Superko/Discussion

Sub-page of Superko

Robert Jasiek: Some players do not like superko because the rule has been explained to them badly. It is better to explain the rule well instead of replacing it by much longer rules.

Superko can be written as rule like this:

A play may not recreate any earlier position.

If superko is not used, then the basic-ko rule is as follows:

A play may not recreate the position just before the last opposing move.

As you can see, the basic-ko rule requires a longer rules text than the superko rule. Even worse, if superko is not used, then more rules are required:

Exceptionally and immediately after a play it lets end the game with the result tie / win of Black / win of White if the position just after the play is the same as a position just before an earlier move and if the number of white-stones minus the number of black-stones that have been removed during plays since the first occurrence of that position is equal to / greater than / smaller than zero.

In practice using superko instead of several ko rules including the basic-ko rule leads to strategic differences only in every 5,000th game. (Probably rarer; I have never had any such game!)

mgoetze: This does not mean that there is no other reason to dislike superko. I have been given the following arguments against superko by Olli Lounela?:

  1. It is unimplementable. This I might argue with as the use of PDAs and laptops to record games becomes more widespread.
  2. It can make the game strategically unfathomable. Here I must defer to the experience of a player who is stronger than me and has actually been involved in a game where such rules were relevant.

Andrew Grant: It can also lead to some strange life and death decisions.

Hu: With modern algorithms (ref. Kombilo), the superko rule is quite implementable.

See Superko Anomalies.

John F.: I have a sense that a large body, maybe the majority, of people who reject superko, or other attempts at rules revision, do so for the rather vague reason that they don't like rules that differ from the ones professionals use. It may not be rational but it's real. It's rather like the situation where most amateurs switch komi as soon as pros do.

RobertJasiek: I have never thought of this. Instead my experience is that most players do not read the rules, like best what they were taught as beginners, and reject to reflect what they were taught for as long as possible. John, how do you come to your impression?

John F.: Robert, I already gave part of the answer - the way amateurs instantly follow pro practice on komi despite endless discussions among themselves on the true value of komi. I also base myself on nearly 40 years of reading and talking about the game. But I think it's also to do with status. As I said, this can be slightly irrational. If you are ill and have the same illness as a friend previously had, it doesn't matter how much your friend tells you about how to cure it - you listen, but in the end you go to a doctor (who may never have seen that illness).

PS: I agree that most of us do not read the rules - nor do we read the manuals how to use new software. It may be regrettable, but it's a fact of life, and the software companies appear to have accepted that and have tried to obviate the need for manuals. Maybe rulesetters need to take the same approach.

Robert Jasiek: This approach sound suitable for running atomic power stations. However, a game of go is neither like a machine nor like a software. If you mishandle a machine, you turn on and off power, if you mishandle a software, you close and open it or the operating system. A go game cannot be initialized. Out of fairness towards both players, already created facts need to be considered for the result.

Go is not a non-rules game and a software is not a non-algorithm process. Even if rules or an algorithm are known only verbally to the user, his actions are subject to the rules / algorithm. There is not a sort of romantic freedom. All that there can be is a lower or higher degree of strictness. A lower degree means a higher frequency of results that are not based on the principally highest possible fairness because already created facts need to dismissed the more often. I like fairness as often as possible, therefore I like a higher degree of strictness.

John F.: But there are more factors than fairness and strictness involved. Complexity and tradition among them. So other mixes and this other choices are valid. Not only that, other choices seem to be in the majority. If you want to change that majority, I think you need to ask (empathetically) why they hold their views, and not just keep telling they are irrational and wrong.

Robert Jasiek: Of course, there are more factors. If we follow the majority, then we play without written rules, without referees, and follow the tradition each player has as its very own tradition, i.e. what he perceived as a beginner :) I have asked quite a lot for their reasons and almost all answer that they like what they were told as beginners and that otherwise they do not want to discuss rules at all. This is pretty irrational, I'd say. It is also not social enough to allow tournament play. The majority of organized players is not tournament players but club players. Tournaments, however, do not work if each club imposes its special club tradition on a tournament. So opinions become relevant for tournaments only if there is an agreement to create something common beyond each club's tradition. This excludes the majority's opinions since the majority does not want to discuss such.

