|
FRENCH RULES OF GO
The following is a translation by John Fairbairn of some material on the site of
Denis
Feldmann pertaining to the rules of the French Go Federation (FFG).
It was intended to inform discussion by members of the British Go Association
of a proposal to make American Go Association rules the rules of the BGA. French
rules are essentially AGA rules but day-to-day experience in France is rather
closer to the British experience and may have useful insights.
Introduction - how to use this document This text is aimed at various groups of players (at the risk of satisfying none of them). If you have only vaguely heard about go, or if, having had only a cursory description of the rules, you are stumped by exasperating problems such as knowing when the game has ended properly or how to count precisely, the first part (an adaptation of the rules written by François Mizessyn) should meet your needs. If you know the rules but are looking for a rigorous description of the Federation's official version (for example, in order to resolve a rare point of dispute), go directly to the second part. Some justifications for this ruleset, and comparisons with the Japanese method, are given in the third part (and also I recommend reading my article [an English version exists] on these matters so as to have a clearer idea of what is really at stake. Finally, a few other practical matters (mainly connected with refereeing problems) are touched upon in the last part. Readers of a mathematical bent will no doubt be entirely satisfied only by a complete formalised version. The New Zealand rules (which differ only slightly from ours) should show them how to achieve that. These same readers will perhaps wish to see how, in the French rules, certain tricky positions are counted up, whether these are the frequent questions about the last point or the last ko (before the dames), or more positions such as the famous "bent four in the corner" and the "three points without playing" of the Japanese rules. The final appendix should provide them with the tools to deal with all the questions of this nature.
Part 1: French rules of go - an introduction 1. Preamble The game of go originated in China (where it is now known as weiqi) several thousand years ago. It has been played in Japan and Korea (under the name baduk) for 1200 years, but has spread to the West only recently. The aim of the game is to control an area that is initially empty by using the most modest equipment: a board, called a goban, on which a square grid is drawn, and pieces, called stones, which are placed on the intersections of this grid in alternate turns, the aim being (in rough terms) to occupy as much of this area as possible by following some simple rules. These rules can be learnt in a few minutes and allow beginners swiftly to enjoy thrilling games. Anyone who would then like to explore the subtleties of the game can join a club or even enter tournaments. They will then be able to observe that, beneath its apparent simplicity, which makes it accessible to even the youngest players, the game of go is an inexhaustible treasure trove. In the meantime, these few paragraphs, which have the sole aim of presenting these rules, will guide them on their first steps. The texts on a green background correspond to the official rules. The rest is intended to clarify them and to illustrate them. It has been deemed worthwhile in particular to warn against certain possible errors in interpretation (these warnings are on an orange background), and to highlight (on a pale brown background) some important consequences of these rules. It is necesssary above all to point out, however, that while go is indeed a game of simple rules, it is possible to know them superficially without nevertheless having a clear idea of what needs to be done exactly, and how remote and fuzzy the final goal seems to someone who has just learnt the rules. This text is intended simply to help beginners cross this hurdle. We have not, incidentally, considered it worthwhile to add an example of a commented game (this can be found on François Mizessyn's site). In any event, it is necessary to recognise that nothing will truly replace the explanations that can be given by an established player. However, an interactive presentation, rather more attuned to the techniques of the game proper is featured on the Federation's site. You can, by the way, find pointers to many other resources on this site. 2. Equipment Traditional equipmemt comprises a goban on which is drawn a grid of 19x19 lines, forming 361 intersections, and stones which are either black or white. But there is nothing to prevent players from using other equipment, in particular gobans of 13x13 or 9x9 lines for beginners' games. In general (and in particular for gobans of Japanese or Korean manufacture), the space between two lines on the goban is about 24mm in the lengthwise direction and 22mm across. The goban is thus not quite square. In the case of the stones, they are biconvex and have a diameter of about 22mm. In theory, each player has an unlimited number, but experience shows that 180 stones of each colour suffice, by and large, to complete 99.99% of games.
Here is a goban of 19x19 lines. Note that some points are emphasised. These are called hoshis (a hoshi is a star in Japanese). 3. Strings and liberties Two intersections are said to be adjacent when they are on the same line and there is no other intersection between them.
In this diagram, 'a' and 'b' are adjacent intersections, but 'b' and 'c' are not. Two stones are adjacent if they occupy adjacent intersections. A string is a group of one or several stones adjacent one to another of the same colour. The liberties of a string are the empty intersections adjacent to the stones of that string.
The four white stones marked X are adjacent one to another. They form a string that has five liberties: the intersections marked by the letters 'a', 'b', 'c', 'd' and 'e'. 4. Territory A territory is a group of one or several empty intersections adjacent one to another, of which all the other adjacent intersections are occupied by stones of the same colour (in other words, it is an entirely empty region bordered by stones of one colour or by the edge of the board).
