beating a dead horse: my thoughts on suicide.

[nagano]

That's simplicity, not logic.

While we're at it, exactly what is the value of simplicity? It's a bit easier to teach beginners. That's good, worth trying for, but just one advantage among many. You seem to think it's worth more.

It is illogical to add rules that are not needed and may create complications. It is logical not to do this. You seem to be coming at this from the angle "How does this benefit the player?" But really this is more of a game design issue, and the effect to the player is not immediately obvious. The more rules you try to add to a game, the more undesirable side effects begin to arise. This is really a core principle in the design of perfect information abstract strategy games. I know it sounds a bit vague, really the only way to fully understand it is to try designing some games on your own, and to try to deal with problems that your rules create.

[hyperpape]

I'm coming at this from the angle that this has nothing to do with logic.

You say that adding rules can create complications. That is true. That is a cost to adding rules. When you say that logic dictates that we minimize the rules, you are either not using the word 'logic properly' or arguing in bad faith. I imagine it's the former.

I'm harping on this point because once you say "it's a matter of logic", it sounds like you've stated an absolutely compelling. After all, one must follow the rules of logic (sorry Mr. Fairbairn). But that's only because the laws of logic don't tell us about things like the rules of Go.

I think you should say something like one of the possibilities:

• I, Nagano, really like simplicity
• Experience has shown that the games with better play are simple (and you might add "given my evaluation of better play")
• If your game is not simple, you will incur various costs, which must be balanced against any other benefits of those rules (and perhaps you think there aren't any benefits...)

[nagano]

The word logic is in many cases meaningless apart from the goal it is applied to. Given a certain goal, there is a logically best way to proceed. In this case the goals are:

1. There must be no chance.
2. There must be no unnecessary rules.
3. There must be no unresolvable situations.

If you accept these three rules, then my reasoning is sound. If you do not, then nothing I can say will convince you otherwise. But seriously, try designing an abstract strategy game for yourself, and see if you do not at some point revert to these ideas.

"Everything should be made as simple as possible, but no simpler" - Albert Einstein

[Suji]

The word logic is in many cases meaningless apart from the goal it is applied to. Given a certain goal, there is a logically best way to proceed. In this case the goals are:

1. There must be no chance.
2. There must be no unnecessary rules.
3. There must be no unresolvable situations.

If you accept these three rules, then my reasoning is sound. If you do not, then nothing I can say will convince you otherwise. But seriously, try designing an abstract strategy game for yourself, and see if you do not at some point revert to these ideas.

"Everything should be made as simple as possible, but no simpler" - Albert Einstein

The options that you present point to checkers, chess, go, reversi, and any other number of strategy games all of which are perfectly decent games. Just because you came to the rules of go from designing an abstract strategy game doesn't mean everyone will. In fact, some would say that since go has multiple different rule sets and chess has only one chess would then be the "perfect" game.

P.S. I'm taking the devil's advocate point of view here.

[jts]

The word logic is in many cases meaningless apart from the goal it is applied to. Given a certain goal, there is a logically best way to proceed. In this case the goals are:

1. There must be no chance.
2. There must be no unnecessary rules.
3. There must be no unresolvable situations.

If you accept these three rules, then my reasoning is sound. If you do not, then nothing I can say will convince you otherwise. But seriously, try designing an abstract strategy game for yourself, and see if you do not at some point revert to these ideas.

"Everything should be made as simple as possible, but no simpler" - Albert Einstein

But - I think what only you and a small coterie of like-minded individuals accept is that "no suicide" is an unnecessary rule whereas "suicide is allowed" is the obvious but implicit conclusion of the basic capture rule.

[MarylandBill]

The word logic is in many cases meaningless apart from the goal it is applied to. Given a certain goal, there is a logically best way to proceed. In this case the goals are:

1. There must be no chance.

Well, assuming you mean no truly random element, then yes, this is true since this is one of the basic definitions of an abstract strategy game.

2. There must be no unnecessary rules.

The problem here is that people may not agree on the definition of unnecessary. The no suicide rule cuts both ways. At times it will benefit you and at times it will hurt you. The fact that most Go codes have the rule suggest that many believe it is a necessary rule.