John F.: In nearly 40 years of tournament play I can recall only the tiniest handful of problems, so I can't accept your view that current tournaments don't work. The problems I can recall are: (1) a player not accepting that bent four was dead; (2) suspected collusion leading to a jigo and a distorted shareout of prizemoney; (3) total confusion over an attempted introduction, and early abandonment, of auction komi. I've heard of mix ups and miscounts because of changes of komi not being clearly announced.

Against these very scant incidents, I could cite at almost every tournament problems with people turning up late, people with dodgy ratings, draws having to be redone, rules on komi and byoyomi not being announced or properly explained. It seems to me that the vast effort spent on rewriting game rules could be better spent on getting people to behave more considerately towards tournament directors. If they could be persuaded to do that, they might then more easily accept their authority on the very few occasions where a game-rule dispute occurs. I have to say, too, that I regard reading the fine points of game-rules to see how they can be exploited as NOT behaving considerately towards tournament directors.

Robert Jasiek: I have not said that tournaments did not work. Why do they work? Because each of them already has a standard that overrides every club's own tradition.

During the some 10 years of my tournament play I have seen about 0.5 disputes per round of a tournament with an assumed average of 100 players. This is more than your experience of a the tiniest handful. I should say that I am curious to watch disputes and so I tend to notice many disputes others would not notice. I do not know whether disputes are less frequent in British tournaments than in German or EGF tournaments. However, I have had the impression that Eastern European players are more fond of disputes than Western European players. Since only few Eastern European players might enter British tournaments, this could explain at least part of our different observations. I agree and have always said that tournament conditions (work of the tournament organizers, respect towards them, their announcements, and the tournament rules) cause most of all disputes (my estimate: 75%). Of the remaining 25% (disputes related to rules of play), 90% are caused by unclear or too difficult rules.

Go strategy is the application of the rules. Each such application should observe the "fine print" of the rules. Otherwise a line drawn between rules to be applied and rules to be broken intentionally would have to be arbitrary and players would be punished for applying the rules. If rules shall not be applied, then such rules (within a rules text) ought not to exist in the first place. At the very least, such a rules text ought to state explicitly the rules that are expected to be broken intentionally if that should be their purpose. Rules do not violate human rights or the like but they behave like laws. Mankind has adopted laws to avoid anarchy and not to exercise anarchy formally. Such is a favourite means of totalitarian regimes. The world of Go tournaments is not a place that should support such means.

If rules are bad, then they can be replaced by better rules. Instead of suggesting intentional, systematic breaking of rules, you should support improvement of the rules.

John F.: For people to say they disagree with your emphasis on rule change does not mean (a) that they say there should be no rules or no observation of rules or (b) that they support infringement of rules. There probably is a difference between Britain and (the rest of) Europe. European law is largely governed by the Napoleonic Code approach in which every conceivable thing is thought of in advance and a law written to cover it. In Britain we have case law and operate on a case by case basis. This is changing in the modern EU but I think you can still see the fundamentals of this approach at work in the way we resolve tournament disputes - something I believe we do rather well. This is close to the Japanese approach and so lots of people here can empathise with their attitude to rules. There are big areas outside Europe that share the traditional British approach to rules, to some degree or other - not least America. I don't think it's any coincidence that the bulk of the rather small support you get from people outside Europe for rule change is from mathematician's and the like who esteem concepts such as elegance. Nothing wrong with that, but it's not the majority view.

As I've already said, if you want to sway the majority, you really do need to understand why they speak/act/lurk they way they do, and approach them on their own terms. It is, however, my view that you cannot sway them directly, especially not by insistently telling them they are wrong or irrational. My view of the best (only?) way for you to achieve your goal is to persuade professionals in Japan to change. The western majority would then convert overnight.

RobertJasiek: It is probably a good guess that British jurisdiction has a solid history. IMHO, most Western Go players would not know the fundamental principles of Japanese law. Concerning support of me from outside Europe, your conclusions are too rash and simplifying. E.g., the AGA uses reasonable rules and so why there not be support for them from US players? E.g., mathematically interested players are more often rules fans than other players, so why I not get an above average support from such players? Anyway, how could you assess "the majority view" of non-European, Western amateurs on go rules? I have talked with maybe hundreds such players about rules but still I could not determine such a majority view. I wonder how you could. Have you inquired thousands of non-European, Western amateurs about go rules?

