Life In 19x19
http://www.lifein19x19.com/

Japonese counting
http://www.lifein19x19.com/viewtopic.php?f=45&t=18300
Page 18 of 20

Author:  jann [ Sun Sep 19, 2021 11:13 am ]
Post subject:  Re: Japonese counting

Gérard TAILLE wrote:
My proposal : in order to avoid mentionning continuously this big flaw in J89 pass-each-time can't we agree to say that the correct interpretation is really this J89 pass-once with a ko pass required only for the first recapture?

The closed loop in double kos was never a thing, this is clear from the examples. The question is only about whether whether this only a precedential ruling/restriction at the moment (and subject of a slight rule correction in the future), or whether J89 already has a good logical explanation for that.

At the moment I'm 75% sure that pass-once is the official J89 rule. The commentary mentioned here spells this out almost literally (after passing for it, the ko becomes like if the game resumed - and no pass-for-ko required in resumption).

Another reason I feel more and more confident is the amount of noticeable work went into J89. Have you seen the official Korean or Chinese rules (or many other for that matter)? The richness of J89 commentary, the number of L/D examples and all the different aspects of L/D they examine show that J89 is the result of a lot of work and attention. It's hard to imagine such an obvious double ko flaw could have remained unnoticed (though the older version also had a few gross errors in comments and an example which may contradict this - maybe they came from different editors or so?).

Quote:
The number of remaining flaws are now very small indeed. Basically I identified three positions and in each of them the problem is due to the presence of a double ko.
These positions are the following:

You may want to dig through some old bestiaries. There were an original pass-for-ko counterexample a few decades ago. I don't clearly remember now but it was a relatively large position (9x9 or bigger) and it also demonstrated an unnecessary reinforcement move.

Also note ligtvector's example (which I considered the biggest flaw of J89) does not work, as kvasir noticed:

Quote:
Click Here To Show Diagram Code
[go]$$B
$$ ---------------------
$$ | O . O . X . . O . .
$$ | O O O X O X O O . .
$$ | X X X O . O . . . .
$$ | . . . O O O . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ -----------------[/go]

B cannot seem to do anything to the corner without capturing a W stone in the ko first, and this stone will be replayed and become uncapturable, so is an enabled stone.

Quote:
In the third position the problem is due to the pass-for-ko required only ONCE (in this example if you require a pass-for-ko for the TWO first ko captures then the problem disappears).

That doesn't sood like a good idea. Note that pass-once has a good theoretical rationale as is (freeze and simplify the dynamics related to the stop position, and let a pass for it "resume the game" there). Pass-twice sounds ad-hoc without any theory ground.

I also don't see a serious problem in your third example - the behavior will remain reasonable and any difference to pass-each-time is irrelevant if that is not the official rule. Maybe a single email to the Nihon Ki-in could answer the question. :)

Author:  Gérard TAILLE [ Sun Sep 19, 2021 11:32 am ]
Post subject:  Re: Japonese counting

jann wrote:
Gérard TAILLE wrote:
My proposal : in order to avoid mentionning continuously this big flaw in J89 pass-each-time can't we agree to say that the correct interpretation is really this J89 pass-once with a ko pass required only for the first recapture?

The closed loop in double kos was never a thing, this is clear from the examples. The question is only about whether whether this only a precedential ruling/restriction at the moment (and subject of a slight rule correction in the future), or whether J89 already has a good logical explanation for that.

At the moment I'm 75% sure that pass-once is the official J89 rule. The commentary mentioned here spells this out almost literally (after passing for it, the ko becomes like if the game resumed - and no pass-for-ko required in resumption).

Another reason I feel more and more confident is the amount of noticeable work went into J89. Have you seen the official Korean or Chinese rules (or many other for that matter)? The richness of J89 commentary, the number of L/D examples and all the different aspects of L/D they examine show that J89 is the result of a lot of work and attention. It's hard to imagine such an obvious double ko flaw could have remained unnoticed (though the older version had a few gross errors in comments which may contradict this - maybe they came from different editors or so?).

Quote:
The number of remaining flaws are now very small indeed. Basically I identified three positions and in each of them the problem is due to the presence of a double ko.
These positions are the following:

You may want to dig through some old bestiaries. There were an original pass-for-ko counterexample a few decades ago. I don't clearly remember now but it was a relatively large position (9x9 or bigger) and it also demonstrated an unnecessary reinforcement move.