The black stones border a territory of 7 intersections. Note that the edge of the grid forms a natural border to the territory, but it is perfectly possible to have a territory that does not touch the edge of the board (imagine that the grid is a continent surrounded by the sea, that the edge of the board represents the shore, and that the stones represent the frontiers between the countries on this continent). 5. Progress of the game Go is played by two players. The player who starts the game plays with the black stones and the other with the white stones. Taking turns, the players place a stone of their own colour on an empty intersection of the goban, or else they pass. Passing is essentially a way to indicate to the other player that you consider the game to be over. It is important to note, incidentally, that a player is never forced to play a worthless move, unlike what can happen in chess, for example. On the other hand, if there are several moves still left to play it is usual, except perhaps in competition, to alert the opponent that this is the case rather than to take advantage of his premature pass. 6. Capture When a player fills in the last liberty of an enemy string, he captures it by removing the stones in that string from the goban. Further, when placing a stone, a player must not form a string with no liberties, except if, in doing so, he captures an enemy string. When a string has only one liberty, it is said to be in atari. Contrary to what persistent rumours may have us believe, you are in no way obliged to warn the opponent that he is in atari, and furthermore he is not under any obligation to defend against it - defending actually may be indispensable, mediocre or disastrous, because capture in go is only a means and rarely an end. That said, it is nevertheless advisable for beginners to warn each other and even allow moves to be taken back to avoid unwanted capture. Be careful, however, not to acquire thus the bad habit of taking back moves in general, which is, to be sure, strictly forbidden in a serious game.
The three white stones X form a string which is in atari (because it has only one liberty, at 'a').
If Black plays at 1, he fills in the last liberty of the white stones...
... and so Black captures the white stones and removes them from the goban. The placing of the black stone 1 and the capture of the three white stones count as only one move, as specified by official Rule 1 (alternation).
Here there are several other white strings in atari (can you see them all?).
All the Black moves 1 capture one or more strings...
... ending in the result of this diagram.
In this diagram, all the White moves 1 are illegal. White loses his last liberty while not capturing anything.
In this diagram, on the other hand, all the White moves 1 are legal. White loses his last liberty but captures one or more of the black stones. Provided that a string with no liberties is not created, it is perfectly legal for either player to play, now or later, on one of the empty intersections that have just appeared (although, more often than not, to do so would be a complete waste). Similarly, the rule makes possible relatively complex sequences of captures and immediate re-captures, such as can be seen in the following diagram (which should not be confused with the repeating situation that will be analysed in 8):
7. Life and death From the capture rule flows the concept of life and death: dead stones are stones which a player is certain of being able to capture without losing anything elsewhere, whereas live stones are stones that a player no longer has any hope of capturing.
According to the capture rule, White can play at 'a' and take Black. In this case it is said that Black has only one eye (the intersection 'a') and that he is dead.
White is unable to play at either 'b' or 'c' and so he can never capture Black. Black is said to have two eyes (the intersections 'b' and 'c') and to be alive. The most usual way of ensuring the life of a group is two have two eyes (or to be able to make them whatever else happens), but it is not the only way:
If Black plays at 'd' (or 'e'), White will play at 'e' ( or 'd') and capture him. Likewise, if White plays at 'd' (or 'e') Black will capture him. In other words, neither player has any interest in playing at 'd' or 'e'. In this case, the stones X are said to be alive in seki (impasse, from Japanese), and 'd' and 'e' are said to be neutral points. The rules make use of these definitions in order to be able to count up the score of a completed game quickly (as will be explained in section 9), but questions about life and death are not always easy to elucidate, and naturally depend on the strength of the players. In case of disagreement on this matter after the two players have passed (if, for example, one of the players believes he can still save a group that the other considers dead), it is sufficient to resume the game and to continue playing until all the positions have been clarified. 8. Repetition When placing a stone, a player must not restore the goban to a state identical to any state he had created before. The following diagrams show the simplest and commonest case of repetition, which happens to be called ko (the FFG recommends writing this as kô).
If Black plays at 'a', he captures the white stone X which is in atari. White cannot play back at 'b' immediately and take the black stone 1, even though it is in atari, because otherwise he would be reproducing the position of the first diagram. He must therefore play elsewhere. The shrewd thing for White to do, with his move elsewhere, is to try to create a threat sufficiently serious for Black to be keen to respond to it immediately and so not have time to play at 'b' himself. If Black does respond to the threat, White will now be able to play at 'b', since his previous move will have altered the state of the goban. Then it will be Black's turn to find a threat, and so on, so long as neither player connects. 9. End of the game The game stops when the two players pass, one straight after the other. The points are then counted up. Each intersection in a player's territory earns him a point, as does each of his stones still present on the goban. It so happens that it is an advantage for Black to play first. Therefore, in an even game, White also receives, by way of compensation, points called komi. The winner is the one with more points. Komi is normally 5 and a half points or 7 and a half points (the half point is used to avoid drawn games).
At this stage, all the territories have been closed off, and none of the borders can be captured by the opponent. This is the time to pass and to count up the points.
In practice, in order to shorten games without altering the score, the players may, by mutal agreement, remove dead enemy stones from the goban just before counting up the points, without having to add the moves necessary to capture them. In case of disagreement (which is almost always the exception), it is sufficient to continue the game until all the possible disputes are settled.
If Black plays at 'a', he captures the white stones X. If White tries to save them by playing at 'a' himself, Black plays at 'b' and captures them anyway. Since, as it happens, all the territories have been closed off, the two players pass. Black then removes the stones X from the board, and the points are counted up. Confirm that Black wins by 3 points and a half. Confirm also (this is important) that if Black plays the useless move 'a' anyway, he still wins by 3 points and a half. The same would apply, incidentally, if White played first at 'a'. Important note: in practice, it is possible to use a quick method of counting up which avoids having to determine the number of stones that are on the board. This method is described later in this document. 10. Handicap game Sometimes, a handicap is imposed on one of the players by letting the other, who takes Black, play several moves in succession at the start of the game. In this case, White receives a half-point (again to avoid drawn games) and a number of extra points equal to the number of moves that he was unable to play at the start of the game. So, in a 7-stone handicap game Black puts down 7 stones and then White plays. He thus a deficit of 6 points and will receive, at the end of the game, compensation of 6 and a half points. In a one-stone handicap game, White receives compensation of only 0.5 point.