You wonder about my strategy towards area scoring? Simple. I try to convince teachers because a beginner's first rules knowledge is decisive for his preference. Of course I also try to convince professionals, but, as we all know, their speed of changing rules significantly tends to be every 40 years:) I am patient;(

John F.: Robert, you already know I don't go round asking people for their views on rules, because I have no special interest. I simply observe what happens in most tournaments, and read in most books. It is true that there has been some variation in recent years, notably the use of Ing rules, but we all know that's venal and not from conviction.

I wonder whether what you call support for your rules is really much more than polite interest. Talk is cheap, after all. Surely the fact that you state that you are the one in the position of being patient demonstrates that you are on the minority side.

Robert Jasiek: What do you observe in tournaments? You claim something for non-European tournament players but what do you observe in non-European tournaments? Aren't you guessing wildly?

You know very well that Go books for Westerners were dominated by Japanese until very recently, even Chinese Go books. Therefore you should not conclude about go rules teaching from the Go books so far.

If you want to judge support for one or other rules, then it is fair not to overlook support for the other (Japanese style) rules. Typically, the latter is like this: "Can you explain me Japanese rules?" Silence. It is not even admitted that the person favouring Japanese rules cannot explain them. He says nothing. So much for the typical support for Japanese rules. It could hardly be weaker.

I do not enter a meta-discussion about a relation between patience and minority or majority. However, there is no doubt that in Germany I am still in a minority; currently most German players prefer Japanese rules. There is also no doubt that currently more Japanese professionals prefer Japanese rules; so if you included me for the sake of counting, then I would be in a minority there as well. I guess that I would be in the majority in China, the USA, New Zealand, the world. The world is unfair, of course, because that is simply a matter of greater population.

John F.: I'm sorry, Robert, but I don't think you can claim to be in a majority anywhere, unless you are shifting ground. If you ask someone whether Japanese rules are logical and they say no, or if they can't explain rules to you I don't see the link that entitles you to say they support you. I would say Japanese rules are illogical. I would admit I can't remember some rules. If you then said I supported you, I'd say you're wrong because I don't feel, emotionally or objectively, that I'm supporting you. All I'm doing is agreeing with two out of a possible zillion statements.

The question I think you need to be asking is more on the lines of: "If I, Robert Jasiek, convince you that rule set X is flawed, will you change to another one that I, Robert Jasiek, have devised?" Experience seems to show that next to nobody will, and that's what I mean by lack of support. This is not about Japanese rules only. Amateurs who prefer Chinese rules are aping Chinese pros, not supporting you. If Chinese pros switched tomorrow to Japanese rules, I would expect the majority of Chinese amateurs to switch, too.

New Zealand is too small to take into account.

barry: Yes, just ignore us. :)

My understanding of the AGA rules is that they apply mainly to tournaments. Clubs generally follow one of the pro sets, but the USA has been in a special position in having fairly large numbers of Chinese as well as Japanese and Korean immigrants. Their tournament rules have been useful as a compromise that brings these Orientals together. Although rules theorists have informed the AGA rules, my understanding is that the real driving force was cultural, not rules-theoretical.

barry: I would put it the other way round. The driving force was theoretical, but the cultural mix in the USA made it implementable. The initial push for change came from theoreticians. The possibility of unifying the two counting systems made it politically justified.

(I have this impression from reading the AGA archive collections.)

Bill: I believe that one factor in the adoption of the AGA rules, which is of some importance in the West, and particularly in the U. S., is that they are easy for rank beginners to apply without assistance. Many go players in the U. S. have no local club and play only among friends who also learned to play from a pamphlet or book. They need rules they can understand and apply. The AGA rules are right for them.

Robert Jasiek: If you want to assess my being in a minority/majority only for my living environment, then there is no point in considering whether I have support in non-European countries. - If somebody cannot explain Japanese rules, then I do not say that they support me but that they do not (or hardly) support Japanese rules. - If you say something about Japanese rules, then I do not say that you support me about Area Scoring rules.

Who has said that I ask for usage of MY ruleset? I am not Mr. Ing. I suggest usage of particular CONCEPTS like simplicity, clarity, completeness, Area Scoring, Superko, etc. Some of my rulesets are models that use these concepts.

