Superko rules and ko-cycles rules are BAD board game design

[CDavis7M]

UPDATE: just adding a bit here to hopefully clear up the discussion. I will present some fundamental principles of board game design. So far it seems that no one has tried to dispute these principles. If you think about it you will realize that all games follow these principles. But if you would like to dispute the principles then please at least try to provide examples of good board games that violate these principles. I have tried to do this myself, but I cannot even find bad board games that violate these principles.

Principle 1: A board game should not require the players to perform mental bookkeeping of anything in the game besides the last play (e.g., placing a stone to capture a stone in Go, a player playing a card and resolving its effects in Uno, spinning and taking cherries in Hi-Ho-Cherry-O).

Principle 2: The score or victory-conditions should be determinable by the game-state. Even better if the score/victory directly corresponds to the plays game in the game.

Principle 3: The rules and game-pieces of a game should be simple while the strategy and tactics provided by the game should be complex.

If a game would violate these principles, the Designer would provide game-pieces/tokens or a note/score-pad for tracking what is happening in the game.

Go does not include pieces/tokens for tracking cycles. The superko rule covers situations that can not be determined by the existing board state and the last play. The superko rules provide a victory condition that is not based on the game-state.

Arguing that players can already perform mental calculations for life & death and tactical reading so then players should also be able to memorize cycles for the victory condition are missing the point of Principle 3.

----------------------------------

Why are there so many posts trying to "solve" the supposed "problem" of Kos in the board game Go? These rules are bad board game design. And many of the "solutions" are even worse board game design. I do not understand the preoccupation with trying to make this game worse.

It is a fundamental principle of good board game design that any decision making by the player (where/how to play) need only rely on the existing state of the game (the stones positions in Go) and the last change to that game state (the last stone placed in Go, or the last turn/phase in some games). While mental bookkeeping may be a good strategy in some games (e.g., card counting), it should be minimized when determining the viability of a play/move in a game. Any rule requiring the players to perform mental bookkeeping for what happened in past turns is cumbersome. The fact that Go does not have tokens/markers to track when stones were captured shows how incongruent super ko rules are. I'm not suggesting that Go implement tokens/markers to track every capture and which turn. That would also be cumbersome. I'm just making the position that Go with the possibility of a perpetual ko-cycle is the best version of the game.

Just because tournaments require professional Go players to play with poorly designed rules does not make that version of Go better than the version of Go having the small possibility of a perpetually reoccurring game state. Admittedly, a perpetual cycle is poor game design, but in Go it is unlikely and it is not cumbersome to game-play. Even in a tournament setting, it would be better game design to have an alternative victory condition for games that are so close that the players would decide to perpetually continue a ko-cycle. Rather than require mental bookkeeping of super kos, the tournament could simply allow the players to decide the game by nigiri or by a shorter 13x13 or 9x9 game.

And by definition, doesn't 劫 presuppose the game lasting for aeons without end. Adjourning a well designed game that unfortunately entered an unlikely perpetual state is better than playing a game with poorly designed rules or mechanics.

[kvasir]

I can agree that superko is lame. We are also not required to write down our moves during a game, so how do we resolve superko disputes?

I think what all the threads are about is not restricting normal play but to address how to handle such cycles when confirming life and death. It is superfluous in that we never really wrangle over the end of the game in that way, but maybe nice to have a less ambiguous way to decide the result when we think we are done playing. I'd not take it too seriously.

I thought 劫 suggested coercion or robbery, that is usually what I am aiming for

[RobertJasiek]

"the possibility of a perpetual ko-cycle is the best version of the game", where you mean "the possibility of a perpetual ko-cycle is the best version of the game, except a perpetual ko-cycle of two successive moves that are plays", suggest "the tournament could simply allow the players to decide the game by nigiri or by a shorter 13x13 or 9x9 game":

For your opinion to possibly start making some sense, you first need to clarify the following:
a) What about a perpetual ko-cycle of two plays with intervening pass(es), such as the last basic endgame ko when the opponent's only legal ko threat is a pass or filling an intersection of his own two-eye-formation? (I cannot guess what you consider good rules design here.)
b) What about a cycle of exactly 1 play being single-stone suicide (let me guess: you want to prohibit suicide for some reason)?
c) What about the most frequent cycle (longer than 1 play) comprising 3 plays (sending-2-returning-1)? (Let me guess: for your opinion of superko and long cycle ko rules to be bad game design, it is necessary to call traditional territory scoring "good" game design.)

After your clarifications, we can start discussing your contents.