More importantly, I would point out that "necessary" depends on the effect you are looking for in the game. Is the Queen's move necessary in chess? Not strictly since of course it is a relatively recent innovation to the game of chess. On the flip side, if you replace the Queen with a less powerful piece, the game changes considerably. Is either right or wrong?

If suicide is allowed, then one could use it to do all sorts of interesting things, like turning a group of dead stones into a seki position (At least I think this is possible, someone better versed in the game will have to address that). But perhaps that is not what the people who developed the current no suicide rule wanted to prevent.

3. There must be no unresolvable situations.

I am not sure I entirely agree. I do agree that such situations should be rare, but there is, I think, a certain value in having certain conditions that can allow a draw to occur. While normal draws in chess are kind of boring, stalemate (particularly if the player stalemated looked like they were in a lost position), can be quite exciting.

If you accept these three rules, then my reasoning is sound. If you do not, then nothing I can say will convince you otherwise. But seriously, try designing an abstract strategy game for yourself, and see if you do not at some point revert to these ideas.

"Everything should be made as simple as possible, but no simpler" - Albert Einstein

I think I would be careful of comparing game design with theoretical physics. Einstein was talking about modeling the Universe how it actually is.

--
Bill

[Laman]

If suicide is allowed, then one could use it to do all sorts of interesting things, like turning a group of dead stones into a seki position (At least I think this is possible, someone better versed in the game will have to address that). But perhaps that is not what the people who developed the current no suicide rule wanted to prevent.

i think it is the opposite, that the suicide is not very useful, only in rare cases creating more ko threats or changing the outcome of a semeai. if i imagine situations where it can be used, i think both allowing and disallowing suicide keeps the game equally fair, logical, consistent and strategic.

given the two premises above, i would say that from a view of an ordinary player, it doesn't really matter if we (dis)allow suicide. then applying the law of inertia - most players are satisfied with current rules and adding suicide doesn't really improve quality of their game experience (= enjoyment, in my opinion the highest goal of any game), so i vote for staying in the current state

if there were no go and nagano were designing a completely new game, there would be no problem with allowing suicide

[robinz]

Amazing that such a simple (and, from the point of view of the game of go in practice, not particularly important) issue seems to arouse such strong feelings

I must say that I'm pretty much neutral on suicide (and voted that way in the poll). Unlike many others on this thread, I know exactly where nagano is coming from, and agree that (purely as an aesthetic choice, if nothing else) one should not add extra rules to a game unless they are necessary (for balancing the game, or making the gameplay more interesting, and so on). However, I don't think that it's the case that forbidding suicide is an additional rule, and that if it wasn't there then the other rules automatically allow it - which seems to be nagano's assumption. In fact, it can be argued that rule-sets which allow suicide are more complicated - then each turn is typically divided into three stages, namely playing a stone, removing any opposing stones with no liberties remaining, and finally removing any of your own stones with no liberties remaining. With suicide banned, only the first two stages remain, along with a note that one is not allowed to make a play which would leave any of one's one stones without liberties after first removing any opposing stones without any. Which of the two rulesets is "simpler" strikes me as a pretty arbitrary (and subjective) choice.

The more important question is which leads to the better/most interesting game. I'm not a strong enough or experienced enough player to really know, but it seems that at the very least there is no concensus here, given the differences in the various rulesets in existence. I'm not even sure if it has ever really been discussed - but it is clear that, in most normal game situations, whether suicide is allowed or not makes no difference. (And those where it does will almost always be where suicide allows an extra ko threat.)

[hyperpape]

The word logic is in many cases meaningless apart from the goal it is applied to. Given a certain goal, there is a logically best way to proceed. In this case the goals are:

1. There must be no chance.
2. There must be no unnecessary rules.
3. There must be no unresolvable situations.

If 'logic' is meaningless outside of specifying a goal, then that tells against your usage. After all, John F. could walk into the thread and say the Japanese rules are the most logical because they best adhere to tradition, which is his goal (or one of his). And then the two of you could go back and forth, each saying his approach is logical, with no hope of rational discussion. Basically, if your claim is that from your goals, a certain approach logically follows, then that approach isn't so much supported by logic, as supported by your goals. And your goals are not logical (or illogical)--they're just your goals.