Also note ligtvector's example (which I considered the biggest flaw of J89) does not work, as kvasir noticed:

Quote:
Click Here To Show Diagram Code
[go]$$B
$$ ---------------------
$$ | O . O . X . . O . .
$$ | O O O X O X O O . .
$$ | X X X O . O . . . .
$$ | . . . O O O . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ -----------------[/go]

B cannot seem to do anything to the corner without capturing a W stone in the ko first, and this stone will be replayed and become uncapturable, so is an enabled stone.

Quote:
In the third position the problem is due to the pass-for-ko required only ONCE (in this example if you require a pass-for-ko for the TWO first ko captures then the problem disappears).

That doesn't sood like a good idea. Note that pass-once has a good theoretical rationale as is (freeze and simplify the dynamics related to the stop position, and let a pass for it "resume the game" there). Pass-twice sounds ad-hoc without any theory ground.


Click Here To Show Diagram Code
[go]$$B
$$ ----------------------
$$ | P . P . Z . . O O . O
$$ | P P P Z O Z O O O O O
$$ | X X X O . O O . O . .
$$ | X . X O O O O O . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]


see https://www.lifein19x19.com/viewtopic.p ... 51#p267451. It looks the result is seki but I am not quite sure.

Author:  jann [ Sun Sep 19, 2021 11:35 am ]
Post subject:  Re: Japonese counting

I doubt a new permanent stone UNDER a captured one could ever be anything but enabled, neither that there is any restriction about where to look for enabled stones.

Author:  Gérard TAILLE [ Sun Sep 19, 2021 1:11 pm ]
Post subject:  Re: Japonese counting

jann wrote:
I doubt a new permanent stone UNDER a captured one could ever be anything but enabled, neither that there is any restriction about where to look for enabled stones.


Black to play : the result for white is the following
Click Here To Show Diagram Code
[go]$$W
$$ ----------------------
$$ | . X . X X W C W W C W
$$ | . . . X W C W W W W W
$$ | X X X W C W W C W . .
$$ | X . X W W W W W . . .
$$ | . X X . . . . . . . .
$$ | X X . . . . . . . . .
$$ | . . . . . . . . . . .
$$ | . . . . . . . . . . .[/go]


My understanding is that all the area contolled here by white is not due to the capture of the five white stones in the corner. It is white area in ANY CASE. I do see a white stone captured by black and then a white permanent stone under the captured one but I do not see a NEW permanent stone. It just like a nakade : the stone can be captured, is reestablished but is it a NEW permanent stone?

Author:  jann [ Sun Sep 19, 2021 1:19 pm ]
Post subject:  Re: Japonese counting

Compare to snapback. No change in area, the reason the original stone is alive is solely because a new one played under the original IS enabled by the capture.

Author:  Gérard TAILLE [ Sun Sep 19, 2021 1:37 pm ]
Post subject:  Re: Japonese counting

jann wrote:
Compare to snapback. No change in area, the reason the original stone is alive is solely because a new one played under the original IS enabled by the capture.


Click Here To Show Diagram Code
[go]$$W
$$ ----------------------
$$ | . W X X . . .
$$ | X X O X . . .
$$ | O O O X . . .
$$ | . . . . . . .
$$ | . . . . . . .[/go]

For snapback surely :wc: is alive because white can reestablished a stone under it.

Quote:
Click Here To Show Diagram Code
[go]$$B
$$ ---------------------
$$ | O . O . X . . O . .
$$ | O O O X W X O O . .
$$ | X X X O . O . . . .
$$ | . . . O O O . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ | . . . . . . . . . .
$$ -----------------[/go]

For the same reason no doubt at all :wc: is alive.
But then what about the status of the five white stones in the corner? When killing these stones white is not able to create a NEW permanent stone because the marked stone is already alive. My understanding is that the capture of the marked stones and its reestablishement is relevant when looking at the status of this specific stone but, if this stone is alive, it seems not relevant when looking at the status of another group of stone.

