robert Pauli.] |
4 = ko-pass |
|
7 = ko-pass |
9, 10 = pass Hypothetical-sequences like this show that the black ko-stone is not capturable-1 |
|
The local-2 of the black ko-stone. |
3 = ko-pass |
|
6, 8, 10, 11 = pass Hypothetical-sequences like this with the local-2 black permanent-stones 2 and 4 show that each small black string is capturable-2. |
The local-3 of the black ko-stone. The local-2 and the local-3 for the capturable-2 ko-stone are unequal. Therefore it is capturable-3. Due to the direct-ko rule, the alternation may not stop as in diagram 0020. |
Example II.10 (continued)
|
(repeated) The local-2 of the black ko-stone. |
The local-3 of the black ko-stone. |
This is an example where the local-2 and the local-3 of a capturable-2 ko-stone are equal. The alternation may reach the next game stop and the ko need not be dissolved because the ko-stone is not capturable-3.
Rules Text
§ Direct Kos
A "ko-stone" is a string consisting of exactly one stone and being on one intersection of a ko. For a string, "local-3" is local-1 and, recursively, any adjacent intersection without a stone of a string that is either uncapturable or capturable-1. A string is "capturable-3" if it is a capturable-2 ko-stone and if its local-2 and local-3 are unequal. A pass may not succeed a pass to create a game stop if then a string were capturable-3.
Proposals for Future Revisions
Small Step - Changes
The idea is to skip superfluous rules, replace very by less artificial concepts, use a much clearer game end procedure, but add still necessary or useful rules.
Some players might consider it too strict to have to fill dame and teire until the game stop. If such or restoration of the game stop (without resumption) until the tournament rules' result agreement (when, e.g., the players sign a result form) should be allowed, it is better to do that in tournament rules than to spoil the rules of play. Similarly, tournament rules (instead of the rules of play) might treat pure rules technical sekis.
The authors of the Japanese 1989 Rules did not want to admit that a superko condition would be useful. Therefore they allowed infinite hypothetical-sequences. However, their usage is much less practicable than a functionally equivalent superko condition could ever be.
Why is the resulting text still very long? There are too many ko rules with too many conditions, the used construction of "alive" requires three types, and the no territory in so called sekis comes as an exception.
Small Step - Rules
The Japanese Rules of Go
§1 General
1. Go is a game in which two players play on a board, which is a grid of 19 horizontal and 19 vertical lines forming 361 intersections.
2. The players alternate one move at a time. A "move" is either a board-play or a pass. A "board-play" is placing one's own stone on an empty intersection. One player, called "Black," uses the black stones - his opponent, called "White," the white stones. Black starts. When a player makes a pass and his opponent makes a pass in succession, the "game stop" is reached.
3. A "string" is a stone and all stones of the same colour continuously connected to it via grid lines. A "liberty" of a string is an empty intersection horizontally or vertically adjacent to it.
4. The player making a board-play removes all stones of the opponent's strings without liberty. They are called "prisoners."
5. A "position" is the distribution of black or white stones on the board's specific intersections.
6. A shape in which the players could alternately capture and recapture one opposing stone is called a "ko." A "ko-stone" is a string consisting of exactly one stone and being on one intersection of a ko. A "ko-capture" is a play that is in a ko and removes exactly one stone.
7. A player may not make a board-play if, after it and its captures, the player's string would be without liberty.
§2 Restrictions until the game stop
1. If, in a ko, a player's stone has been removed in a ko-capture, then he may not make a ko-capture in that ko on the next move.
2. Recreation of the position after three or more board-plays ends the alternation prematurely with an exceptional result. The result depends on the numbers of black and white stones that have been removed from the board since first leaving the position and until including recreating the position:
3. A pass may not succeed a pass if then a string were capturable-3.
§3 Definitions for life and death
1. A "ko-pass" announces the word ko-pass. A "hypothetical-move" is either a board-play, a pass, or a ko-pass. A "hypothetical-sequence" is a sequence of hypothetical-moves that a) starts with the opponent of an analysed string from the game stop's position, b) lets the players alternate hypothetical-moves, and c) ends either with the pass succeeding a pass or just after the board-play recreating an earlier position since the hypothetical-sequence's start.