Here is the start of a 9-stone handicap game. Black begins by placing 9 stones on the board. Only then does White place his first stone (move 1 in this example). By tradition, Black places the handicap stones on the hoshis. 11. Quick method of counting up To determine the score without having to count up the stones of each player still on the goban, this is possible:
In this way, in an even game, each player will at the end have used the same number of stones, which will all be on the goban; there will then be no need to count them. In a game with n handicap stones, the total of black stones on the goban will be equal to the total of white stones plus the n-1 extra points. So again there will be no need to count up the stones. In these two cases, the winner will be whoever has more empty intersections, not forgetting, in even games, to add the komi to the White total, and in handicap games likewise to add half a point to the White total.
This even game has just finished, and it is Black who has made the last move. During the game, Black captured 5 white stones, and White captured 2 black stones. The stones marked X are removed from the board because they are bound to be captured sooner or later.
First method of counting up in an even game The empty intersections and the stones that are on the board are counted up, the captured stones being ignored.
Second method of counting up in an even game Each player puts the enemy stones in his possession inside the territory of the other (the stones marked X). Black has 9 white stones to place (5 stones captured during the game, 2 stones removed from the board after the two passes signifying the end of the game, 1 stone that White has given him because Black played last, and 1 stone for the pass by White). White has 5 (2 stones captured during the game, 2 stones removed from the board after the two passes signifying the end of the game, and 1 stone for the pass by Black). Only the remaining empty intersections are counted up, with no need to know how many stones are on the board.
This game, which was at three stones handicap, has just finished. White played the last move. During the game, White captured 5 stones. Black captured no white stones. After the two passes signifying the end of the game, White removes the 2 black stones X from the board because he is bound to be able to capture them. The points can then be counted up.
First method of counting up in a handicap game The empty intersections and the stones that are on the board are counted up, the captured stones being ignored.
Second method of counting up in a handicap game Each player puts the enemy stones in his possession inside the territory of the other (the stones marked X). Black has 1 white stone to place (which corresponds to the pass by White at the end of the game). White has 8 (5 stones captured in the course of the game, 2 stones removed from the board after the two passes signifying the end of the game, and 1 stone for the pass by Black). Then only the remaining empty intersections are counted up, with no need to know how many handicap stones there were nor how many stones remain on the board.
All the calculations above sometimes feature deplorable mistakes in arithmetic. Experienced players have therefore invented methods of counting up which in no way change the score but make it easier to calculate and to check. Given that the number of empty intersections in a territory does not depend on its shape, it is simplest to rearrange it in order to create, for example, one or more rectangles. Similarly, a stone can be moved from one territory to another. We strongly suggest to beginners, however, that they do not "break up" the borders, as experience shows that it may then become hard to recall who owns such and such a space. Part 2: Official rules of the French Go Federation (FFG) Preamble The "official" text that follows is accompanied by several comments (in italics and on a pink background) intended to clarify certain points that might appear ambiguous. In theory, this commentary should not be necessary, or, more accurately, should be relegated to an appendix in the definitive version of this text. I have allowed myself to take advantage of the opportunity to give my own point of view on certain options. These personal comments will no doubt also disappear in the definitive version. 1. Equipment The objective of the game is to control a gridded space (grid, or goban) by means of markers of two colours that the two players place alternately on the intersections. In official competitions only square grids formed of 19 x 19 = 361 intersections (or, exceptionally, 9 x 9 or 13 x 13 intersections) are used. The markers take the form of white and black discs (called stones), available (in theory) in unlimited quantities. 2. Terminology We refer to a string as a (maximal) set of stones of the same colour connected via the paths on the grid, to liberties of a string as the set of intersections at which a further stone (of the same colour) may be placed without forming a new string, and to territory as a (maximal) set of empty intersections similarly connected and having adjacent to them only stones of the same colour. For even more rigorous definitions, see for example the New Zealand rules. It is reasonable to suppose that no dispute will ever arise on this point in an official match (and I personally would advocate a lifetime tournament ban for anyone who had the nerve to try it), but, in case of need, a rigorous formalistion of these concepts could always be added to the official rules. Progress of the game A game of go is a contest between two identifiable entities (the players, referred to henceforth in this document as Black and White), most often represented by two actual individuals, but team matches (rengo, pair go, etc.) are possible. Preliminary rule (determining the conditions of play) The game begins on an empty goban. Black makes the first move. It is possible to allow Black a certain number of prior moves (this is then referred to as a handicap game at n stones): he may then play n consecutive moves (and does not have the right to pass), whilst his opponent (White) is obliged to pass n-1 times. Finally, it is possible to give one of the players (most often White) compensation (in points) called komi. The FFG recommends setting this compensation at 5 points and a half (or 7 points and a half) for even games, and to n-1/2 points for games at n points handicap. These conditions of play must obviously be fixed before the start of the game. Likewise, the choice of colours may be decided by drawing lots or at the discretion of the event organisers (who will also decide the size of any handicap). The handicap stones (if any) may be placed anywhere, but the FFG nevertheless suggests keeping to the traditional placement on the hoshis, as detailed (and justified) in Part 3. The size of