Author:  jann [ Sun Sep 19, 2021 1:54 pm ]
Post subject:  Re: Japonese counting

In J89 a newly played permanent stone is enabled, whether under the original or elsewhere. The condition is only that it must be related to / made possible by the capture: it couldn't have been played originally, if the stone in question weren't captured - like you cannot play another stone on top of the earlier one.

I don't see how you can have different opinion on snapback and this one. Either you recognize a replayed stone under an original (alive) one as what was enabled by the capture, or you don't.

Also note that this whole dead stone business is about completely dead stones in territory, that could be picked up anytime without problem or any strings attached - except the score penalty on cleanup moves. You do want to recognize any kind of negative consequences or problem with such capture (before granting for free).

Author:  Gérard TAILLE [ Sun Sep 19, 2021 2:30 pm ]
Post subject:  Re: Japonese counting

jann wrote:
In J89 a newly played permanent stone is enabled, whether under the original or elsewhere. The condition is only that it must be related to / made possible by the capture: it couldn't have been played originally, if the stone in question weren't captured - like you cannot play another stone on top of the earlier one.

I don't see how you can have different opinion on snapback and this one. Either you recognize a replayed stone under an original (alive) one as what was enabled by the capture, or you don't.

Also note that this whole dead stone business is about completely dead stones in territory, that could be picked up anytime without problem or any strings attached - except the score penalty on cleanup moves. You do want to recognize any kind of negative consequences or problem with such capture (before granting for free).


I understand Yann but I have to think about it before agreeing. I do not see clearly what could the consequences of this new interpretation (new interpretation for me OC).

Author:  jann [ Sun Sep 19, 2021 2:45 pm ]
Post subject:  Re: Japonese counting

One more small detail: when you play a "new uncapturable stone" under an original, it IS new even in the sense that the original was not an uncapturable stone, apparently.

Author:  Gérard TAILLE [ Mon Sep 20, 2021 7:16 am ]
Post subject:  Re: Japonese counting

What about this modification of lightvector's position ?

Click Here To Show Diagram Code
[go]$$B
$$ ---------------------
$$ | O . O . X X . . O . . O X . .
$$ | O O O X O X X O O O O O X . .
$$ | X X X O . O O O X X X X X . .
$$ | . . X O O O X X X . . . . . .
$$ | . . X X X X X . . . . . . . .
$$ | . . . . . . . . . . . . . . .
$$ | . . . . . . . . . . . . . . .[/go]

Author:  jann [ Mon Sep 20, 2021 7:58 am ]
Post subject:  Re: Japonese counting

Nice, this seems to work again since B first can capture all in confirmation.

However, since W not independently alive on either side, this seems less bad and easier to accept than the original (about the same as the other reinforcement examples).

Btw, I tried to dig up the old reinforcement example, but haven't find it, only a different one:

Click Here To Show Diagram Code
[go]$$B
$$ ---------------------
$$ | . . X X X X O X . .
$$ | X X O X O X O X . .
$$ | X O . O . O O X . .
$$ | O . O . O O X X . .
$$ | O . O O O X X . . .
$$ | O O O X X X . . . .
$$ | X X X X . . . . . .[/go]

Author:  Gérard TAILLE [ Mon Sep 20, 2021 8:12 am ]
Post subject:  Re: Japonese counting

jann wrote:
Nice, this seems to work again since B first can capture all in confirmation.

However, since W not independently alive on either side, this seems less bad and easier to accept than the original (about the same as the other reinforcement examples).


I do not understand what you mean Jann. Could you explain a little?

Author:  jann [ Mon Sep 20, 2021 8:38 am ]
Post subject:  Re: Japonese counting

I meant your example removes W's "enabling" defence, since even the ko and the other/both side of W will become captured (so no permanent W stone anywhere). But this is also why this example feels less unreasonable and the reinforcement easier to accept: W is less safe.

All these examples (including the one I posted above - not sure about the other one I faintly remember though) show the same: being barely safe on double ko isn't enough for J89 confirmation.

A related question is reinforcement where one is not necessary in normal game due to more ko threats. I seem to recall such differences mentioned in Korean rules, though I'm not sure since in modern times pass can always serve as ko threat. So disputes like these seem impossible today regardless of L/D rules.

