Ko ban goes away if you can't agree on which stones are dead?


#1

I’m not sure if this is a bug or a rule. I’m assuming it’s a rule because I’m still new, but it seems odd.

In a game I was playing recently (http://online-go.com/game/738220), I got into a ko fight. I won the ko fight, and then my opponent passed because the rest of the board was settled. I then passed, and my opponent immediately rejected the marked stones and played as though there hadn’t just been a ko fight (i.e. his play would have been illegal if he’d played there without passing).

This seems like the wrong outcome. If my opponent had another move, I wouldn’t have passed; I’d always understood it to be the case that two passes in a row ends the game. If there’s a dispute about the state of the game, shouldn’t the ko state of the game be preserved when play resumes? Does the game not actually end when both players pass? Is this just because of the software, or is it really the rule that if both players pass and then dispute the stones the ko threat goes away.


#2

In most rulesets passing lifts the ko ban. So if your opponent passes, you should fill the ko.


#3

So two passes in a row doesn’t end the game?


#4

As @KillerDucky said, you need to resolve the ko. In your game you could have also taken the white group.

Since it made no difference in the outcome of the game, I’d have just marked everything else and let the game end.

Since two passes is used to let the computer know that we wish to score and because a dispute restarts play, I feel ko state should also be preserved. In the new game tree, there is no longer any move after the flip of the ko.


#5

Thank you both for your answers. I agree with you, Traveller. I was caught off guard because I understood myself to be agreeing to end the game, not to be offering my opponent another move with the ko removed. It seems like a technicality that is beneath the simplicity and beauty of the game. I suppose those kinds of technicalities come up frequently around ko, but this one is unnecessary.


#6

What rulesets are you thinking of, @KillerDucky and @Traveller? I’m not an expert on the subtleties of different rule sets, but I shared Wax’s belief that a ko couldn’t be re-initiated immediately after two passes like that. So seeing this, I became curious and I looked at the AGA’s statement of the rules and the Japanese statement. Under AGA rules, Rule 6 states that a move which would return the board to an identical position from earlier in the game with the same player to play is illegal. This is precisely what happened in Wax’s game: At move 128 white returned the board to the same position as move 124, and it was black’s turn to play again. This should’ve been an invalid move under the AGA ruleset. In the Japanese rules, the comments to Article 7, Clause 2 state that when the game stops, ko recapturing also stops, even if play resumes to determine the life/death status of a group. So under the Japanese rules it shouldn’t have been legal to restart the ko after both players passed.

I’d appreciate it if anyone wants to illuminate these rules or could discuss the other major rule sets (Chinese, Ing, others?), but based on Japanese and AGA rules, I think this is a glitch in the software. And while it wasn’t outcome determinative in the game Wax cited, it theoretically could be in any ruleset that counts only the empty space you’ve surrounded (Like Japanese & AGA) since these rule sets penalize you for playing inside your own territory. Forcing a player to fill the ko even if the opponent passes is forcing that player to sacrifice 1 point.

I found the AGA rules here: http://www.usgo.org/files/pdf/completerules.pdf
I found the Japanese rules here: http://www.cs.cmu.edu/~wjh/go/rules/Japanese.html (Translated by James Davies)


#7

Also tagging @Wax.


#8

[Please note that I will yield to about anyone on proper interpretations of the rules.]

I wasn’t really focusing on a particular ruleset. I tend to agree (as I indicated above) that the ko state should also be preserved if the scoring is rejected. This would also agree with the section of the rules you referenced. This game was under Japanese rules. Under AGA with the passing stone, there is no advantage to passing if there is any question of the status of any stones, like a ko. Same would also apply for Chinese, but here you would have to resolve to ko before you filled any dame.

The status of the white stones has no impact on the outcome of the game, I would have left them dead and accepted the score (and the win). That was already White by more than 90.

That being said. It IS proper to resolve all kos before ending the game. Fill, or kill, would have worked here.

Notes for future play:

  • If you know you have lost and there is nothing you feel you can learn from continuing, you should resign.
  • If you know you have lost and want to play to gain experience, ask your opponent if they mind continuing.
  • If you don’t know that you have lost, keep making the best moves you can.

It rarely matters what the score was. Usually, only the outcome - win, lose, or jigo (tie) - matters.


#9

I was thinking of the Article 7 which deals with determining life and death:

In the confirmation of life and death after the game stops in Article 9, recapturing in the same ko is prohibited. A player whose stone has been captured in a ko may, however, capture in that ko again after passing once for that particular ko capture.

I guess it doesn’t really apply during play. So it’s probably technically a difference. The end result should be the same: Just because your opponent doesn’t have another ko threat doesn’t give you the right to not fill in a ko and end the game. That’s the spirit of the tradition for Japanese rules. Also Japanese rules are honestly a mess if you really get into the details about it.