E.g., the Japanese 2003 Rules are not invented to be used by amateurs but - besides other purposes - to explain Japanese rules so well that one realizes that using them at all is not particularly desirable:)

Tournaments. I am concerned mainly about the rules used in tournaments, by far not so much in clubs. In clubs it does not matter if I lose because of a rules problem. In tournament games it does matter.

You might be right about significance of the cultural aspect for adopting the AGA 1991 rules.

barry: For thousands of years go was played without any written rules. It led to variations in the rules used around the world. There were always disputes, but mostly in amateur games. Professionals had a more uniform appreciation of what rules they played by, and in any case the number of professionals in the world prior to 200 years ago was very small.

Now the world is changed. There are many more professionals in many more countries and even amateurs are more often playing in different countries. In a local tournament everybody is likely to have learned from the same source, and new players can quickly learn the "house rules". In international tournaments with larger prizes at stake, it is harder to explain the variations in rules that are accepted, so unambiguous standards are needed.

This is largely the difference between John's and Robert's approaches. John has one foot in the 12th century, and sees things on a local scale. Robert is more in the modern chess tradition. Because of differences in traditions what is/isn't allowed needs to be spelled out.

Robert Jasiek: barry, do you say that there were reported quite some amateur disputes prior to 1900? Any details?

John F: Barry - small is beautiful! Yes, I agree with you that in the USA the rules theoreticians came first (Olmstedt & Robinson, was it?), and also that quite a bit of notice of their work was taken in Japan, but I would still rate the cultural/political element as dominant.

Robert - I didn't mean to imply Ing-type behaviour, though I'd be surprised if there wasn't a big debt to you in some future rule change.

Barry mentioned my "approach". I don't dispute his view of where I'm coming from, but that word may suggest I have a destination in mind. I don't and I'm often unsure whether Robert understands that. I am not advocating any rule set. I am a follower. If the rules are changed tomorrow (by the pro organisations), I'll change, too. I have no interest in rules in the sense meant here (I did compile a long feature on rules incidents on the latest GoGoD CD, but that was from a journalistic stance).

On the other hand, I can see perfectly well why Robert and others find the current game rules irritating, unsatisfying. I can and do admire the arguments they make about them (though I rarely understand).

Where I take issue, in practical terms, is in two areas. One, as I've stated above, is that I think the effort is misdirected - you need to convince the pros, not r.g.g or SL.

The other area is that that some people who have this special interest extend that interest to tournament rules (Robert is one, I know, but I've never had the impression that Barry is, for example) and then in turn feel that it is legitimate to use the esoteric points of either the game rules or tournament rules (which they may even have helped draft) as a means of winning games. I've read Robert's descriptions of some of the disputes he's been involved in. If I had been the other person in those games, as soon as it became plain that he was arguing about rules esoterica, I would have said (as I have done elsewhere), "If you want to win the game that badly, have it - I resign." Even though I know, in Robert's case, that he's really just searching for an ultimate truth, I'd still regard his behaviour as either unsporting or anti-social (an old debate we've had, which we better not resurrect!). Robert's stance seems to be that the result is the important thing. To me, it's more important that at the end of the game the two players should feel content with each other - they don't have to become friends, but they should be happy to play each other again.

That stress on harmony is, of course, inherent in the way Japanese play go, and for that reason Japanese rules in the widest sense - not the game ruleset - appeals to me. Conversely, the frequent disputatious behaviour by Chinese pros puts me off their rules, again in the widest sense, even though to my eye the Chinese game rules have lots of advantages over Japanese game rules.

I don't think this is a fondness for tradition, a word that is often associated with me. It's the behaviour pattern, which may or may not have roots in Japanese tradition, that I like. I wouldn't want anyone to be starry-eyed about Japanese players - club players can be very rude and obnoxious - but my ideal of go behaviour exists somewhere in Japan.

Robert Jasiek: IIRC, Olmsted is the spelling. - On average, games without dispute are friendlier than games with dispute. This is another reason to have rules that only rarely lead to disputes. However, the current Japanese rules text follows a different pattern: It seems to be designed so that one can study the players' behaviour in case of disputes.