But also, you're wrong about the point of definition. The Merriam Webster definition 1a. is a pretty good stab at a definition of logic. It might not be crystal clear if you've never encountered the subject, but it's roughly on target. More or less, logic studies which sentences can be inferred from which other sentences.

It's no more a logical truth that some rules of Go are the best than it's a mathematical truth that this bottle of wine costs more than I should spend (though logic and math will be relevant to each bit of reasoning). If my brother buys the bottle of wine, I can't say "that's unmathematical!" though I can use math to explain to him that he will no longer be able to pay his rent.

[jts]

More or less, logic studies which sentences can be inferred from which other sentences.

Not that I disagree with you very strongly - I think your position makes more sense than Nagano's - but this is a very recent innovation in what "logic" means which was really only pioneered in the late 19th century. Before that, if someone wrote a book on "Logic", only a very small part of it would be about analysis, and most of it would be about the nature of mind, knowledge, rationality, intension, etc.

[nagano]

Boy, you guys give me a lot to respond to.

The options that you present point to checkers, chess, go, reversi, and any other number of strategy games all of which are perfectly decent games. Just because you came to the rules of go from designing an abstract strategy game doesn't mean everyone will. In fact, some would say that since go has multiple different rule sets and chess has only one chess would then be the "perfect" game.

P.S. I'm taking the devil's advocate point of view here.

Chess and checkers actually violate rule 2, because having different piece types (or promotion to different piece types) are not necessary. I also made the mistake of leaving out 4. The game must be as equal for both players as possible, which may at least create difficulties for many of these games. By the way, the assertion that Chess has only one ruleset is wrong. There are many variant rulesets of Chess, and its rules have been changed many times, even if you aren't counting the many games more properly described as variants.

But - I think what only you and a small coterie of like-minded individuals accept is that "no suicide" is an unnecessary rule whereas "suicide is allowed" is the obvious but implicit conclusion of the basic capture rule.

There is no other possible conclusion. Any move in go which is not expressly prohibited is allowed.

The word logic is in many cases meaningless apart from the goal it is applied to. Given a certain goal, there is a logically best way to proceed. In this case the goals are:

1. There must be no chance.

Well, assuming you mean no truly random element, then yes, this is true since this is one of the basic definitions of an abstract strategy game.

Correct.

2. There must be no unnecessary rules.

The problem here is that people may not agree on the definition of unnecessary. The no suicide rule cuts both ways. At times it will benefit you and at times it will hurt you. The fact that most Go codes have the rule suggest that many believe it is a necessary rule.

More importantly, I would point out that "necessary" depends on the effect you are looking for in the game. Is the Queen's move necessary in chess? Not strictly since of course it is a relatively recent innovation to the game of chess. On the flip side, if you replace the Queen with a less powerful piece, the game changes considerably. Is either right or wrong?

If suicide is allowed, then one could use it to do all sorts of interesting things, like turning a group of dead stones into a seki position (At least I think this is possible, someone better versed in the game will have to address that). But perhaps that is not what the people who developed the current no suicide rule wanted to prevent.

Here I define necessary as no more than is required for the game to function. In Chess, you do not need multiple piece types for the game to function, so it is unnecessary. Suicide rarely causes any change to life and death, aside from ko threats and some capturing races. Besides, this should not be a factor in deciding whether or not suicide should be allowed.

3. There must be no unresolvable situations.

I am not sure I entirely agree. I do agree that such situations should be rare, but there is, I think, a certain value in having certain conditions that can allow a draw to occur. While normal draws in chess are kind of boring, stalemate (particularly if the player stalemated looked like they were in a lost position), can be quite exciting.

A draw is not an unresolvable situation because the game has an outcome. Stalemate is an unnecessary rule (one that does not exist in Xiangqi, the Chinese Chess), but that is another issue.

I think I would be careful of comparing game design with theoretical physics. Einstein was talking about modeling the Universe how it actually is.

