Life In 19x19

It can be found stated many times around the web that a board which would allow two players to play according to the Ing rules cannot be programmed because of the ko rules. My question is, what is the more precise limitation, whether:
a)It is totally impossible
b)It is possible only if the given boardsize is solved
c)It is theoretically possible with enough computing power (brute force check)
d)It is possible but nobody has found a way to do it yet
e)It has been done, but the web is full of out of date info
Just out of curiosity

The answer is: it is still impossible because the Ing ko rules are still not fully understood. You can start here:
http://home.snafu.de/jasiek/ko.pdf
http://home.snafu.de/jasiek/ko_types.pdf
http://home.snafu.de/jasiek/external.pdf

This allows to define ko so that the ko concept can be understood for Ing ko rules. However, they also have lots of superfluous terms, such as "ko position" and "invariation", whose ambiguity even I have not resolved on the definition level yet. I could if I invested another few thousand hours but there is too little benefit also on the intellectual level. I have solved the core of the Ing ko rules here:
http://home.snafu.de/jasiek/newko.html
http://home.snafu.de/jasiek/ingkolec.txt
http://home.snafu.de/jasiek/ingko_epilogue.pdf
This together with the ko definition tells us how Ing ko rules should have been designed. It is not so interesting to find exact definitions for how they ARE designed.

Computational complexity is another hurdle. Not to mention thousands of hours of implementation.

I have searched the internet and these files, but I haven't yet actually seen a definition of "hot stone", even through Ing rules seem to stand on this concept. Have I overlooked it somewhere or is this term still undefined like the other terms?

Ing rules have been codified on IGS, but I don't think they're accessible any more via its new Pandanet interface (and I can't say if they've been implemented to the extent you guys are discussing)

xed_over wrote:Ing rules have been codified on IGS, but I don't think they're accessible any more via its new Pandanet interface (and I can't say if they've been implemented to the extent you guys are discussing)

I have also seen OGS claiming to have it implemented, but experimenting with the demo/review feature reveals it applies superko.

If you try to implement Ing's text without the examples, programming the Ing rules is nigh impossible, because of the ambiguity and inconsistency of the text. However, I think that the Ing '96 rules ( http://www.usgo.org/files/pdf/IngRules2006.pdf ), along with the KSS paper ( http://www.cs.cmu.edu/~wjh/go/rules/KSS.html ) are enough to produce a program that would satisfy the Ing Foundation. (One thing. Having seen the Chinese version of the '96 rules, I am reasonably certain, despite not reading Chinese, that it uses proximity scoring.)

About "invariation": Invariation is a made up word. AFAICT, invariance works just fine. E. g., replacing invariation with invariance in one paragraph of the '96 rules yields this:

Invariance:
Invariance is the cyclic repetition of board positions, when both sides refuse to yield. Invariance results in games with no outcome. How to prevent all invariance was the enigma which puzzled rule makers for almost five thousand years.

That reads a lot better, IMO. To continue with the ko section of the '96 rules:

Ko:
Ko prevents invariance. These rules classify ko as fighting or disturbing. Repeated removal of ko stones after intervening board or pass plays is a ko fight, cyclic removal of ko stones is disturbing. The hot stone rule prevents invariance by disallowing the removal of hot stones (i.e., immediate recapturing of ko stones without an intervening board or pass play) in a ko fight. The disturbing ko rule prevents invariance by forbidding the disturber from disturbing after one complete cycle (of repeated board positions). Since all types of invariance are accounted for, every life and death shape can be resolved.

Ing distinguishes "repeated removal" from "cyclical removal", but which term to use appears to depend upon whether the ko is fighting or disturbing. Those terms do not help up to make that distinction. Again, trying to decide whether stones are hot or not does not help us to tell the difference between a fighting ko or disturbing ko. But in a fighting ko, plays that do not capture stones are permitted. The hot stone rule only disallows captures. The disturbing ko rule allows a disturbing ko cycle to be played only once.