TJ: John F's comments above about preference for Japanese rules expresses, very well, something I've been trying to express many times myself. Thank you. I grow tired of feeling that I must defend my preference of Japanese rules to anyone who likes to have things spelled out and codified to the very minutiae. When I say that I like Japanese rules in a conversation on a Go server, there's always someone who assumes I just don't know REAL rules. Usually they bail out and stop responding to me when I start trying to explain why I like them, and what, exactly, they are, and how disputes are actually handled and why I like that. I'm not saying that Robert thinks or acts this way, of course! I'd just like to ask that he stop spreading, as a side-effect of his argument, the impression that one who disagrees with NZ, AGA, area scoring, super-ko, or any other of the more "Napoleonic" rule sets... in short, that one who prefers Japanese style rules and scoring must not know what they're talking about. I may not be as good at the game, or as experienced in tournaments as, well, anyone else on this page, but I can (and I hope do) still have an informed opinion on what rules set I prefer from a philosophical or aesthetic stand-point.

Robert Jasiek: I have not said that all who prefer Japanese rules cannot explain them but that the most frequent reaction of those who prefer Japanese rules is silence when asked to explain them.

TJ, what are your motivations (philosophical, aesthetic, or other) to prefer Japanese rules?

TJ: Very fair question. First, I would like to make sure to say Robert there knows more about rules, including my preferred rules, than I do, and that I'm under no illusions that I understand things better than he. My objection above was mostly due to seeing someone very knowing and esteemed possibly contributing to people who know even less than me rushing to judge me as an idiot because I disagree with an authority on the subject, as Robert is. There, that's out of the way.

Motivations for liking the Japanese rules are, in my case, I think, mostly aesthetic and philosophical in nature. I can only speak for me. I don't like a preponderance of rules. Platonically speaking, overall I identify more with The Republic than The Laws. Ethically, I don't enjoy the philosophy inherent in the rule of law. Developmentally/ psychologically, I don't think the rules of Go need to be grounded in formal operations, but can leave that behind as unnecessary. That's all ways of trying to say the same thing, trying to put my finger on an aesthetic preference.

I think that a system which asks that the people within it behave rationally is less unwieldy than a system which attempts to block every recourse to irrationality. Blocking routes to irrationality treats people as immoral creatures with base impulses which must be controlled by authority and more rational institutions placed above the individual in every area of his/her life, every moment, all the time.

For the game itself, I like the idea of things being based on agreement. Agree that a game is scrubbed due to triple-ko no one wishes to break. Agree that bent four in corner is dead. Agree on things until you cannot agree, then worry about solving the disagreement. This seems more trusting to me, putting more faith in humans to rise to the occasion.

To do with territorial scoring: I like the minimalist nature of such play. Dame need not be filled, dead groups should not be removed even if there's no other point on the board (no freebie safety move at the end). From this arises possible dispute, as mentioned above with the dead bent four, of course, and a system for conflict resolution of some sort needs to be in place. I hope I'm not wrong in thinking resolution by proof (actually playing out dead bent four on a separate board, for instance) is still a (little known, it seems) part of the ruleset you mentioned was changed just in 2003? Irregardless, I find the possibility of dispute to be part of the charm; the possibility of dispute must be present in order for the joy of avoiding dispute to be claimed, personally, by the participants. Leaving in the possibility of such everyday transcendental moments is maybe my strongest, if impossible to prove, argument in favour of such a ruleset.

I apologize for any error of fact in this little statement about my own odd affection for a set of rules I certainly don't understand as well as many here. I hope I'm not proved wrong in that the rules I like contain such aesthetics and philosophical overtones, for that would make me quite sad. If I'm to be proven wrong, I do hope that stress is on how other rulesets can maintain the ethereal something touched on while fulfilling any need for standardization.

I hope I've expressed that "ethereal something" well enough, I've not tried to put it into words before. It was difficult. Perhaps the ethereal nature of this preference is not only my own, and others fall silent not out of ignorance, but rather the difficult nature of forming such a sentiment in coherent language.

Robert Jasiek: TJ, your philosophy is a refreshingly new one, I have not seen before, except tiny aspects of it. What follows is my doubts.