As for AGA, you’re right those rules do not have pass lifts ko-ban. Instead a player should fill his own territory as a “ko threat”, and force the other player to fill the ko. Actually it’s sorta useless in that case usually because filling the ko doesn’t cost points in AGA.


#10

I have the feeling that on japanese rules the player who wants to continue the game has to “give” the turn to the other person first. So in this case the turn should have been for the person who took the ko. Then if he passes the opponent is free to retake the ko.

btw all this trouble because too lazy to fill a ko when the game is ending?-)


#11

Japanese rules do seem to indicate that I should have been given the first move, which would have let me end the ko. Which makes sense, because my move (to pass) was based on an understanding that my opponent would have no subsequent moves (since s/he had already passed); if my opponent wants to move again, i.e. s/he’s passing-unless-I-pass, it seems right that I should get the chance to reconsider.

Pempu, in some sense it was laziness, but my opponent passed, and there were no more real live questions on the board. I could have played and forced him to pass again before I passed, but I had no reason to suspect I should have to. The ko fight was settled, and the game was over.


#12

So your plan was to pass, and then when he resumes the game, fill the ko? In that case why not just fill the ko before passing?

When he passed, it’s with the understanding that he’s letting you win the ko. When you don’t actually follow through and win the ko, he disputes that and argues that you need to add a defensive move. At this point the spirit of the rules are to figure out if you needed to add a defensive move or not.

BTW the ko in your game is rather complex, during this discussion I’ve been assuming that if your opponent is allowed to take the ko first it’s going to cause a big problem.


#13

Many of us have the view that filling in the ko is a “follow through” of some kind, and that it’s “lazy” not to fill it. But to my mind the exact opposite it true. All the scoring systems that only count empty intersections penalize one for a play within one’s own territory. Intuitively I’ve always considered this to be part of the elegance of the game–it encourages the player to optimize their groups’ shapes to live with the fewest stones possible. It is scoring systems that count occupied intersections as points that encourage laziness because they do not penalize a player who can’t be bothered to consider whether there is still room for a successful invasion of his or her territory or not. Similarly, leaving the ko open with the understanding that there’s no point in filling it when the opponent’s only “threat” left is to play within his or her own territory strikes me as more in the spirit of the game as I take it. What purpose, either tactical or social, is served by having white pass, forcing black to play one last stone, having white pass again and then having black pass? It is redundancy in a game whose virtue is to demand efficiency.

Looking at it another way: forcing black to fill the ko after a pass effectively turns the pass into a small ko threat of it’s own which strikes me on a gut level as wrong. White should have to play a stone in order to insist that black fill the ko.

Aesthetic matters aside, I still think this could matter in a game that is within 1 point, and I’d love it if someone can point to a reference or article discussing the issue in different rule sets.


#14

Actually there is a precedent of this. The first example of http://senseis.xmp.net/?RuleDisputesInvolvingGoSeigen
There is no “point” to be gained by not filling the last ko.

Under modern rules the ko should be filled. There is some complicated “hypothetical play” thing that comes if something is disputed over the board. Then the one who can eat the ko will do so and the other player cannot play any ko threats and must use pass as a ko threat.

PS. I hate & love rules


#15

Under area-scored rule systems and AGA-like systems filling the ko doesn’t cost you anything, so you may as well. If one of those systems has a local ko rule instead of the AGA’s super-ko rule as cited above, then filling the ko becomes mandatory.

Japanese rules technically don’t allow play to continue to resolve disputes, since under pure territory scoring forcing a capture can change the score. Hence complicated rules like “bent 4 in the corner is dead.”


What are the rules?
#16

Japanese rules do have “hypothetical play” that will resolve most weird situations from the board. There is no specific “bent 4 in the corner is dead” rule. It all comes from the “hypothetical play”


#17

This is an old thread, but nobody mentioned it so I think I should:

Under Korean (NOT Japanese) rules, if one side has “obviously more ko threats” (this is not precise in the rules), then filling the last ko is not necessary, and the player can claim that one extra point of territory (just like in your case: you would have one extra point under korean rules as I understand, if there are no more ko threats and all dame are filled).

Under all other (“common”) rulesets, you should fill the ko at the end (although for different reasons, depending on the ruleset).

This is likely the only “important” (so to speak: it’s not very important at all, although it could change the result of super-close games) difference between korean and japanese rules.


#18

Saw this same thing in another game here:


@miyakawa vs. @xipuwu

It arguably makes a difference here. If black gets the point a E1, black wins. If it’s dame or, as happened in the game, white captures and fills the ko, white wins.

Dame feels fairest, but that’s off the cuff and not based on any rule set. @Benbob makes a good case above that black should get the point.


#19

Black simply made a mistake by not filling it.


#20

Agreed. I think you’re crazy not to fill a ko if your opponent passes. I understand not filling in all the dame before counting, but passing with stones unsettled would set my teeth on edge.