Fighting ko:
A fighting ko determines the life and death of the opposing groups involved. The ko stones in the repeated fight are called hot stones. Hot stones cannot be removed until after an intervening board or pass play.

Clear enough, except for how to tell whether the ko determines the life and death of the groups.

Disturbing ko:
A disturbing ko is the cyclic removal of ko stones initiated by the disturber by either fighting an unnecessary hot stone fight or recycling ko threats. If either side refuses to give in, cyclic removal of ko stones is resulted. Not only does this prevent the game from ending, it serves no purpose. After one complete cycle, the disturber is never allowed to continue disturbing. For any disturbing ko, every move in the second cycle or subsequent cycles causes invariation, and the disturber can be forbidden from disturbing at any time, enabling the game to end.

By inference, a disturbing ko is one where the life and death of opposing groups is not at issue. There seem to be hot stones in some disturbing kos, but they can be captured.

So life and death is key to distinguishing between fighting and disturbing kos. More later, but I am going to take a break now.

The reader may be interested in what I wrote some time back, based upon the previous version of the Ing rules before the '96 version. See http://senseis.xmp.net/?FightingKoAndDi ... Definition and http://senseis.xmp.net/?IngSpightKoRule . Note that the Ing-Spight rule would not be acceptable to the Ing Foundation.

In Ing rules, life and death are the key to whether a ko is fighting or disturbing. The Ing '96 rules make an important change (in the text) about life and death.

Life and Death
Life and death are determined by breath type; verify by removal.

Breath is used for liberty in the sense of adjacent empty point. In previous versions, the Ing rules stated that life and death were determined by removal, even though that did not work. Finally Ing has admitted that life and death are not determined by removal. The rules do not say so, but verification by removal assumes the correct determination of fighting and disturbing kos. That means that fighting and disturbing kos may be distinguished by inspection.

Breath Type:
All external breaths are false breaths unless there is a pairing up of the internal breaths and a pairing up of the external breaths on both sides. All internal breaths are real breaths with a minimum of two real breaths for a live group: permanent breaths for independent life, balancing breaths for coexistence [seki]. Breaths (real or false breaths) involving groups where life and death have not been settled are fighting breaths. When one side recycles ko threats, a ko fight becomes a disturbing ko. In a disturbing ko, the breath types have been decided (real or false), but one side insists on disturbing the game.

A group needs real breaths for life, except in the case of some rare seki positions where the player who can capture a group in atari cannot afford to do so. For pairing of breaths see the examples or the KSS paper. If there is, say, a quadruple ko in which each side has two breaths, it is a disturbing ko, because the breaths are classified as real and each side, with two real breaths, is alive. That determination cannot be verified by removal.

An algorithm to recognize disturbing kos only has to distinguish internal from external breaths and know how to pair up breaths.

It is easy to get lost in the Ing rules, because Ing claims that they do what they do not. If you try to implement his claims you will go astray. In the '96 rules Ing finally admits that fighting and disturbing kos are not distinguished by play. So you do not have to start off by asking whether stones are hot or whether removal is repeated or cyclical.

If you expect the Ing rules to be logical, in the sense that they are derived from first principles, they are not. Nor are they elegant, IMO. But Ing did succeed in liberalizing the rules of go without making the game too complex for human players.

Bill Spight wrote:Ing did succeed in liberalizing the rules of go without making the game too complex for human players.

This can be achieved with much simpler ko rules.

Bill Spight wrote:Ing did succeed in liberalizing the rules of go without making the game too complex for human players.

What? How much more complex compared to Ing ko can a ko rule get? Ing criticises the Japanese rules throughout his ruleset and then decides to base the ko rule on Japanese style special status assessment principles? Simple! I am sure strong dan players have a fairly good understanding of the way it works, but at our longitude a weak SDK can be a tournament referee even at a fairly significant tournament. How are Ing rules supposed to be applied here?

So life and death is key to distinguishing between fighting and disturbing kos.

One would then have to call the last endgame ko a disturbing ko. That doesn't seem to be right.

tiger314 wrote:

Bill Spight wrote:Ing did succeed in liberalizing the rules of go without making the game too complex for human players.