2. A "permanent-stone" is a stone that is played during a hypothetical-sequence and then not removed to its end.
3. A player's string is "uncapturable" if the opponent cannot force capture of its stones.
4. For a string, "local-1" is all the string's intersections. A player's string is "capturable-1" if the opponent cannot force both capture of the string's stones and no local-1 permanent-stone of the player.
5. For a player's string, "local-2" is local-1 and, recursively, any adjacent intersection without a stone of a string that is of the player and either uncapturable or capturable-1. A player's string is "capturable-2" if the opponent cannot force both capture of the string's stones and no local-2 permanent-stone of the player.
6. For a string, "local-3" is local-1 and, recursively, any adjacent intersection without a stone of a string that is either uncapturable or capturable-1. A string is "capturable-3" if it is a capturable-2 ko-stone and if its local-2 and local-3 are unequal.
7. A string is "alive" if it is either uncapturable, capturable-1, or capturable-2. A string is "dead" unless it is alive.
§4 Rules after the game stop
1. Hypothetical-moves and hypothetical-sequences are only imagined. Analysis is done hypothetically, i.e. without physically executing moves or captures.
2. An uncapturable string is neither capturable-1 nor capturable-2. A capturable-1 string is not capturable-2.
3. A player may make a ko-pass only if a ko-capture by the player in some ko is currently available but prohibited. In a sequence of hypothetical-moves, between a player's ko-capture on one intersection of a ko and the next ko-pass by him or his opponent, the opponent may not make a ko-capture on the other intersection.
4. "force" is left undefined but shall abide by the following guidelines: One status question of one string is analysed at a time. If a player forces something, then his opponent pursues the opposite objective. Detailed analysis is done only where the players disagree. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting hypothetical-sequences.
§5 Territory
1. A "black-eye-string" is an intersection that is empty and, recursively, any adjacent intersection that is empty or occupied by a stone of a dead white string. A "white-eye-string" is an intersection that is empty and, recursively, any adjacent intersection that is empty or occupied by a stone of a dead black string.
2. An intersection of a black-eye-string is a "black-eye-point" if the black-eye-string is adjacent and only adjacent to intersections with stones of one or more than one alive black strings. An intersection of a white-eye-string is a "white-eye-point" if the white-eye-string is adjacent and only adjacent to intersections with stones of one or more than one alive white strings. An "eye-point" is either a black-eye-point or a white-eye-point.
3. A "dame" is an empty intersection that is not an eye-point.
4. A "black-region" is an intersection with a stone of an alive black string and, recursively, any adjacent intersection that has a stone of an alive black string or is a black-eye-point. A "white-region" is an intersection with a stone of an alive white string and, recursively, any adjacent intersection that has a stone of an alive white string or is a white-eye-point.
5. A black-region or a white-region is "in-seki" if at least one of its intersections is adjacent to a dame.
6. An intersection is "black-territory" if it is a black-eye-point that is not part of an in-seki black-region. An intersection is "white-territory" if it is a white-eye-point that is not part of an in-seki white-region. "territory" are the intersections of black-territory or white-territory.
§6 Score and result
1. A player's score is one point for each empty intersection of his territory plus two points for each occupied intersection of his territory plus one point for each prisoner stone of opposing colour made earlier.
2. After the game stop and according to §§1 and 3 to 5, the players use hypothetical means to determine the scores. If the players agree, they may skip this.
3. The player with the greater score wins. If both players have the same score, the game is a tie.
Intermediate Step - Changes
Now we make more fundamental changes. In rare positions, they can violate current professional Japanese rules tradition. This is the price for a significant reduction of text length and relative increment in conceptual elegance.