the komi in the case of a handicap has been selected to facilitate the quick counting up provided for by Rule 6a, but any other compensation (including giving it to Black) may be envisaged in order to allow finer adjustment of the chances of winning. Rule 1 (alternation) Each player in alternation places one of his stones (of which, in theory, he has an unlimited number) on one of the legal (see Rules 3 and 4 below) empty intersections and removes (Rule 2) any of the opponent's stones thereby captured (the totality of these operations constitutes what is referred to as making a move), or he passes his turn. Passing can be done by handing over a stone to the opponent, so that there can be no possible dispute. This is especially well suited to the quick method of counting up outlined below. Rule 2 (capture) If, after a stone has been placed, one or more strings of the opposite colour no longer has a liberty, these strings are immediately removed from the goban. It is traditional to keep possession of the removed stones (called prisoners), in view of the opponent, in order to facilitate estimation of the score. Rule 3 (suicide) If, after implementation of Rule 2, a string remains on the goban and has no liberties, that move was illegal. We say that such a move would be suicide, and suicide is prohibited. The French rules do not make any strict provision as to what should happen in the case of an illegal move. This point will be covered in more detail in Part 4. Rule 4 (repetition) If, after a player has made a move, the goban is in the same configuration as after a previous move or a pass by the same player, the move that has just been made (and which has reproduced this configuration) was illegal. This rule (also called the rule of "situational superko") can lead, in rare circumstances (triple ko, chosei, etc.) to considerable difficulties for referees which will likewise be discussed in Part 4. Rule 5 (end of the game) The game stops when the two players pass one straight after the other. It is possible, however, to tolerate a resumption (especially when using the quick method of counting up) if it is discovered that there remain on the goban some parts that did not have a clear status and could be affected by moves of one of the two players (this is the point mentioned below under the heading of Rule 5a). Of course, the refereeing problems that arise from this remain tricky. The fairest thing would probably be to envisage a uniform procedure (quick counting up or otherwise) in which, both players having passed in succession, one of the two can request a resumption (and it is then the other player who would resume). But, truth to tell, I am personally rather biased towards the simpler version described at the end of Rule 6a. However, attentive readers and connoisseurs of certain peculiar positions (such as molasses ko) will no doubt have noticed already that the important advantage that accrues in this case from the immediate cessation of the game as soon as the two consecutive passes have been made is lost if use is made of this possibility of resuming (and in fact it is impossible to count up a deviant position of this type correctly using the method of quick counting up). It would be necessary therefore to allow the players (or perhaps the referee) the possibility, when a rare situation of this type appears (triple ko at the end of the game is another possible example), of requesting strict application of Rules 5 and 6. Rule 6 (counting up) Once the game has ended, each player registers a point for each intersection of territory bordered by his colour, and a point for each stone of his colour in position on the board. Any compensation provided for by the preliminary rule is added to this score. The player who has the greater number of points has won. In the event of equal scores, the game is declared drawn ( which should not happen if the official recommendations envisaged in the preliminary rule are followed). In friendly games in particular, this way of determining the score can seem too fastidious, and the method of quick counting up explained in Part 1 will surely seem more natural and more agreeable to the majority of players. Here is the official definition of it: Rule 6a (quick count-up) First of all, Rule 5 must be replaced by the following Rule 5a: the game ends provisionally (we say it has been paused) after two consecutive passes. The dead stones (that is, those that the two players consider, by joint agreement, no longer able to evade capture) are removed from the goban, then counting up is done as envisaged in Rule 6. In case of disagreement, the game resumes as normal. It is also possible to count up even more quickly, if the players have kept track of the captures made during the game. It is then possible to stop the game after the last neutral intersection has been played, removing the dead stones and adding them to the prisoners, and to count up only the territories, reduced by the number of prisoners lost. It will then be necessary, in order to obtain exactly the same count as the official one envisaged by Rule 6, for a a compensation stone to be paid by any player who passes instead of making a move on the board. More precisely, the game must end with an exchange of consecutive passes in the form [Black passes; White passes]. Any other exchange requires a penalty of one point for the player who passes, including the most common final exchange [Black play; White passes], which must, strictly speaking, then be completed by [Black passes; White passes], this final exchange alone not being penalised. A simpler alternative formulation and one more in accord with the presentation given in Part 1 is: "The game ends with two consecutive passes. Any player who passes (including in this final exchange) pays a compensation point. If Black has made the last move, White pays an extra compensation point." The main disadvantage of the second formulation is that it seems even more artificial and unfair to players used to counting "Japanese style." It may be noted, however, that in the majority of cases, this is, for example, a good opportunity for White, instead of passing and "paying" one point, to repair (needlessly) a weakness in his territory, and this observation (which will be covered in detail in the final appendix) should compensate a little, psychologically, for the "injustice" of the passing penalty, since it allows White from time to time to connect without having to reflect on the necessity of doing so. Although the equivalence of the two counts is fairly easy to demonstrate (as has been done in Part 1 and as will be done in the final appendix), in cases of doubt (or stray prisoners) a player can always insist on application of the official counting-up method of Rule 6, which alone takes priority in competitions. This last observation, by the way, should in fact be specified