Why do you consider complicated rules design (to decide the game by nigiri or by a shorter 13x13 or 9x9 game) to be good rules design? Simple rules design other than to simply apply the same (superko-like) rules (which you dislike for other reasons than the rules being simple as rules) is to declare a game with special behaviour a tie (equivalent to jigo).

"Board game design" is a strong phrase. Discussion is already complicated enough if we restrict it to "go game design".

You say that the possibility of a perpetual ko-cycle was the best version of the game but what it actually means is infinitely long, endlessly repeated play. Good rules design for human players prevents the possibility of a perpetual ko-cycle! Practically speaking, it also handles play longer than executable within reasonable time, such as a cycle with more than 19.500.000 moves. So presumably you mean the contrary: the PREVENTION of a perpetual ko-cycle. Since you call superko-like rules bad, more specifically you want ko restriction rules somehow restricting long cycles. This brings us to another clarification you need to make:

d) Do you want to allow or prohibit occurrence of long cycles with 4+ plays, such as in eternal life or triple ko? Do you call Japanese style no-result ko rules good or bad? (Let me guess: since such are (long) ko-cycle rules, you call them bad. This raises the next clarification.)
e) How to let ko rules aka go game design detect any long cycle at all if such rules are called bad and therefore may not be part of go game design? (Let me guess: you have overlooked this contradiction in your opinion.)

After all your clarifications, let us discuss the central topic of your opinion:

Why do you consider superko rules and ko-cycle rules bad?

I consider the simplest such rules good.

The simplest is the positional superko rule. It is good because
- just one rule handles all,
- almost all practically occurring cycles are at most 6 plays long and can be handled well,
- the only significant, practically occurring cycles with more than 6 plays occur in quadruple or quintuple kos, in which the skill to avoid a rule violation by accidentally completing a long cycle is still much easier than the skill required for tactical reading in difficult life and death problems without cycles.

I also consider the Basic-Fixed Ko Rules (basic ko rule and no continuation of recycling on completing a long cycle) good. They are good because
- just two rules handle all,
- not playing a long cycle is strategically equivalent to playing it once.

You suggest that long cycle kos were not a problem at all. Uh, but they are - otherwise you would not even have started this thread:)

You suggest that people were trying to solve long cycle kos rules. Uh, no, not exactly. They have been solved long ago! Positional superko is a possible solution. The Basic-Fixed Ko Rules are a possible solution.

The problem is rather that some people including you a) consider the superko solution bad and b) do not explain why the Basic-Fixed Ko Rules solution was not good and why not better than Japanese-style no-result rules.

"It is a fundamental principle of good board game design that any decision making by the player (where/how to play) need only rely on the existing state of the game (the stones positions in Go) and the last change to that game state (the last stone placed in Go, or the last turn/phase in some games)":

Sorry, but this cannot be so for go because cycles longer than 2 moves do exist, do occur and do need ruling. You can only achieve this on the strategic level.
- Superko does not achieve this on the strategic level.
- Japanese-style no-result rules do not achieve this on the strategic level because it is necessary to make strategic decisions whether to choose or avoid variations with the result no-result.
- The Basic-Fixed Ko Rules achieve this on the strategic level because not playing a long cycle is strategically equivalent to playing it once.

So you might like my Basic-Fixed Ko Rules, although they do not achieve it on the rules level. The latter must be accepted because no go rules ever could achieve it on the rules level if strategic ko play shall be allowed in the game of go (hint: go could be designed with the prohibition to play twice on any intersection but the game would be fundamentally different and hopelessly boring; go as we love it is a game with cycles).

What is 劫?

[kvasir]

What is 劫?

The Chinese character for ko

[Cassandra]

What is 劫?

The Chinese character for ko

劫 = 刧

= KYÔ, GÔ = threat, manace, harassment; extremely long time (Buddhist)
= KÔ = extremely long time (Buddhist); experience gained over many years; certain constellation (Go game)

[John Fairbairn]

劫 = 刧

= KYÔ, GÔ = threat, manace, harassment; extremely long time (Buddhist)
= KÔ = extremely long time (Buddhist); experience gained over many years; certain constellation (Go game)

Go had kos long before Buddhism reached China. The application of the meaning of aeon to go is Japanese folk etymology (or sacerdotal etymology if you prefer). In origin, the term is Chinese (jie), not Japanese. The Chinese for to start a ko is dajie, which happens to be Chinese also for mugging.

[kvasir]

The best ko rule is 趁火打劫 (chen huo da jie) - start fires to fight ko.