"alive" means "two-eye-alive". Strings in a seki are not alive for the purpose of these rules. (One might rename "alive" into "two-eye-alive" and call the alive strings in a seki "seki-alive", if one prefers that.) The definition of "alive" is a weak form of two-eye-alive but it is closely related to the alive of the Japanese 2003 Rules, version 35a in regions that are not in-seki.
The rules use a different kind of locality. Instead of local-1 / local-2, a two-eye-formation must be established on at least one of the intersections of a string tested for alive.
The "adjacent and only adjacent to" condition in the definitions of black-territory and white-territory mean, simplifying a bit, the same as being surrounded. Thereby there should not be many surprises, although the rules have not been tested to many types of shapes yet. Testing of "alive" is recommended for all the scarce ko shapes though.
Under these rules, it is recommended to force one's opponent to defend all so called teire. Otherwise they might be territory and score.
The ko rules until the game stop have been tested successfully for Area Scoring and differences to the ko rules of those Japanese 1989 Rules valid until the game stop are rare, if "no result" and "tie" are equated. (The shape sending-three-returning-one can be different.)
Intermediate Step - Rules
The Japanese Rules of Go
§1 General
1. Go is a game in which two players play on a board, which is a grid of 19 horizontal and 19 vertical lines forming 361 intersections.
2. The players alternate one move at a time. A "move" is either a board-play or a pass. A "board-play" is placing one's own stone on an empty intersection. One player, called "Black," uses the black stones - his opponent, called "White," the white stones. Black starts. When a player makes a pass and his opponent makes a pass in succession, the "game stop" is reached.
3. A "string" is a stone and all stones of the same colour continuously connected to it via grid lines. A "liberty" of a string is an empty intersection horizontally or vertically adjacent to it.
4. The player making a board-play removes all stones of the opponent's strings without liberty. They are called "prisoners."
5. A "position" is the distribution of black or white stones on the board's specific intersections.
6. A player may not make a board-play if, after it and its captures, the player's string would be without liberty.
7. A board-play immediately succeeding a board-play may not recreate the position.
§2 Ko rules until the game stop
1. Three successive board-plays may not recreate the position, even with one pass in between them.
2. If otherwise a board-play recreates the position, then the game ends prematurely and the result is a tie.
§3 Alive and territory
1. In a position, a "two-eye-formation" is a set of one or several strings of the same player and exactly two empty intersections so that each of the strings is adjacent to each of the two intersections, none of the strings is adjacent to another empty intersection, and each of the two intersections is adjacent only to the strings.
2. A string of a player is "alive" if the opponent cannot force no intersection of the string with a two-eye-formation of the player on.
3. A "mixed-space" is an empty intersection and, recursively, any adjacent intersection that is empty or occupied by a stone of a not alive string. An "empty-space" is a mixed-space consisting of one or more than one empty intersections. A "black-space" is a mixed-space consisting of one or more than one empty intersections and one or more than one intersections occupied by white stones. A "white-space" is a mixed-space consisting of one or more than one empty intersections and one or more than one intersections occupied by black stones.
4. An intersection is "black-territory" if it is in an empty-space / black-space that is adjacent and only adjacent to alive black strings. An intersection is "white-territory" if it is in an empty-space / white-space that is adjacent and only adjacent to alive black strings. "Territory" is either black-territory or white-territory.
§4 Rules after the game stop
1. A "hypothetical-sequence" is a sequence of moves that a) starts from the game stop's position with the opponent of the string currently tested for alive, b) lets the players alternate moves, c) ends either with the pass succeeding a pass or just after the board-play recreating an earlier position since the hypothetical-sequence's start, and d) is done only hypothetically without physically executing moves or captures.
2. The alive strings are determined by means of hypothetical-sequences.
3. "force" is left undefined but shall abide by the following guidelines: One string is analysed at a time. If a player forces something, then his opponent pursues the opposite objective. Detailed analysis is done only where the players disagree. The players should strive to understand the omniscient player's choices. In particular, a player should not consider obvious failures but also should not reject the opponent's suggestions of other possibly interesting hypothetical-sequences.