at the level of tournament regulations, in a form such as: "The quick method of counting up shall be used, but in cases of disagreement the referee reserves the right to require that they shall revert to 5a and 6, or even 5 and 6." Similarly, in cases of disagreement on the status of certain stones (are they dead or not?), the game resumes (Black having first move) and is counted up by strict application of Rules 5 and 6 (but it is possible, especially in friendly games, to continue using the quick method of counting up, even when, in case of further disagreement after two more "passes" resuming the game afresh in the manner just described). It may finally be noted that in a certain sense the quick method of counting up is simply the Japanese method of counting up, but (in contrast to more debatable aspects of their rules which I will come back to in the following part) this method without doubt represents an improvement: experience shows that there is nothing simple in counting stones during the game (in order to estimate the score) even for Chinese players. Part 3: Comments and additions Historical remarks The rules used in France up to the beginning of the nineties were (but not really officially, and for good reason) those of the all-powerful Japanese federation, the Nihon Ki-in. In this era, the exact (and very poorly known in Europe) version of these rules included a considerable number of exceptions, and proved, even for Japanese professionals, extremely difficult to handle in cases of unforeseen disputes. A more up-to-date version was produced in 1989, and immediately drew severe criticism. Even now, there is still no satisfactory version of these rules, and it seems as if they are hardly ever applied, even in high-level Japanese competitions. During this period, we poor amateurs sometimes had to resort to conniving to make our "bent fours in the corner" alive. The restoration of order engendered by the Federation's decision to follow the (Chinese inspired) rules of the American Go Association (of which François Mizessyn and Jean Hossenlopp produced a French adaptation in 1994, which was very widely distributed) allowed us to put an end to these potential refereeing problems, but had a serious perverse effect: in normal positions, the Japanese style of counting up seems more natural, and the majority of strong French players did not wish to forsake it. Now the quick method of counting up allows it to be done easily - provided we accept paying the famous penalty stone, though to be sure, no "old" player willingly pays it, despite all the rigorous proofs they have been given that it changes nothing ("if it changes nothing, why change?" obviously being the inevitable response to this argument). You will find here a discussion of the various relative stances, and I hope that the well disposed reader, after having seen things such as "three points without playing", will allow of a certain superiority in the Chinese philosophy (and thus in the French rules) in cases of dispute. The equivalence of the two methods of counting up is, as already said, easy to demonstrate. But, on the other hand, it may not be true that the winner under Japanese rules is the same under French rules, and this is not due to the problem posed by playing the last stone. The situations where, in effect, the two rulesets are not totally equivalent are fairly numerous (even if they occur very rarely). By far the most likely is the appearance of points in certain sekis, which surprises many of the older players already mentioned, although none of them has yet tried to raise an objection on these grounds (though this is surely only a matter of time). Technical adjustments The exact formulation of certain points in the rules (for example, Rule 4, prohibiting repetitions) may cause surprise. It is a question there of minor adjustments which will shock only the purists seeking a formulation as concise and as homogeneous as possible, but which have been made necessary by the discovery of barely believable positions leading to serious difficulties. The novice reader would be well advised not to look too closely. The game is difficult enough to master already, and and it would be better to leave this nitpicking to specialists. On certain points the rules above are not complete. The positioning of the handicap stones is officially free, but the FFG recommends the use of traditional handicaps (on the hoshis). Everyone will do as they please, but it should be noted that not only does the traditional choice allow a unformity of teaching practice, but also very strong players (such as Fan Hui, for example) also recommmend these handicap-stone positions as optimal, even for players of the same strength! Likewise, the size of the komi is actually left to the discretion of organisers, even though it is fixed at a uniform (though not the same) value in Japan, Korea and China. Of course, you need to know what you want: to imitate oriental players (technically more competent) or to create a system of rigorous and fair rules (in which case the western mindset often gives better results). I have, for example, proposed a system of determining komi (the method of "komi bidding") which, in my humble opinion, ought to solve the problem entirely. The perverse effects that became apparent when it was first tried did, I confess, disappoint me somewhat (in 2005 I will try a new version suggested by Stéphane Dhur during the 2004 Montpellier tournament, but that's not to say it will then be plain sailing...). Part 4: Practical matters The official rules make provision only for the normal method of playing a game and establishing the winner. Practical methods of handling organisation and refereeing will form the subject of a separate document (after all, these problems are not really the concern of the players). However, some important matters are covered here. Irregularities In theory, any irregularity in the conduct of a game (illegal moves, bad behaviour by players, etc.) can (and often must) be penalised by the official referee (whose decisions, in a first approximation) have the force of law, even if, in accordance with European directives, it means envisaging a more complex procedure for truly important games, culminating for example in the setting up of an appeals committee). However, it is advisable not to penalise players too heavily for unwitting irregularities (we would wish neither to align ourselves with the world of chess with their complex considerations regarding the obligation to play a touched piece, etc. nor with the world of professional go in Japan, where any irregularity (such as immediate re-taking of a ko, for example) leads instantly to loss of the game). From this point of view, the trickiest situation for referees is connected to the prohibition on repetition (Rule 4) in cases which are not kos, and where it is possible to