You describe a philosophy that would apply much better to pure Territory Scoring rules than to a Japanese version of Territory Scoring rules. Japanese rules contain more law than you like, lead to filling of dame where you like them remaining unfilled, lead to removals of dead stones where you like them to remain on the board. IMO, you would like a hybrid of the Territory Scoring 2003 Rules II and Japanese rules more than any Japanese ruleset. The Territory Scoring 2003 Rules II (II stands for letting a stone in snapback be alive instead of dead because it can be captured) omit all the legislative exceptions of Japanese rules (so far that, of course superko is used, for which reason I guess that you prefer a hybrid) and allow dame to remain unfilled (because there is no technical rules seki but asymmetrical sekis have an internal territory difference other than zero).

That dame need not be filled under Japanese rules is heard from time to time as a false argument of preference for them, false because under Japanese rules dame are filled, just technically optionally after instead of necessarily before passes.

I wonder: Do you let dead stones remain on the board while you count? Otherwise you pretend a philosophy that you do not follow. Rulesets like the Logical Japanese Rules of Go, the Japanese 2003 Rules, or any Territory Scoring 2003 Rules are more generous here since they do not require a particular counting method, i.e. during counting dame can remain unfilled (except J2003, where they should be filled) and dead stones can remain on the board.

Your philosophy about agreements/disagreements, disputes is also available under most Area Scoring rulesets, but I understand why it can be more convincingly applied under Territory Scoring rulesets.

TJ: From what you're saying, it seems that I DO in fact prefer Japanese Rules because they are the closest to what I like that I know of under any common rulesets. Common in that I, who play almost exclusively online, know about them. Reflecting on this, I think that I'd perhaps drawn the conclusion that only Japanese rules uses territory scoring, and that it did this quite well and logically (what I thought about this logic is found below).

I don't fully understand a couple points: Why must dead stones remain on the board for my philosophy to be true? Rather, how can they remain on the board. On the occasions I get to play a game on an actual goban, dead stones get removed and, with prisoners, fill opponent's territory. This for the sake of simplicity in counting, not as a rule. It seems that, short of this, they'd at least have to be made prisoners for any ease in counting at all, in name if not in fact by being removed from the board. Perhaps my outlook on this will be better clarified below, when I talk about what scoring rules I do like.

A stone in a snap-back lives? This one really confuses me. If it would die if it was played out in isolation, I'd think it should be dead.

I've never had a liking for any rules that counts any points in seki, although this one's quite possibly because I certainly have never taken the effort to learn any such rule.

What I thought I'd like about Japanese rules, that apparently isn't contained in Japanese rules anymore, if it ever was, was that the so-called special rulings were, I'd thought, no longer a set of rules set out as being chiseled in stone arbitrarily because some rule is needed, but were consequences of units in question being played for in isolation, which, upon request in disputes during the scoring phase, must be done. This meaning that only internal ko threats were valid ones. If the unit could be killed, but then replaced by a living group, they still live. The opponent should have killed them if it was to his/her benefit, too late now.

Without referring to any rulesets to hybrid, because I don't know enough about these rulesets you mention, what I'd like, or THINK I'd like, as I liked it about what I thought were Japanese rules:

Territory storing, points in seki don't score, dame do not get filled (I knew it was Japanese rules to fill them, but I try to ignore that rule as being annoyingly arbitrary. Filling in the scoring phase is just a way to make the counting easier, though I never thought it made it much easier.), and a unit is dead during the scoring phase if it can be captured in isolation, to be proved if there is an honest dispute. Oh yes, and no super-ko, the problems of repetition being solved by honourable agreement should such a sequence arise.

I had thought that this best joins together minimalism of territory scoring with a mechanism for dispute resolution, all with the fewest possible rules which must be formulated and known for the sake of playing at all. Is this ruleset somehow naive or impossible to play under? If possible, is it somewhere formulated?

It is sad that what seems to be my shallow or incorrect understanding of Japanese rules does indeed seem to be what I like, but the actual Japanese rules still seems to be the closest to ideal for me which is available in online Go, at least. I'm hoping there's something closer already being used somewhere; if not, Japanese rules still seems to be the ruleset I'll have to use as being the system closest to my ideal which is actually used in play by any significant number of players.