§5 Score and result
1. A player's score is one point for each empty intersection of his territory plus two points for each occupied intersection of his territory plus one point for each prisoner stone of opposing colour made earlier.
2. After the game stop and according to §§1, 3, and 4, the players use hypothetical means to determine the scores. If the players agree, they may skip this.
3. The player with the greater score wins. If both players have the same score, the game is a tie.
Intermediate Step - Examples
Note that, in the examples for the Intermediate Step, "alive" means "two-eye-alive", i.e., strings in so called sekis are not alive.
Example 0000 (continued)
|
(repeated) |
Is the upper left white string alive? |
|
After hypothetical-sequences like in diagram 0000h, White has a two-eye-formation on at least one intersection of the string: White's outer strings and the empty corner intersections are a two-eye-formation because this includes exactly two empty intersections, each of the included strings is adjacent to each of these empty intersections, none of the included strings is adjacent to another empty intersection, and each of the two included empty intersections is adjacent only to the included strings. Hence the upper left white string is alive. (Similarly, the lower right white string is alive.) |
Is the small black string alive? This is a typical hypothetical-sequence for answering the question. |
|
|
|
|
Black has got a two-eye-formation on the small string's intersection because also the two empty intersections are part of the two-eye-formation. Due to this and similar hypothetical-sequences, the small black string is alive. (Similarly, the big black string is alive.) |
Is the small upper white string alive? No, because White (!) cannot get a two-eye-formation on its intersection. (Similarly, each of the small white strings is "not alive".) |
|
A = alive. M = mixed-spaces. |
E = empty-spaces. B = black-spaces. (There are no white-spaces.) |
|
B = black-territory. W = white-territory. |
Score = (6 + 6 + 0) - (2 + 0 + 0) = 12 - 2 = 10 |
Example 0004 (continued)
|
(repeated) (Only a representative fraction of rules application is shown.) |
Is the left black string alive? |
No, because Black does not get a two-eye-formation on its intersection. |
|
Is the right white string alive? |
White 4 ends the hypothetical-sequence because the position at its start is recreated. |
Note that also the marked part of the board is not a two-eye-formation because one of the two empty intersections is adjacent also to another string. |
|
Hypothetical-sequences like in diagrams 0004k1-2 do not give White a two-eye-formation on the right white string's intersection. Thus that string is not alive. |
M = mixed-space (There is no alive string.) |
(There are neither empty-spaces, black-spaces, nor white-spaces because the only mixed-space consists of intersections that are empty, have a black stone on, or have a white stone on.) |
|
(There is no territory because there are neither empty-spaces, black-spaces, nor white-spaces.) |
Score = (0 + 0 + 0) - (0 + 0 + 0) = 0. |
Example 0010 (continued)
|
(repeated) (Only a representative fraction of rules application is shown.) |
0010d1 Is the small white string alive? |
0010d2 This is a typical hypothetical-sequence. |
|
0010d2 |
0010d3 Black 13 - 45 = pass White 46 = pass |
The small white string is alive because White gets a two-eye-formation on at least one of its intersections. |
|
A = alive M = mixed-spaces |
E = empty-space W = white-space (There is no black-space.) |
W = white-territory (There is no black-territory.) |
|
Score = (0 + 0 + 0) - (10 + 16 + 0) = 0 - 26 = -26 |
Example 0011 (continued)
|
(repeated) Is the upper black string alive? |
|
|
|
|
Yes, because, with hypothetical-sequences like this, Black gets a two-eye-formation on the upper black string's intersection. |
Example 0015 (continued)
|
(repeated) (Only a representative fraction of rules application is shown.) |
Is the lower big black string alive? Yes, because, after a hypothetical-sequence like shown, a two-eye-formation is on at least one of its intersections. (Note that the white ko stone's intersection is not empty and so does not endanger the definition.) |
|
Is the lower right white string alive? |
This is one possible hypothetical-sequence with that White gets a two-eye-formation on at least one of the lower right white string's intersections. |
|
This is another possible hypothetical-sequence with that White gets a two-eye-formation on at least one of the lower right white string's intersections. So that string is alive. |
Is the lower small black string alive? No, because Black cannot get a two-eye-formation on its intersection. |
|
Is the lower white ko stone string alive? This is one possible hypothetical-sequence with that White does not get a two-eye-formation on its intersection. |