ask how a reasonable decision can be reached without a computer if, on top of everything else, the two players are in time trouble. It would perhaps be convenient to make explicit provision and set up a procedure for this case before the inevitable day when we have to make a ruling on a triple ko during a decisive game in the Paris tournament (presupposing, of course, that by then they are playing with French rules!). Counting-up procedure In the event of a dispute about the status of groups (the two players having passed as envisaged by Rule 6a), the game resumes (Black to move as White passed last). It is possible then to continue counting up by the quick method (and, indeed, this is the recommended procedure for a friendly game), but it is without doubt more prudent not to trust in that in a competitive situation and to apply to the letter the normal method of counting up (Rule 6). The two players must not forget then actually to capture all the stones they consider dead, the question of knowing how to count up in case of a further oversight being left to the discretion of the referee (but, in general, any stone remaining on the board after the two final passes is considered alive). Incidentally, it is possible to ask how other possible counting-up errors are to be refereed (starting with simple errors of arithmetic). Several stances are possible, from the one consisting of leaving the players to be responsible for all their actions (including declaring the wrong winner by mistake) to the one which considers that, once the final two passes have been uttered, the game no longer belongs to the players and any official may count it up. The FFG recommends that in cases of dispute, it is, in effect, the referee who re-counts the game (insofar as that is still possible). Other organisational problems The time for a game of go is always (since the end of the 19th century) limited: the most common method is a main time (for example, one hour per player), counted down by means of a chess clock or its equivalent, followed by a "period of grace" (called byoyomi, that is, in Japanese, counting off the seconds), corresponding for example to the obligation to play each move in less than 30 seconds, or instead to play each sequence of 15 moves in less than 5 minutes (Canadian byoyomi). The use of chess clocks to formalise these obligations is not perfectly suited to our game, though the important thing is that the passage of the time allowance is counted during the game, and this may be a matter of regret (in particular for lightning tournaments), though ways of getting round the constraints might perhaps be envisaged. The most important of the problems raised by these questions of time is the handling of the clock during the discussion phase allowed for in the case of quick counting up (to say nothing of what might happen if the game is resumed). Once again, the final decision will no doubt be left to the referee (or to the tournament organisers). I personally would advocate a rather strict version for cases of dispute (it is not so easy to play correctly in order to capture properly a group dead because of "bent four in the corner", and there is no reason why a player finding himself in this situation and in byoyomi should have his task made easier by the rules), but conversely the equivalent of a free-kick for obstruction against anyone who would wish to exploit the fatigue and byoyomi status of his opponent in order to oblige him to capture all stones trivially dead (at the risk of forgetting some, making a mistake, or losing on time). Be it noted in passing that French rules often allow playing (with no loss of points) fairly long sequences of moves of no value before the two final passes (fighting over an unwinnable double ko, or over a "bent five in the corner," for example) - perfectly legal behaviour, therefore, and of a type not really possible to penalise (although it would be necessary to be somewhat dishonest for an established player to play such a sequence by pretending to believe "it might work"). I suggest in that case, too, exhibiting considerable tolerance for any time overshoots by a player subjected to such an ordeal. Decisions concerning komi and handicapping are left to organisers (which has allowed me to experiment with komi bidding as mentioned above), but the FFG recommends, as far as possible, using an odd-numbered komi and to make alignment with the position of the strongest oriental players (and so to move to 7.5 points, which will be made official in 2005) and to give priority to even games in tournaments wherever possible, especially for games between strong players. Finally, the organisation of tournament events is also codified (for example, mention may be made of the MacMahon system). The FFG recommends using the Gotha draw program (by Luc Vannier) which is currently employed in all major French tournaments (except, of course, the Grenoble tournament), and can easily be set to meet most organisers' needs. Appendix A: New Zealand rules, a complete, rigorous and concise ruleset In order to show how it would be possible to formalise the French rules completely, here, by way of an example, is a translation of the New Zealand rules (with the kind permission of their author, John Tromp). [JF: we use the original English direct from John Tromp here, and John has also advised us that it is incorrect to call them New Zealand rules; he said, "I was co-author of the New-Zealand-rules-inspired Tromp-Taylor rules of Go, but certainly had nothing to do with the design and phrasing of the New Zealand rules themselves."] Their compilation in this form is extraordinarily compact and rigorous, but will maybe satisfy only mathematicians and other bizarre creatures! In other words, they need at least the following commentaries (by Bill Taylor). These rules are essentially equivalent to ours. The reader may enjoy verifying what exactly the differences are between both. The Logical Rules 1. Go is played on a 19x19 square grid of points, by two players called Black and White. 2. Each point on the grid may be coloured black, white or empty. 3. A point P, not coloured C, is said to reach C, if there is a path of (vertically or horizontally) adjacent points of P's colour from P to a point of colour C. 4. Clearing a colour is the process of emptying all points of that colour that don't reach empty. 5. Starting with an empty grid, the players alternate turns, starting with Black. 6. A turn is either a pass; or a move that doesn't repeat an earlier grid colouring. 7. A move consists of colouring an empty point one's own colour; then clearing the opponent colour, and then clearing one's own colour. 8. The game ends after two consecutive passes. 9. A player's score is the number of points of her colour, plus the number of empty points that reach only her colour. 