[John Fairbairn]

"Design" is a red herring and so a bad choice. It's better to talk of go having evolved.

There is a simple thought experiment. It is based on the simple premise that go is a game played for fun.

Awkward positions are rare in go. Those that might occur with any sort of noticeable frequency, such as bent four have special rules. These special rulings are very few in number and are easy to understand. "Bent for is dead." There is no need to explain it, demonstrate it or justify it. We just look and say "dead." Likewise "triple (or higher) ko is a draw or replay." Most other freak positions are so rare they will only be seen in the lab. Because it is so easy to follow these very few rules, go has become massively popular. Furthermore, rather than being a blemish, the special rules (amongst which we can include ko itself) are very few in number and so easy to handle that they are seen by most people as a feature, not a bug. They add a little salt and pepper to the meal. The result has been that we now have a game which is not utterly trivial yet which is not over-complicated, and which as a result has become massively popular, not only with players but artists and poets and politicians. It has become an integral part of various culture, and has sustained that status for 2,000 years.

Imagine now, instead, a similar game where freak lab-type positions occur in almost every game, and require pages of written rules and a special sort of mentality to apply them. If we can imagine that, I think we can easily imagine also that such a game might appeal to certain kinds of people - say, those who avoid walking on the cracks in pavements, those who have OCD, those who have dictatorial tendencies, or just those that have nothing better to do with their lives. But do even they really, really, really believe such as game would apply to the the majority of people and could become part of culture.

As I often like to say, the problem with logic is that you get different answers depending where you start from. If you start from the simple premise that you want a game that is suitable for most people, you end up with a different game from the one where you start with the premise that every possible game played must have rules with predictable result. Putting the rules ahead of people does bring carts and horses to mind.

But that's just the overall dichotomy. Even within the minor details of the debate there are startling errors of logic by people who think they are being rational. Just to mention a couple, there is the obsession with getting a result in terms of winner. Yet go for most of its existence go has happily gone on with a mode of play in which jigo can occur. If a draw can occur because of equal points, why not because of triple ko? Tournament rules (including komi) are not actually game rules, remember. You are then starting from a different point. So the logic changes.

Another irritating argument is that working out cycles is easier than doing life and death calculations. But the important difference is that in doing tactical calculations you are allowed to play where you like - to make mistakes even, but making a mistake does not mean you automatically lose. When doing a cycle, however, you have to learn and follow a quite separate algorithm rigidly and if you make a misstep you automatically lose. In other words, you are not supposed to park your car too near a corner. But you misjudge what is too near, and from the ordinary people in charge you get a fine or other slap on the wrist. With the rules mavens in charge, however, you would be sent before a firing squad.

Since I know from long experience how this thread will continue (logic choppers are programmed not to listen), I'll say at this stage I will no longer contribute to it. I just wanted, by pointing up some absurdities, to make this present post as an excuse to make yet another plea for this forum to move away from rules (and AI). Take a step back - go has thrived for 2,000 years as a game with a culture, as a game played by real people with interesting lives. These things still exist. Let's concentrate on them instead.

[RobertJasiek]

While there is the historical explanations of precendental ko rules, there is no need to go back to them.

Awkward ko positions do not only occur in the rarities but occur also in the frequent shapes, starting with the dead ko, which occurs in about every second game.

People do not per se have difficulty to understand set rules. People new to go can easily grasp the concept "repetition is prohibited" and its necessity that "the game proceeds". It is rather so that some people have unwillingness to accept change to other rules they learnt earlier, even if they must newly learn more exceptional rules to maintain those rules they already know.

If you allow taktical mistakes in life and death problems for the sake of the argument, then life and death is the easiest thing because you may play move 1 wrongly. However, a purpose of go is to play well. This includes appreciating the skill of correctly playing in life and death problems at all, and increasingly complicated LD problems with increasing playing strength. Such problems quickly become more demanding than playing quintuple ko correctly. Your objection is one of unwillingness of doing so and your excuse of appreciating weak play in life and death problems is weak.

[kvasir]

[...]to make yet another plea for this forum to move away from rules (and AI).

In the end people simply post in the sub-forums that they see fit and on the topics that they do like. I certainly miss more activity in Improve Your Game but if someone likes to post about square stones - whatever! Just as go is a game that doesn't have much purpose if you don't enjoy it, it is the same with writing and reading forum posts. If you don't enjoy it - it is of no use! Why does everyone else need to conform?