What? How much more complex compared to Ing ko can a ko rule get?

Don't ask. But the question is the complexity of the game.

For instance, if you have two double ko deaths on the board, under Japanese and Korean rules (IIUC) and the player with the "dead" stones has the move, she can hang the game by going back and forth between them. Under AGA and New Zealand rules, and the first Ing rules, they form a superko. Under the last Ing rules they are simply dead. The Ing rules simplify the game.

Ing criticises the Japanese rules throughout his ruleset and then decides to base the ko rule on Japanese style special status assessment principles?

No, the Japanese '49 rules are deader than a doornail. Even the Japanese '89 rules are based upon general principles. The Ing '96 rules determine life and death in a new way for complex kos, which apply throughout the game. Ing wanted to determine life and death by play, and claimed that he had done so in previous versions of the Ing rules. Finally he admitted that he does not.

Simple! I am sure strong dan players have a fairly good understanding of the way it works, but at our longitude a weak SDK can be a tournament referee even at a fairly significant tournament. How are Ing rules supposed to be applied here?

Being an Ing referee does not require much go expertise. If you really want to host an Ing tournament, contact the Ing Goe Foundation, 887 Oak Grove Road, Suite 203, Menlo Park, CA 94205. I am sure that they will be very helpful.

tiger314 wrote:

Bill Spight wrote:So life and death is key to distinguishing between fighting and disturbing kos.

One would then have to call the last endgame ko a disturbing ko. That doesn't seem to be right.

No, that remains a regular ko about the life or death of a single stone. The Ing rules do not alter the classification of regular kos.

Here is a game that might at first glance challenge a referee who is not a very skilled go player. (See https://gogameguru.com/quadruple-ko-chi ... qi-league/ )



The quadruple ko starts at move 266.

First, a weak referee should realize that it is not a superko, because there are no superkos per se in Ing rules and there are none that combine separated kos. So she can rule that there is a disturbing ko on the board. The obvious candidate is the double ko on the right. It takes a certain degree of go expertise to see that it is a double ko death, and that White is dead. However, the players know that it is a double ko death. Otherwise Black would simply have filled a ko at move 267 or later. So if the referee does not see that, she can and should ask the players what is going on. (She should do that anyway, as soon as she arrives, but referees are not always taught that, because of the assumption that go rules are simple to apply.)

Edit: Note that this game was played by Chinese rules, which ostensibly have a superko rule, but the referee ruled no contest and had the players play another game.

Edit2: Note also that the Ing-Spight rule prohibits White from taking a ko at move 272. It does not always produce the same results as the Ing rules, but there are only two known exceptions.

Bill Spight wrote:The Ing rules do not alter the classification of regular kos.

What do you mean by this with respect to what I call "dead ko".

tiger314 wrote:How much more complex compared to Ing ko can a ko rule get?

You are asking for a ko ruleset. The upper bound for its textual complexity is infinity because, in principle, one can enumerate all finite and infinite game sequences and assign a different rule to each;)

RobertJasiek wrote:

tiger314 wrote:How much more complex compared to Ing ko can a ko rule get?

You are asking for a ko ruleset. The upper bound for its textual complexity is infinity because, in principle, one can enumerate all finite and infinite game sequences and assign a different rule to each;)

Bill Spight wrote:For instance, if you have two double ko deaths on the board, under Japanese and Korean rules (IIUC) and the player with the "dead" stones has the move, she can hang the game by going back and forth between them. Under AGA and New Zealand rules, and the first Ing rules, they form a superko. Under the last Ing rules they are simply dead. The Ing rules simplify the game.

That would probably be the simplest ruleset according to Bill's definition

RobertJasiek wrote:

Bill Spight wrote:The Ing rules do not alter the classification of regular kos.

What do you mean by this with respect to what I call "dead ko".

It is true that some dead kos are alive under AGA rules. Do you know of any that are alive under Ing rules?

WRT Ing ko rules, the problem is not whether they are dead or alive but whether they are a fighting or disturbing ko.