Triple ko in Tiger's Mouth Tourney

[HermanHiddema]

From the AGA E-Journal comes this game:

http://www.usgo.org/news/wp-content/uploads/2013/05/TM-Triple-Ko1.sgf

Although an AGA event, the game was played with Japanese rules. At move 257, the players reached an impasse in the upper left. Three kos, how now? They asked the TD for a resolution, because they did not want a void game. The TD ruled the situation seki, so that they could continue their game, after which white won. Personally, I think that ruling seki in case of long cycles is a nice idea (In the past I posted it to SL and called it the Long Cycle Seki Rule.

The E-Journal used it as context to talk about AGA rules with its superko, and how such rules would have avoided the need for a ruling altogether.

But there is a problem. It is not actually triple ko. After move 255, black can tenuki. Black is alive in double ko anyway, and the outside ko is irrelevant. At game end, under Japanese rules, the white group would then be dead.

It remains a tricky beast, this ko rule

[jts]

Huh. So how are you imagining this would be resolved in hypothetical play?

[oren]

My suggestion is to get rid of superko and void the game, but I don't think that will go well with AGA rules head.

[Boidhre]

Isn't it a mutual double ko with two of the kos being for one side only? Or am I completely misunderstanding this?

[shapenaji]

The big issue with superko is that it's impractical for many players to remember where the cycle started. It's an elegant solution, but part of me just likes the existence of chusei and voided games.

In this case, I would say "Well, black doesn't need to fight the triple ko... unless he's losing by less than the size of the white group, in which case he has the ability to void the game"

For white, the best result is then a voided game, but he still has to play for it! I actually think that's kind've cool.

[pwaldron]

The big issue with superko is that it's impractical for many players to remember where the cycle started. It's an elegant solution, but part of me just likes the existence of chusei and voided games.

In principle the server could keep track of the repetition and would mark the illegal moves. Like you, though, I'm of the opinion that there's nothing wrong with a voided game.

[yoyoma]

Huh. So how are you imagining this would be resolved in hypothetical play?

If the position remains on the board to the end, see this example:
http://www.cs.cmu.edu/~wjh/go/rules/Japanese.html
life-and-death example 8

[jts]

Isn't it a mutual double ko with two of the kos being for one side only? Or am I completely misunderstanding this?

The issue (in Japanese rules) is that when you do hypothetical play, the only way to lift the ko ban is to say pass and also name the specific ko you are passing for.

So if I understand correctly, if W goes first in hypothetical play, W takes, B takes, W passes for the internal ko that B just took, and then B is forced to pass again for the other internal ko. (If B took the external ko, W could take the second internal ko, and B would have no move.) So if W goes first the cycle never stops.

But if B goes first, he can take the external ko first (atari), W takes an internal ko, B takes the other internal ko (atari again), and then W's only move is to pass for one of the ko, and B captures.

This brings up a point of hypothetical play I had never actually thought about. Do both players have the right to a round of hypothetical play in which their color moves first?

[RobertJasiek]

Since the game was played under (some) Japanese rules, the players did not have any right to ask for seki and the TD did not have any right to declare it a seki.

Depending on which Japanese ruleset is used, rules application differs:

http://home.snafu.de/jasiek/rules.html

However, there are two basic possibilities under Japanese rules:
- The players recycle and let the game be void.
- The players stop the game using passes and apply hypothetical status analysis.

Tournament rules overriding 'void game' could be a factor.

[Bill Spight]

Since the game was played under (some) Japanese rules, the players did not have any right to ask for seki and the TD did not have any right to declare it a seki.

Depending on which Japanese ruleset is used, rules application differs:

http://home.snafu.de/jasiek/rules.html

However, there are two basic possibilities under Japanese rules:
- The players recycle and let the game be void.
- The players stop the game using passes and apply hypothetical status analysis.

Tournament rules overriding 'void game' could be a factor.

Since the tournament was on KGS, were the rules Nihon Kiin Japanese rules or KGS "Japanese" rules?

[Mef]

Since the game was played under (some) Japanese rules, the players did not have any right to ask for seki and the TD did not have any right to declare it a seki.

As far as I know, under the AGA tournament guidelines, if there is a request by both players for the TD to make a decision, they are free to use their discretion in making that decision. There's a provision for making an appeal in the case that one player thinks the decision is in error, but here it seems that both players found this ruling to be satisfactory. I would say the TD had every right to declare a seki. I can only hope that in a given tournament the worst thing a TD does is resolve a situation in a way that both players find equitable.

[Bill Spight]

Although an AGA event, the game was played with Japanese rules. At move 257, the players reached an impasse in the upper left. Three kos, how now? They asked the TD for a resolution, because they did not want a void game. The TD ruled the situation seki, so that they could continue their game, after which white won.

The TD ruling is obviously incorrect (except perhaps by local KGS rules, which I do not know). It may not be obvious that White is dead, but it is obvious that a triple ko is not seki. Still, the TD's decision stands, I suppose.

I am not exactly sure what the players asked for, but it is always appropriate for the TD to inform them of the rules. Whether it is appropriate for him to tell them the status of groups is another question. Generally he shouldn't, but if both players want to know, I do not see anything that prohibits him from doing so. He should get it right, however. In this case, suppose that Black found out that White was dead before the game result was finalized. The misinformation by the TD could have affected the play.

Ing rules are the only ones in which the corner is seki, AFAIK.

[Bill Spight]

This brings up a point of hypothetical play I had never actually thought about. Do both players have the right to a round of hypothetical play in which their color moves first?

That's how it works. If either player, playing first, can enforce a capture (without thereby enabling another uncapturable stone to be played by the opponent) then the captured stone or stones are dead. That's how you get what is normally called double ko seki to have two dead stones, one for each side. The Japanese '89 rules are either clever or bizarre, depending on your viewpoint.

[Bill Spight]

Since the game was played under (some) Japanese rules, the players did not have any right to ask for seki and the TD did not have any right to declare it a seki.

I would say the TD had every right to declare a seki. I can only hope that in a given tournament the worst thing a TD does is resolve a situation in a way that both players find equitable.

IMO, the players have a right to ask any question they want to. They do not have the right to get the answer, however, unless the question is about the rules. The TD had a right, if I understand what happened, to offer his or her opinion about the status of the corner. That is not the same thing as declaring the corner to be a seki. That was wrong.