This is another possible hypothetical-sequence. |
|
White does not get a two-eye-formation on the lower white ko stone string's intersection, whatever he tries. Therefore it is not alive. |
Is any of the upper right white strings alive? No, because White does not get a two-eye-formation there. |
|
Is the black ko stone string in the upper right alive? |
|
|
The black ko stone string in the upper right is alive because Black gets a two-eye-formation on its intersection. It is immaterial what White does on the rest of the board since the definition of "two-eye-formation" applied to the black ko stone string in the upper right is satisfied nevertheless. |
A = alive. M = mixed-spaces. |
|
E = empty-spaces. B = black-space. (There are no white-spaces. The dubious mixed-space on the lower side contains both a black and a white stone and therefore is neither a black-space nor a white-space.) |
B = black-territory. W = white-territory. |
|
Score = (10 + 6 + 0) - (12 + 0 + 0) = 16 - 12 = 4. |
If White wanted to make white-territory around the lower ko, he should have used a board-play during the alternation to dissolve the ko. It was a strategic mistake of White to not do that, however, even then White could not have won the game (the score would have been 1 then). |
Example 0017 (continued)
|
(repeated) Is the big black string alive? |
0017g1 |
0017g2 Black has not got a two-eye-formation. |
|
Explanation: This is not a two-eye-formation because one of the two empty intersections is also adjacent to another string. |
This is not a two-eye-formation because the ko stone string is not adjacent to both of the empty intersections. |
So the big black string is not alive. (Similarly, none of the strings is alive. Hence there is no territory.) |
Example 0019 (continued)
|
(repeated) Is the small black string alive? |
0019f1 |
0019f2 |
|
Black does not get a two-eye-formation on the small black string's intersection, which is therefore not alive. |
B = black-territory W = white-territory |
Big Step
In a big step, one can move to Modern Territory Scoring or Area Scoring. The latter is off-topic here while the former may be discussed a bit. There are in particular two promising types of Modern Territory Scoring:
Testing and research for both types is pretty advanced (much more so than for the rules of the Intermediate Step). Of course, there are differences to tradition. For Control Territory Scoring, they are very scarce or rare. For Ikeda Territory I Scoring, one-sided plays (like one-sided dame, intersections in asymmetrical sekis, for removal of dead stones from sekis, or defence in a capturable-2 seki) gain points like under Area Scoring. This is the price for getting short applicable rules.
For Ikeda Territory I Scoring, the side effect pass-fights could be imagined but in the meantime it has already been proven that they are rare if only one is careful about the ko rules and number of passes that end the alternation. The current research is about reducing the necessary number of successive passes ending the alternation from three to two while still always forcing the filling of basic endgame kos before the playout.
More information and rulesets are available on the author's webpages and the archives of rec.games.go.
Conclusion
Historical Explanation of the Low Quality of the Japanese 1989 Rules
During the Edo period, go in Japan was played especially by an only small number of elite persons. In such an environment, verbal usage of difficult rules may have been reasonably possible, except for a few incidents about truly rare shapes like moonshine-life. Quite contrarily, spreading of the game to the world was delayed by centuries mainly because of four factors: 1) this exclusiveness, 2) Japan's general isolation, 3) the necessity of translation from Japanese, 4) the difficulty of the Japanese go rules. Historical evidence has indicated that even the last mentioned factor alone was a major hurdle in Europe: Rough rules descriptions had been available much earlier than their even approximately reasonable understanding.
While history, which may be associated with written remainders, is rather old, the history of written Japanese go rules started as late as 1949. Still the contrast between professional playing skill and quality of the rules was like day and night, although the ruleset was formulated in a highly educated language. Why was there this great difficulty and is also seen in the 1989 rules? Apart from the extreme desire of preserving traditional scoring results as precisely and often as possible, the objects demanding description in Japanese style rules are of strategic nature. Mathematicians and computer scientists are having the greatest difficulties with them - so why should players, who are skilled at strategic planning, be skilled to define those objects? These two skills are of very different nature!