Here we are talking about posting about go rules in the go rules sub-forum. There are already a few ongoing threads with titles like "subject blah blah - no mention of yata yata", do we really need to control and try to make people conform to our perspective?

[Cassandra]

... do we really need to control and try to make people conform to our perspective?

No.

However, it should not be wrong to discuss the (very few and rarely occuring) differences between the various rule sets with an open mind.
Just to help people from different worlds what to look out for when they enter another.

I think the different rule sets have a lot to do with location, history and culture.

Probably the difference between Chinese rules and Japanese rules was due to the very different landscape of the large mainland of China and the small island(s) of Japan. And the consequences that have resulted from this in the people's life.

Same with the differences to Western-style rule sets, as Western history and culture is very different from East-Asian history and culture.

--------------------------------

Probably the J89's authors intended a backlash to the Western world by trying to force their understanding of the game of Go into a logical corset.
However, it seems to me that they were not fully aware of the difficulties involved in transposing a matter from one culture area to another. Or to do justice to TWO worlds in ONE treatise.

[RobertJasiek]

Probably the J89's authors intended a backlash to the Western world by trying to force their understanding of the game of Go into a logical corset.

J1989 was motivated by Japanese criticism on J1949 being illogical. J1989 moved a bit towards logic within the limited logic skills and limited knowledge of go rules theory of its authors.

Conceptually, J1949 / WAGC were a failure to describe life by eyes logically so J1989 made the capturability attempt and failed again.

Independent of that, Japanese spread go in Western countries and with that effort came their rules. Partly, early European go players got information about go mostly from Japan; also that way Japanese rules fragments spread.

[CDavis7M]

For your opinion to possibly start making some sense, you first need to clarify the following:

I feel like your entire post is just trying to troll me. Maybe you felt that I was specifically trolling you? I wish that were the case now because I might have tried harder.

a) What about a perpetual ko-cycle of two plays with intervening pass(es), such as the last basic endgame ko when the opponent's only legal ko threat is a pass or filling an intersection of his own two-eye-formation? (I cannot guess what you consider good rules design here.)

Is this a question? Indeed, what about it such a ko-cycle? I see no issue with it beyond what has been stated. Is this a situation where the ko cannot be filled? In that case, it does sound like it might be a perpetual ko if the players choose. But wasn't that an assumption of the question?

b) What about a cycle of exactly 1 play being single-stone suicide (let me guess: you want to prohibit suicide for some reason)?

Again -- what about it? What is the question? I know that some rules prohibit playing a move that would be suicide. But it would be better game design to just not have a rule prohibiting suicide. Such a rule is unnecessary. So I don't know why you think I would want to prohibit it.

c) What about the most frequent cycle (longer than 1 play) comprising 3 plays (sending-2-returning-1)? (Let me guess: for your opinion of superko and long cycle ko rules to be bad game design, it is necessary to call traditional territory scoring "good" game design.)

It's these questions that don't make sense. What is the question? And no, traditional territory scoring is not good game design, but it is not necessarily bad game design. Think for a second. It is Stone Scoring that is good game design because the scoring is simple (count pieces on a board) and it directly corresponds to play-mechanic which is used to build the game state. Territory scoring requires the additional game-mechanic of prisoners. Though it is not usually cumbersome and in fact it is a bit satisfying for the player. Area scoring requires scouring both stones, places where stones could be, and eyes. The basic game mechanic is placing a stone and Stone Scoring counts placed stones. It is the best game design for scoring because it requires the fewest mechanics and it directly relates to the fundamental play-mechanic.

Why do you consider complicated rules design (to decide the game by nigiri or by a shorter 13x13 or 9x9 game) to be good rules design?

I never said that I consider this to be a good rule design. Also, I don't see what is complicated about allowing the players to agree to decide the game by nigiri or a smaller sized board? The mechanic itself is simple.

Simple rules design other than to simply apply the same (superko-like) rules (which you dislike for other reasons than the rules being simple as rules) is to declare a game with special behaviour a tie (equivalent to jigo).

Sure, that is a simple mechanic but it is still an additional game-mechanic compared to simply adjourning the game, which is not a game mechanic. If a game cannot be adjourned because of tournament requirements, then sure, a jigo is a simple alternative just like nigiri.

"Board game design" is a strong phrase. Discussion is already complicated enough if we restrict it to "go game design".

There are thousands of examples of board game and game mechanics from which to draw comparison. There are many game mechanics that have been improved on over time and many game mechanics that have been created to solve a problem. Discussing board game design simplifies the discussion about Go because it clearly shows that certain propositions are just bad game design. The fact that there are complicated discussions about ko-cycles in the first place shows that any rules would be overly complex and thus bad board game design.