The only difference between physics and game design is that in physics, you are trying to determine the rules the universe follows. In game design, you create the rules of the "universe".

if i imagine situations where it can be used, i think both allowing and disallowing suicide keeps the game equally fair, logical, consistent and strategic.

It is not logical to add rules that are unneeded, and decrease the amount of options in the game.

given the two premises above, i would say that from a view of an ordinary player, it doesn't really matter if we (dis)allow suicide. then applying the law of inertia - most players are satisfied with current rules and adding suicide doesn't really improve quality of their game experience (= enjoyment, in my opinion the highest goal of any game), so i vote for staying in the current state

If your goal is what the players want, you can do anything you want. If your goal is to make a game with as few flaws as possible, then you should not create arbitrary rules.

if there were no go and nagano were designing a completely new game, there would be no problem with allowing suicide

This statement shows that you are giving preference to tradition over game design considerations.

However, I don't think that it's the case that forbidding suicide is an additional rule, and that if it wasn't there then the other rules automatically allow it - which seems to be nagano's assumption.

This is not an assumption. The fact that we are even arguing about whether suicide should be allowed or not proves that it is a distinct concept that has to be defined as a rule. It may be possible to write a ruleset in such a way that capture is not a distinct rule, but this would require modification of the other rules.

In fact, it can be argued that rule-sets which allow suicide are more complicated - then each turn is typically divided into three stages, namely playing a stone, removing any opposing stones with no liberties remaining, and finally removing any of your own stones with no liberties remaining. With suicide banned, only the first two stages remain, along with a note that one is not allowed to make a play which would leave any of one's one stones without liberties after first removing any opposing stones without any. Which of the two rulesets is "simpler" strikes me as a pretty arbitrary (and subjective) choice.

The process of play has no bearing on the complexity of the rules.

The more important question is which leads to the better/most interesting game. I'm not a strong enough or experienced enough player to really know, but it seems that at the very least there is no concensus here, given the differences in the various rulesets in existence. I'm not even sure if it has ever really been discussed - but it is clear that, in most normal game situations, whether suicide is allowed or not makes no difference. (And those where it does will almost always be where suicide allows an extra ko threat.)

The goal should be to remove all flaws.

If 'logic' is meaningless outside of specifying a goal, then that tells against your usage. After all, John F. could walk into the thread and say the Japanese rules are the most logical because they best adhere to tradition, which is his goal (or one of his). And then the two of you could go back and forth, each saying his approach is logical, with no hope of rational discussion. Basically, if your claim is that from your goals, a certain approach logically follows, then that approach isn't so much supported by logic, as supported by your goals. And your goals are not logical (or illogical)--they're just your goals.

It may not be obvious, but logic actually has to follow from goals. Is it logical to want to live? This is not a question that logic can answer. But you can make use of it in order to survive. Even in science, if your goal is not to understand the way the world works, then the methodology is totally irrelevant. But I by no means intended to imply that this was the only application of logic (note the words "in many cases"). Yes, logic can and should be applied to the choice of some goals. I think that we can easily say that tradition has no bearing on whether a game is good or not. That is based solely on the design of the game. But the primary goal has to be chosen. There is no way around this. But the goal must have some functional reason which is not wholly based on emotional preference. My primary goal is to create a game that is self-consistent, has no unnecessary rules, has a finite conclusion, and is as fair as possible to both players. The rules decisions that I make do follow logically from that premise.

But also, you're wrong about the point of definition. The Merriam Webster definition 1a. is a pretty good stab at a definition of logic. It might not be crystal clear if you've never encountered the subject, but it's roughly on target. More or less, logic studies which sentences can be inferred from which other sentences.

It's no more a logical truth that some rules of Go are the best than it's a mathematical truth that this bottle of wine costs more than I should spend (though logic and math will be relevant to each bit of reasoning). If my brother buys the bottle of wine, I can't say "that's unmathematical!" though I can use math to explain to him that he will no longer be able to pay his rent.

Something tells me you are referring to formal logic here, which I am not. Certainly one can make the argument that it is not logical, in an informal sense, to spend more money than you have.