Three more stones strength, and I'll be in a place where I'd decided some time ago that I'd like to try to spread Go to my fairly go-free locale, so it looks like, like it or not, I should be finding a complete ruleset to follow soon, and learn it at a less naive level. This may take some time yet at my current rate of advance, but any help on picking a ruleset I'd like to learn and implement would be appreciated, since it seems I knew even less than I thought I did!^^

RobertJasiek: Online go rules are not Japanese rules even if they are called Japanese rules on every go server. They all make that same mistake. So what you like is some sort of online Territory Scoring rules. On some servers like IGS, dame need not be filled because sekis ARE scored. This is a further derivation from Japanese rules.

Dead stones must remain on the board according to your philosophy because it asks for minimality - it should end before any superfluous and extra handling like filling dame, removing stones, rearranging stones. With dead stones on the board, the score is as given as with them removed from the board. Just the counting method will be different. E.g., an intersection with a dead stone counts 2 instead of 1. Online this is even done by the server. All you need is - what you said to like - AGREEMENT about which intersections to score for whom. Online this could be done by letting both players click on each dead string. This is already enough information for the server program. The dead stones are not removed but marked as a sign of common agreement, as you like.

If a snapback stone lives, then its player does not add another stone. If a snapback stone is dead, then its player should add another stone shortly before passing to actually remove the opposing string. Thereby the snapback is dissolved and the stone becomes alive.

What does your go server do with sekis during scoring?

I do not understand your "What I thought I'd like about Japanese rules" paragraph.

When you think you liked Japanese rules, you err IMO. You imagine something that they are not.

For teaching I suggest simple rules but you do not like them because there is the danger that your pupils might actually get hold of a complete written rules text they would understand;)


dnerra: Why is it unimplementable? I don't see a problem there. (Btw, it is pretty much impossible to take care of in reading search trees in go programs, as soon as one uses hash tables (i.e. always).)

RobertJasiek: Superko is as much or as little unimplementable as the extra rules besides a basic-ko rule that also need to refer to recreation of a position. In both ko rules designs (superko or basic-ko plus extra ko rules) one has to be able to recognize recreated positions. In both designs the problem occurs in only every 5,000th game or rarer. "unimplementable" is not a suitable description.

mgoetze: They differ in the amount of previous positions that a player is required to remember.

Robert Jasiek: In practice this is true only in every 50,000th game or rarer. Why? In all other games either cycles with more than 2 plays do not occur or cycles with more than 2 plays have a ko threat and answer in between. After each ko threat answer the position has changed. So one needs to remember only those few positions that are played during the cycle. In practice almost all cycles are of length 3, 4, 6, or 8. In every 50,000th game or rarer longer cycles might occur. Do you really insist that you cannot remember the last 8 moves of a game?! Can you not live with 1 exciting game in your life where winning depends on your ability to remember more than 8 moves?

RobertJasiek: It is right that strategy differs. One should note, however, that it differs significantly only every 5,000th game or rarer. In almost all practical cases of these rare positions, strategy has the same behaviour like in a basic ko fight. E.g., under superko a triple ko is fought like a basic ko: One play in the basic ko, then ko threat and answer, etc. Cases require new strategic insight occur in every 50,000th game or rarer.

mgoetze: How many games of go are played every year? Surely more than 50,000.

Robert Jasiek: Is your argument that no such game may ever occur? It should be that each player "survives" the 1 game in his life in that he himself is faced with such a strategically unusual situation. Don't tell me that we would need several rules instead of one rule because each player has 1 difficult game during his life. I would be glad to have at least that 1 game!

TJ: Perhaps I have an odd point of view, or misunderstand the rules being discussed, but I don't like superko or pretty much any "ko rule" beyond basic and necessary rule-of-ko. Then just add agreement to the mix, and forget about it.