10. The player with the higher score at the end of the game is the winner. Equal scores result in a tie. Commentaries 1. The grid of points is usually marked by a set of 19x19 lines on a wooden board. Each player has an arbitrarily large set of stones of his own colour. By prior agreement a rectangle of different dimensions may be used. 2. Using boards, colouring a point (intersection) black or white means placing a stone of that colour on the point. Colouring a point empty, i.e. emptying a point, means removing the stone from it. 3. Connected stones of the same colour, sometimes called strings, all reach the same colours. Reaching empty means having empty points adjacent to the string, called liberties. 4. Strings without liberties cannot exist on the board at the end of a turn. 5. For handicap games, the weaker player, taking black, may be given an n stone handicap; these are n consecutive moves played before the first white move. 6. This is the positional superko rule, that ignores whose turn it is. 7. For any specific move, at most one of the clearing processes can have effect; the first is called capture, the second suicide. 8. As a practical shortcut, the following amendment allows dead stone removal: After only 2 consecutive passes, the players may end the game by agreeing on which points to empty. After 4 consecutive passes, the game ends as is. 9. This is called area scoring. An almost equivalent result is reached by territory scoring where in addition to empty surrounded space we count opponent stones captured instead of own stones not captured. 10. By prior agreement, for games between equals, a fixed amount can be added to white's final score. This is called komi, and can be chosen a non-integer such as 5.5 to avoid ties. Appendix B: A little arithmetic - counting up typical positions under French rules The aim of this appendix is to give a rigorous proof of the equivalence of the two methods of counting up, then to show how the rules apply in certain more or less common disputed positions. The quick method of counting up The following analysis may seem unnecessarily complicated (the argument expounded in Part 1 is essentially correct), but will enable us, amongst other things, to resolve the trickier problem of comparison with the Japanese method of counting up. I have not, however, generalised it to the case of handicap games, leaving this exercise to the reader, who will see for himself that, with a suitable komi, it remains entirely valid. Let us begin first by showing that Rule 5a changes nothing, that is to say it is possible to get away in practice without having to capture dead groups before passing. In fact, any capturing sequence inside a closed territory, let us say black, will lead in the end to the placing of a certain number of black stones inside this territory and the disappearance of white stones. The final score will therefore be the same as if the white stones had been removed directly, since the total interior space is either empty, or filled with black stones, and so the count-up rule makes no difference between these two cases. It may be noted already that this leads (once the equivalence of the two count-ups has been shown) to a result that is by no means obvious in quick counting-up (or in Japanese rules), namely that any sequence played from dead white stones inside a black territory, including accounting for the capture of black stones, cannot change the score (unless a player passes or plays outside during this period, of course). Let us suppose, therefore, that the game has ended (with two consecutive passes), and that, with both players agreeing on the dead groups, the relevant stones have been removed. The Black territory is tb (we are talking therefore of the set of empty intersections surrounded exclusively by his stones), and the White territory is tw. There are Xb black stones (regarded as alive) and Xw white stones on the goban, and finally there are d empty intersections belonging to no-one (dames). There are, furthermore, pb Black prisoners (created during the game or during the final clean-up) and pw White prisoners. Finally, Black has passed x times (therefore paying x points), White has passed y times, and, if necessary, White has paid an extra penalty e (of one point) if Black played the last move. The score given by Rule 6a is (for Black) S' = tb - tw + pw - pb - x + y + e. We wish therefore to show that S' = S = tb - tw + Xb - Xw, as Rule 6 would have it. Now, Black has played Xb + pb moves and passed x times, White has played Xw + pw moves and passed y times; and Black has played one more "turn" than White if e = 1, or else the same number of moves. This gives us therefore Xb + pb + x = Xw + pw + y + e, and S' - S = 0. The details of this analysis will enable us to go much further. First of all, tb + tw + Xb + Xw + d = 361. Since d = 0 in the majority of cases, except seki or inadvertence by the players, it can be seen that S = 361 - d - 2tw - 2Xw is of necessity odd. More generally, under French rules, the scores change by two points at a time for a given value of d, an observation that we will often make use of again. Suppose then (as is the case in the vast majority of games with no inadvertent mistakes at the end) that d is even (there is no seki between two groups having only one eye). Then S being odd, with a komi of 5.5 points the score for White must be in the list (...; +2.5; +0.5; -1.5; -3.5; ...) and therefore: 1) changing the komi to 6.5 would not alter the winner; 2) if White loses, this is not "because of French rules" (if at any rate x = y = 1, that is if there have not been any passes other than the two final passes, which happens only very rarely), since he was losing already by at least half a point before paying the penalty e (you can find another calculation leading happily to the same result in my article on rules). Finally, it may be noted that if d = 0, S = 361 - 2tw - 2Xw = S = 361 - 2tb - 2Xb, that is it is possible to count up only one of the two territories in order to know whether Black has won, and even be satisfied with just that, using the following procedure: the Black territory is filled in with black stones and a check is made that the totality of the black stones present can now cover more than half the board. This method is not so bad, and even allows a very simple double check (the White score can be counted up independently), but requires some less than obvious adjustments for komi and the handicap, and also for the cases where d is not zero, which means it is not really to be recommended. The problem of the last move It is rather commonplace to fight over the last point (tedomari in Japanese). Since the above remarks show that the score varies by two points at a time, would these battles not be futile under French rules? To understand clearly what is happening, let us analyse a typical example (though a slightly simplified one) in detail.