You say that the possibility of a perpetual ko-cycle was the best version of the game but what it actually means is infinitely long, endlessly repeated play. Good rules design for human players prevents the possibility of a perpetual ko-cycle!

I already admitted that allowing for the possibility of a ko-cycle was bad game design. It's just not as bad as implementing ko-cycle rules.

So presumably you mean the contrary: the PREVENTION of a perpetual ko-cycle. Since you call superko-like rules bad, more specifically you want ko restriction rules somehow restricting long cycles.

I never said that and nothing that I did say would suggest that. ...I feel like you're just trolling me.

d) Do you want to allow or prohibit occurrence of long cycles with 4+ plays, such as in eternal life or triple ko? Do you call Japanese style no-result ko rules good or bad? (Let me guess: since such are (long) ko-cycle rules, you call them bad. This raises the next clarification.)
e) How to let ko rules aka go game design detect any long cycle at all if such rules are called bad and therefore may not be part of go game design? (Let me guess: you have overlooked this contradiction in your opinion.)

Haha! Come on. Anyone else (but you) reading my post will understand my thoughts on these points. A long cycle is not ideal but it is not bad, and certainly not as bad as requiring players to perform mental bookkeeping. Either player can choose to end a cycle if they want. If the game is too close for that, then alternative victory conditions (or a jigo) would be better than requiring mental bookkeeping.

Why do you consider superko rules and ko-cycle rules bad?

I've already said why: they require the player to perform mental bookkeeping of what happened beyond the last turn. And there is no game-mechanic for tracking the cycle.

"Cycles" are not even a concept of Go. There is literally no game mechanic to track cycles. Any rule about cycles goes beyond the game of go. It's like having a rule about what the players can have for lunch while playing the game. It's irrelevant to the actual game mechanics. If cycles were part of the game of Go, there would already be a mechanic for tracking cycle.

The simplest is the positional superko rule. It is good because
- just one rule handles all,
- almost all practically occurring cycles are at most 6 plays long and can be handled well,
- the only significant, practically occurring cycles with more than 6 plays occur in quadruple or quintuple kos, in which the skill to avoid a rule violation by accidentally completing a long cycle is still much easier than the skill required for tactical reading in difficult life and death problems without cycles.

The superko rules is NOT "simple." Just because there is one rule that can cover multiple situations does not make it simple. Superko is complex because the state of the board game does not present the viability of the next move that can be played.

You suggest that long cycle kos were not a problem at all. Uh, but they are - otherwise you would not even have started this thread:)

It's not long ko cycles that are the problem. I have no problem with that. It's long posts that yours that are my problem.

"It is a fundamental principle of good board game design that any decision making by the player (where/how to play) need only rely on the existing state of the game (the stones positions in Go) and the last change to that game state (the last stone placed in Go, or the last turn/phase in some games)":

Sorry, but this cannot be so for go because cycles longer than 2 moves do exist, do occur and do need ruling. You can only achieve this on the strategic level.

No, I'm sorry. But this is true for EVERY board game, including go. Your failure is that you have yet to explain why cycles longer than two moves would need a ruling in the first place. And if longer cycles were to be ruled on, why would there not be some game mechanic to track them?

[CDavis7M]

劫 = 刧

= KYÔ, GÔ = threat, manace, harassment; extremely long time (Buddhist)
= KÔ = extremely long time (Buddhist); experience gained over many years; certain constellation (Go game)

Go had kos long before Buddhism reached China. The application of the meaning of aeon to go is Japanese folk etymology (or sacerdotal etymology if you prefer). In origin, the term is Chinese (jie), not Japanese. The Chinese for to start a ko is dajie, which happens to be Chinese also for mugging.

Fair point. Even if the acceptance of the perpetuality of Kos came later, that acceptance has been around far longer than the attempt to contrain kos. Even if the game mechanic were named "jie" at inception, at least jie doesn't suggest anything about cycles or prevention of cycles.

This fact highlights that tracking "cycles" is and was never a mechanic in the game of Go.

[CDavis7M]

However, a purpose of go is to play well. This includes appreciating the skill of correctly playing in life and death problems at all, and increasingly complicated LD problems with increasing playing strength. Such problems quickly become more demanding than playing quintuple ko correctly. Your objection is one of unwillingness of doing so and your excuse of appreciating weak play in life and death problems is weak.

Ah! No wonder you don't know about game design. You don't even know what a "game" is.