I will end with another quote:

While the Baroque rules of Chess could only have been created by humans, the rules of Go are so elegant, organic, and rigorously logical that if intelligent life forms exist elsewhere in the universe they almost certainly play Go. - Edward Lasker

What do you think this means? Where did he get this idea? Do you agree with it?

[MarylandBill]

I am going to do a whole lot of snipping, to answer only a few of Nagano's points (I am sure others will more ably answer the other points).

Here I define necessary as no more than is required for the game to function. In Chess, you do not need multiple piece types for the game to function, so it is unnecessary. Suicide rarely causes any change to life and death, aside from ko threats and some capturing races. Besides, this should not be a factor in deciding whether or not suicide should be allowed.

You are living in a logical tautology here. You have set up "rules" (really your preferences), and then use them to prove your "rules".

My point was, and remains, that the functioning of the game is dependent on the rules. Even if the rules make only a minor change in the way the game functions that change might be necessary to produce the game that the designer (or in the case of old games like Go and chess that have evolved, its players) desired. The game game of Go plays slightly differently with and without the suicide rule, so whether it is necessary or not depends on what you are looking for.

Lets look at your chess example. You are absolutely wrong that chess does not need multiple piece types to function. The one thing that all the games that have descended (or are presumed to have descended) from Chaturanga have in common is that victory is achieved by check-mating one particular piece. If there is no piece to mate, you now might have a great board game, but it is not in any sense of the word, chess.

Thus, at the very least, that piece needs to be a different type than the rest of the pieces on the board. Further, while they all appear to be descended from a common ancestor and have a common object, chess, shogi, xianqi, and janggi are all very distinctive games. A big part of what makes them different games is the different piece types.

Indeed, in chess, and I suspect its cousin games (since I have never really played them), the differences in the types of the pieces forms an important part of the strategy in the game.

I appreciate your desire for a game where great strategic depth is developed from very simple rules. Its one of the things that I like about Go as well. But it would be wrong to say that Go is necessarily a better game than chess or xianqi because of those simple rules. In fact, since xianqi, shogi and chess are all played by more people in their respective home territories could be the most persuasive argument against Go. One would think that if Go was truly the better game, it should be most popular everywhere (Though I would give a pass in the West since its only been known here for the last century).

3. There must be no unresolvable situations.

I am not sure I entirely agree. I do agree that such situations should be rare, but there is, I think, a certain value in having certain conditions that can allow a draw to occur. While normal draws in chess are kind of boring, stalemate (particularly if the player stalemated looked like they were in a lost position), can be quite exciting.

A draw is not an unresolvable situation because the game has an outcome. Stalemate is an unnecessary rule (one that does not exist in Xiangqi, the Chinese Chess), but that is another issue.

In most cases, it is called a draw because of an unresolvable situation. Chess could do the same thing that Go does and order immediate replays of such unresolvable situations, but instead they opted to call them draws.

I think I would be careful of comparing game design with theoretical physics. Einstein was talking about modeling the Universe how it actually is.

The only difference between physics and game design is that in physics, you are trying to determine the rules the universe follows. In game design, you create the rules of the "universe".

Well, the important difference I was referring to is, you think that no game should have more than a few rules to dictate play. The Universe on the other hand probably can never be described in just a few paragraphs, or pages or even books. General Relativity and Quantum Mechanics are not nearly as simple as a board game and if they are ever unified, I expect the result will be more complicated still.

--
Bill

[EdLee]

A draw is not an unresolvable situation because the game has an outcome.
Stalemate is an unnecessary rule (one that does not exist in Xiangqi, the Chinese Chess), but that is another issue.

Pardon my ignorance, but what happens in a "stalemate" situation in Xiangqi? Does the side who cannot move lose?
Either way (draw/stalemate or lose), a rule was necessary here both in Western Chess and Xiangqi, otherwise the game would be unresolved.

[HermanHiddema]

A draw is not an unresolvable situation because the game has an outcome.
Stalemate is an unnecessary rule (one that does not exist in Xiangqi, the Chinese Chess), but that is another issue.