Reasons For or Against Using the Japanese 1989 Rules
There is only one reason in favour of the Japanese 1989 Rules: They are the valid ruleset for the Japanese go associations.
Reasons against the rules have been discussed elsewhere in great detail, so here is just a summary: Countless easy / easier alternatives of high design quality are available and some of them are even very close to the Japanese 1989 Rules' strategic consequences. The rules contain many mistakes and gaps. They are too difficult for almost all players. The rules are inapplicable as rules because a correct application would demand arbitrarily much time for its execution.
What about applying one of the revision texts? The may have eliminated the mistakes and filled the gaps, however, they are still (by far) too difficult for almost all players. Why use difficult rules when there are simple rules? Because one could not bear a triple ko resulting in a tie?
Which Ruleset Should be Used?
If one still believes that rules may originate only from professional Japanese go associations, then the original Japanese 1989 Rules are the only choice. (The World Amateur Go Championship 1979 Rules are even worse.)
If one wants to interpret the Japanese 1989 Rules as closely as possible, then one should use the Revision of the Rules of Play.
If one wants to interpret current professional Japanese Rules tradition as closely as possible, then one should use the Revision of the Rules of Play together with the Direct Ko Rules.
For the same reasons but theoretical research, one should use the Japanese 2003 Rules, version 35a.
If one wants both to interpret current professional Japanese Rules tradition as closely as possible and skip the most superfluous rules, then one should use the Small Step's ruleset.
If one wants to keep Japanese style rules that define "alive", agree to Japanese rules tradition except for some rare shapes, but do the go players and referees at least some favour, then one should use the Intermediate Step's ruleset after a bit more testing of scarce ko or seki shapes at the end of the alternation.
If one wants to approach the carelessness of usage of Japanese style rules on go servers or if one thinks that the Intermediate Step's ruleset is still too difficult but a ruleset without any "alive" or "dead" is too non-traditional, then one might consider a written, simplistic model of verbal Japanese rules.
If one wants to avoid a definition of "alive", still model Japanese rules tradition closely except for some rare shapes, and accepts, if the players disagree, two analysis sequences on extra boards, then one should use the New Amateur-Japanese Rules.
If one admires the short length of the rules (because no definitions of life, death, or territory are needed), wants only one playout sequence in case of the players' disagreement, tolerates to end the alternation by three successive passes, and accepts the usage of the playout for getting the points of one-sided plays, then one should use Ikeda Territory I Scoring together with, e.g., the Basic-fixed-ko-rules.
If one does not mind to fight over dame and dame ko threats for basic endgame kos, to have one-sided plays of value, and to use pass stones and an equal number of black and white moves (to let Japanese-fill-in-counting still be possible), then Area Scoring is the choice.
For every ruleset, one can alter some details like the ko rules. E.g., the Intermediate Step ruleset might use the long cycle ko rules of the Revision of the Rules of Play during the alternation (but not after the game stop). However, inexperienced designers shall be warned: It is just too easy to do harm, if one has not understood the theory and does not test carefully.