In this position, we assume that the only thing left to do, before filling in the n final dames, is to play the boundary moves at 'a' and 'b', and that White has sente. If he plays at 'a' (the correct move), Black replies at 'b' (registering one point), White plays to the left of 'a' and the only thing that remains is to fill in the n + 1 dames, Black going first. On the other hand, if White plays at 'b', Black plays at 'a' (registering 2 points) and there remain likewise n + 1 dames, but it is White who starts filling them in. So, if n is odd, to begin at 'b' costs 2 points for White (one point of territory and the penalty for the last dame). If n is even, it makes no difference whether you begin at 'a' or 'b'. (By way of an exercise, you might try to analyse what happens if it is Black who starts. You should be able to convince yourself that, now, playing at 'a' or 'b' never changes anything.) All that may seem paradoxical to players used to counting up in Japanese fashion. Playing at 'a' is clearly worth one point more than playing at 'b' (although few players notice that this is true only if White plays first). But the battle for the last point overlaps with an invisible battle for the last dame, and it is by no means guaranteed that it is possible to win both! Let us begin by observing that if there are four similar positions on the goban, and if White makes an error four times, he loses 4 points: he in effect is creating four extra dames, which changes nothing in their final alternating filling-up. It is only if there is an odd number p of such positions that making an error in all of them has a total cost of p-1 or p+1 points, depending on the parity of the number of dames. Moreover, Black has no need for regret if White is not "punished" if n is odd. The analysis in the first paragraph shows that in this case the game was being played for only one point (and that therefore the "Japanese" sanction would not have been sufficient to cause Black to lose if he was not losing already). For the same reason, if there is an odd number of dames and Black begins filling them in, White loses nothing by needlessly protecting one of his weaknesses (but not several!) inside his territory, even if he does this before the filling in of the last dame. Once again, there is no "injustice" in that. On the one hand, a too prudent move by White would make him lose (under Japanese rules) only one point, and we have seen that that would not transform his victory into defeat. On the other hand, if White is mistaken and there is an even number of dames. the move now costs him two points (and perhaps the game), and that naturally enough requires a certain amount of reflection... The last ko The battle over the last ko (called a "half-point ko" or han-ko) is, in general, of the same type: there is one point of territory at stake, therefore 0 or 2 in fact, depending on the parity of the number of dames remaining. But there is an important exception. Supposing that Black, for example, has 2 ko threats more than White, and that 3 dames remain. Instead of connecting the ko, then filling in the dames (White therefore paying no penalty), Black must play the last dame, White takes back the ko, Black plays a threat (to which White responds) and takes back the ko, White plays the second dame, Black plays the last dame, White takes back the ko, Black plays his last threat and takes back the ko, White passes, Black fills in the ko and White passes again. It can be seen then that Black registers 2 points more than if he had simply connected (in general, it is possible to verify that in all similar cases, the "value" of the ko is 2 or 4 points, instead of 0 or 2). Other rarer situations are described in my article on the rules, and it would be similarly useful (although...) to study what would happen with indirect kos, staged kos, etc. Two examples of exceptional positions: "Bent four in the corner" and "Three points without playing"
Bent four in the corner The well-known position in the diagram above is not a seki. White can, whenever he wants, start a ko fight there. Japanese rules ("bent four in the corner") say that the black group is dead (without any ko proviso). Let us see how it is necessary to play under French rules against a Black player who refuses to consider the group dead (as is, by the way, his absolute right) and demands rigorous application of the rules. The analysis in the first paragraph allows us in any case to apply the method of quick counting up (which allows an easier check on the score). White begins by filling in all the dames (and pays after this any penalty point necessary), then eliminates all Black's ko threats by playing inside his own territory, which costs him nothing since, during this period, Black must do likewise or pay penalty points. Then, Black passes (and pays a point), White connects at the 1-1 point, Black pays another point, White puts Black in atari, Black captures four stones, White plays back at the 1-2 point, Black replies at 1-1, White takes this ko, Black passes (and pays a final point), White captures the group, and the two players pass. It is easy to verify that, in total, the score is exactly the same as if Black had resigned himself at the beginning to the loss of his group, and had accepted the count-up proposed by Rule 5a. In this latter case, Black would register 17 points (10 points of territory and 7 prisoners), whereas in playing out the sequence Black has had to hand over three points, plus 9 prisoners (do not forget the ko), and White has 9 points of territory but has lost 4 prisoners (and so has 17 = 3+9+9-4).
There is, nevertheless, a small flaw in this analysis. What happens if Black has indestructible ko threats, as for example in the diagram above? The answer is not so simple. White must now eliminate only the big threats (and Black likewise, by the way), then play the ko if he does not lose out in the exchange (and if he does, accept the seki, that is treat the black stones as alive). But we are now talking about questions of go and not of arithmetic...)
The rare position of this diagram (has it ever even appeared in a tournament? It is known in the Japanese rules under the name "three points without playing") exhibits a behaviour close to a seki: each player would prefer the other to play first. Black must play, however, because if both pass he does not score a point. Let us assume that there are no more dames. Black captures the four stones. White cuts (this is an example of playing under the stones, or ishinoshita. Black sacrifices a third stone at the 1-2 point (yes, yes: this hard-to-see tesuji will bring him 2 extra points), White captures (at the 1-3 point), Black puts this stone in atari, which White cannot save because of the snapback; White connects at the 2-1 point, Black captures, White plays at 1-1, Black connects, and White passes and pays a final point. The balance sheet: Black has registered 3 points (which may seem totally contradictory to our initial observations on the fact that the scores change two points at a time - except that here d has changed, since the initial liberty at 1-1 has disappeared in the course of the sequence). By way of an exercise, the reader, if not yet exhausted, may verify that the result is exactly the same if the colours are reversed. A useful theorem to know for this type of analysis states that the value of a position never changes if the colours are reversed, but the general demonstration of this too is left to the reader. JF: The above text was written by Denis some time ago. A very slightly modified version of the FFG introductory text has since appeared, translated here (and kindly formatted by Richard Mullens). |