In chess, repetition is a draw, but three times, to avoid error. Superko is too unforgiving, and even less respectful than chess of all things in this matter (pardon me for being facetious there). If, in a game or mine, we repeat the position, but it is realized that the first move in a long sequence can be changed and my opponent can still win, I'd not want to claim a draw, or force my opponent to break the pattern and lose, or whatever. Going through it three times is still darned silly and arbitrary, too, really, I always thought that about chess. But, if it is agreed that no one wants to break the repetition, basically what happens in chess if a position is repeated three times, call it a tie, jigo, no result, invalidated, or whatever you will, and throw a party or something to celebrate the rarity of such an occurrence in go, as opposed to chess. Whoever repeated first can maybe buy the first round. Then repeat rounds until everyone agrees it's time for bed.^^

Oh, yes, the other thing I find interesting is the case where there is a long repetition and a change in score occuring every repetition (one side gaining prisoners in exchange). I actually HAD a game like this one. If superko was being used, I would possibly have to be the one to avoid repetition, and against my own interest? That's silly, if I'm understanding correctly and it's true! I'm happy with the state of things, since I can repeat the shallow position (not counting change in score) until I've gotten an insurmountable lead. My opponent should be forced to break the cycle (or resign), since I'll never agree to no result or jigo if I'm happy with the current state of things.

Robert Jasiek: Superko is as much a basic and necessary ko rule as is the basic-ko rule or the Fixed Ko Rule. - The basic-ko rule is, to discuss your argument, too unforgiving. Why don't you proclaim the Fixed Ko Rule (it allows a cycle and only then prohibits further recycling, even if the cycle includes only 2 plays)?

Bill: I had always wondered about the three repetition rule in chess, too. Not too long ago a reason came to me that might be right, historically. Three repetitions of a whole board position guarantees the repetition of a whole board situation, i. e., with the same player to move. So the three repetition rule in chess is a kind of situational superko rule. :-) (Not that there are not other factors, such as allowing the players to consider variations.)

PurpleHaze: The "three-fold repetition" rule in chess requires that the positions be identical which includes having the same player to move as well as having all the special cases (en passant, castling, etc) in the same state.

It is optional, the draw only occurs if the player whose move it is announces that the position will occur for the third time and then plays the move to cause it.

The history is that occasionally the chess player with sente will discover that they have played an unpromising variation, find a forcing line to cause the position to recur, then choose another line.

All well and good until the invention of the chess clock, now a player with a difficult but advantageous position could force multiple repetitions to build up a bank of time in order to solve the position.

There are several games from the turn of the century with difficult endgames that have five plus fold repetitions or even multiple thereof.

The three-fold rule exists to counter this abuse. Even so there are often two-fold repetitions just before a time control is reached.


Tom: I have a question for the rules experts. I very much agree with Robert when he expresses a desire for simple unambiguous tournament rules. However, my personal preference is against a superko rule, or anything else which would complicate the description of the state of the game.

My question is: is it possible to devise a set of rules without superko, that are simple and unambiguous? These rules should leave the game recognisable as go, in the sense that in the vast majority of cases, good play under this rule set would be the same as good play under Japanese or Chinese rules (in the same way that the superko rule does).

As a price for this gain, I would accept rules for which good play leads to a never ending game occasionally, but I would like it to be rare that a player who is behind could use the threat of such an outcome to extract an advantage. This, I believe, is similar to the situation in Japanese rules, where a player who is behind can occasionally use the threat of a 'no result' through triple ko to gain an advantage.

Jozsef: Simplicity and unambiguity usually means avoiding territory scoring and it's problems with life and death. On the other hand, logical area scoring rules usually rely on superko. I know two attempts of area rules using basic ko: Erik van der Werf's Migos rules (used in his solver), and my attempt. If you aim for maximum simplicity, you could also try dropping my ko-pass rules.

Regarding the Implementation of Superko

isd: Recently there was an article in the BGJ which indicated problems with the implementation of superko in AGA rules. Specifically, it is possible to try to force a winning position in a superko cycle by refusing to acknowledge the point the game was at in this cycle. This was brought up by the application of a traditional triple ko = draw ruling from a BGA tournament. Whether or not this could be resolved by recording the game was not discussed.


Robert Jasiek: Practically speaking, all other ko rulesets are "related" to superko. Therefore it is misleading to speak of related rules in case of Long Cycle Rules or No Result. With the same right, one might list dozens of other ko rulesets as "related". Better move such to a different page like the general page about Ko Rules or a new page.


Superko/Discussion last edited by 50.23.115.116 on January 23, 2015 - 09:26
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
RecentChanges
StartingPoints
About
RandomPage
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Goproblems.com
Login / Prefs
Tools
Sensei's Library