0001
0001a
0001b1
0001b2
0001c1
0001c2
0001d1
0001d2
0002
0002a
0002b
0002c
0003
0003a
0003b
0003c
0003d
0004
0004a
0004b
0005
0006
0006a
0007
0007a1
0007a2
0008
II.10
II.10a1
II.10a2
II.10b
II.10c
II.10d
II.10e1
II.10e2
II.10f1
II.10f2
II.10f3
II.10f4
II.10f5
II.17
II.17a1
II.17a2
II.17b1
II.17b2
.GIF)
II.17c1
II.17c2
0009
0009a
0010
0010a
0011
0011a
0012
0012a
0010b1
0010b2
0010b3
0010b4
0010c1
0010c2
0010c3
0010c4
0010c5
0011b1
0011b2
0011b3
0011c1
0011c2
0011d
0012b1
0012b2
0012b3
0012b4
0012c1
0012c2
0000
0000e
0000g1
0000g2
0000g3
II.10i1
II.10i2
II.10i3
II.10j1
II.10j2
II.10j3
II.2c1
II.2c2
0001e
0001f
0001g
0001h
0001i
0001j
0013
0013a
0013b
0013c
0013d
0014
0014a
0014b1
0014b2
0001k1
0001k2
0001l1
0001l2
.GIF)
0001m1
.GIF)
0001m2
0001n1
0001n2
0001n3
0001n4
.GIF)
0001o1
0001o2
0001p1
0001p2
0001q1
0001q2
.GIF)
0001r1
.GIF)
0001r2
0001s1
0001s2
0001s3
0001s4
II.16
II.16d
II.16e
II.16f
II.16g
II.16h
II.16i
II.16j1
II.16j2
II.16j3
II.16l1
II.16l2
II.16l3
II.17
II.17d
II.17e1
II.17e2
II.17e3
II.17e4
II.17e5
II.17e6
II.17e7
II.17e8
II.17e9
II.17f1
II.17f2
II.17f3
II.17f4
II.17f5
II.17g
II.17h
II.17i
II.17j
II.17k
II.17l1
II.17l2
II.17l3
II.17l4
II.17m
II.17n1
II.17n2
II.17n3
II.17n4
II.17n5
II.17n6
II.17n7
II.17o
0015
0015a1
0015a2
0015b1
0015b2
0016
0016a1
0016a2
0016b1
0016b2
0016c1
0016c2
0002a
0002d
0002e
.GIF)
0002f
0002g
0002h
.GIF)
0002i
0002j
0002k
0003a
0003e
0003f
.GIF)
0003g
0003h
0003i
.GIF)
0003j
0003k
0003l
0004c
0004d
0004e
0004f
0004a
0004g
0004h
.GIF){kind=small}
0004i
0017
0017a
0017b
0017c
0017d
0017e
0017f
.GIF)
0017g
.GIF)
0017h
0005
0005a
0005b
.GIF)
0005
.GIF)
0005a
.GIF)
0005b
_32.GIF)
0005
.GIF)
0005a
.GIF)
0005b
.GIF)
0005
.GIF)
0005a
0005c
.GIF)
0005
.GIF)
0005a
.GIF)
0005b
0005d
0005e
0005f
0005g
.GIF)
0005
0005h
0005i
.GIF)
0005
.GIF)
0005h
0005m
.GIF)
0005
.GIF)
0005h
.GIF)
0005m
.GIF)
0005b
_34.GIF)
0005
.GIF)
0005h
0005j
.GIF)
0005
.GIF)
0005h
.GIF)
0005
0005k
0003
0003m
.GIF)
0003n
0003o
0006
.GIF)
0006b
0006a
0006c
0017
_36.GIF)
0017g
.GIF)
0017h
.GIF)
0017i
0004
0004i1
0004i2
0005
0005a
0005l
0018
0018a
0018b
.GIF)
0018c
_38.GIF)
0018
.GIF)
0018a
.GIF)
0018b
.GIF)
0018c
0015
0015e
0015f
0015g
0015h
0019e
0020
0020a1
0020a2
0020a3
0020b
0020c1
0020c2
0020d
II.10h
II.10k
0000
0000h
0000i1
0000i2
0000i3
0000i4
0000j
0000k
0000l
0000m
0000n
0004
0004j1
0004j2
0004k1
0004k2
0004l
0004m
.GIF){kind=small}
0004n
.GIF){kind=small}
0004o
.GIF){kind=small}
0004p
0010
0000e
0000f
0000g
0000h
0011
0011d1
0011d2
0011d3
0015
0015i
0015j1
0015j2
0015k
0015l
0015m
0015n1
0015n2
0015o
0015p1
0015p2
0015p3
0015q
0015r
0015s
0015t
0017
0017h
0017i
0019
0019g