Pardon my ignorance, but what happens in a "stalemate" situation in Xiangqi? Does the side who cannot move lose?
Either way (draw/stalemate or lose), a rule was necessary here both in Western Chess and Xiangqi, otherwise the game would be unresolved.

Yes. If a player has no move in Xiangqi, he loses.

It is very possible to remove the need for stalemating rules by removing the concept of mate altogether, as well as removing all rules that make it illegal for the king to end up in check on the player's own move.

You can simplify the rules to: First player to capture the king wins (and you must make a move).

Then in stalemate, you cannot avoid putting yourself in check and your opponent will capture you. End of game.

Any current checkmate situation also involves being unable to avoid capture, so ends the game one move later.

[nagano]

Here I define necessary as no more than is required for the game to function. In Chess, you do not need multiple piece types for the game to function, so it is unnecessary. Suicide rarely causes any change to life and death, aside from ko threats and some capturing races. Besides, this should not be a factor in deciding whether or not suicide should be allowed.

You are living in a logical tautology here. You have set up "rules" (really your preferences), and then use them to prove your "rules".

I am beginning to think this whole dispute is a misunderstanding. I am not saying that all games should be this way. I like many games that do not fit many of the criteria I listed. But if the goal is to design a game with as few flaws as possible, my criteria may not be the only way, but it's insanely hard without them. Therefore I consider them to be the best and most elegant way.

My point was, and remains, that the functioning of the game is dependent on the rules. Even if the rules make only a minor change in the way the game functions that change might be necessary to produce the game that the designer (or in the case of old games like Go and chess that have evolved, its players) desired. The game game of Go plays slightly differently with and without the suicide rule, so whether it is necessary or not depends on what you are looking for.

Believe it or not, I never disagreed with this.

Lets look at your chess example. You are absolutely wrong that chess does not need multiple piece types to function. The one thing that all the games that have descended (or are presumed to have descended) from Chaturanga have in common is that victory is achieved by check-mating one particular piece. If there is no piece to mate, you now might have a great board game, but it is not in any sense of the word, chess.

Thus, at the very least, that piece needs to be a different type than the rest of the pieces on the board. Further, while they all appear to be descended from a common ancestor and have a common object, chess, shogi, xianqi, and janggi are all very distinctive games. A big part of what makes them different games is the different piece types.

Indeed, in chess, and I suspect its cousin games (since I have never really played them), the differences in the types of the pieces forms an important part of the strategy in the game.

First of all let me clarify: I meant one movement type. Of course the King is technically a different type of piece. I also agree with you here. I am very familiar with chess history and many variants. In fact I'm one of the few proud owners of the second edition of D. B. Pritchard's Encyclopedia of Chess Variants.

I appreciate your desire for a game where great strategic depth is developed from very simple rules. Its one of the things that I like about Go as well. But it would be wrong to say that Go is necessarily a better game than chess or xianqi because of those simple rules. In fact, since xianqi, shogi and chess are all played by more people in their respective home territories could be the most persuasive argument against Go. One would think that if Go was truly the better game, it should be most popular everywhere (Though I would give a pass in the West since its only been known here for the last century).

It's not so much that the rules are simpler, but that these rules do not lead to troublesome flaws.

In most cases, it is called a draw because of an unresolvable situation. Chess could do the same thing that Go does and order immediate replays of such unresolvable situations, but instead they opted to call them draws.

There are other ways to deal with blocked positions, but a draw that results from both players playing equally well that still allows the game to finish is fine.

I think I would be careful of comparing game design with theoretical physics. Einstein was talking about modeling the Universe how it actually is.

The only difference between physics and game design is that in physics, you are trying to determine the rules the universe follows. In game design, you create the rules of the "universe".

Well, the important difference I was referring to is, you think that no game should have more than a few rules to dictate play. The Universe on the other hand probably can never be described in just a few paragraphs, or pages or even books. General Relativity and Quantum Mechanics are not nearly as simple as a board game and if they are ever unified, I expect the result will be more complicated still.

The universe actually appears to get simpler the smaller the scale; but many simple systems interacting can lead to an appearance of great complexity. Really the only reason anything seems complex is because our brains cannot hold it all.