Author:  Gérard TAILLE [ Mon Sep 20, 2021 10:22 am ]
Post subject:  Re: Japonese counting

jann wrote:
I meant your example removes W's "enabling" defence, since even the ko and the other/both side of W will become captured (so no permanent W stone anywhere). But this is also why this example feels less unreasonable and the reinforcement easier to accept: W is less safe.

All these examples (including the one I posted above - not sure about the other one I faintly remember though) show the same: being barely safe on double ko isn't enough for J89 confirmation.

A related question is reinforcement where one is not necessary in normal game due to more ko threats. I seem to recall such differences mentioned in Korean rules, though I'm not sure since in modern times pass can always serve as ko threat. So disputes like these seem impossible today regardless of L/D rules.


In the exemple you gave the dispute was there because one player has more ko threats than the other.
Here it seems quite different because the double ko is a dead double ko and that means that one player must have an infinite number of ko threats to survive. In such case my understanding is the following : if in normal play the player (with the almost dead group) is able to force a NO RESULT then it it fine for her, otherwhise the group is dead and the opponent is not forced to add a move. Why there is here an issue? In normal play it is really a dead double ko but in confirmation phase the pass-for-ko brings a side effect modifying the normal handling of the double ko.

Another (symetrical) example:

Click Here To Show Diagram Code
[go]$$B
$$ -------------------
$$ | O O X . X X O . .
$$ | O O O X . X O . .
$$ | X O O O X O O . .
$$ | . X O . X X O . .
$$ | X . X X . X O . .
$$ | X X O X X X O . .
$$ | O O O O O O O . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]


the theme is always the same : a double ko in which one player has taken the two ko and the opponent cannot retake one of the ko due to the required pass-for-ko. That the reason why I said it is a side effect of pass-for-ko because in normal play no problem at all exists.

Author:  Cassandra [ Mon Sep 20, 2021 10:33 am ]
Post subject:  Re: Japonese counting

Gérard TAILLE wrote:
Another (symetrical) example:

Click Here To Show Diagram Code
[go]$$B
$$ -------------------
$$ | O O X . X X O . .
$$ | O O O X . X O . .
$$ | X O O O X O O . .
$$ | . X O . X X O . .
$$ | X . X X . X O . .
$$ | X X O X X X O . .
$$ | O O O O O O O . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]

Please try to give us the solution to the following reverse tsume-go:

What were the last seven moves before the position shown was reached (INSIDE White's fencing group, as a matter of course)?

Author:  jann [ Mon Sep 20, 2021 11:06 am ]
Post subject:  Re: Japonese counting

Gérard TAILLE wrote:
the theme is always the same : a double ko in which one player has taken the two ko and the opponent cannot retake one of the ko due to the required pass-for-ko. That the reason why I said it is a side effect of pass-for-ko because in normal play no problem at all exists.

Sure, no doubt here. These reinforcement moves are undesired (but maybe tolerable) side effect, are incorrect in normal go, and cause one point of scoring error. This is the price J89 accepts in exchange of having its consistent L/D logic.

But note that pass-for-ko is NOT absolutely necessary for territory scoring. It is there 90% because of moonshine, but other solutions are also possible.

Earlier I mentioned a useful reference (imaginary) ruleset: confirmation with normal play plus some moonshine ko rule like yours, and global enable. That would solve these examples naturally (the defender plays the double ko, the attacker sooner or later would need to pass, the defender finishes the double ko and takes pass-alive control after).

Author:  Cassandra [ Mon Sep 20, 2021 10:07 pm ]
Post subject:  Re: Japonese counting

Click Here To Show Diagram Code
[go]$$B
$$ -------------------
$$ | O O X . X X O . .
$$ | O O O X . X O . .
$$ | X O O O X O O . .
$$ | . X O . X X O . .
$$ | X . X X . X O . .
$$ | X X O X X X O . .
$$ | O O O O O O O . .
$$ | . . . . . . . . .
$$ | . . . . . . . . .[/go]

Positions like these are patterns with NO value.


Click Here To Show Diagram Code
[go]$$B :b7: pass
$$ -----------------------------------------
$$ | O O Z . X X O . . | O O . . X X O . . |
$$ | O O O X . X O . . | O O O X . X O . . |
$$ | X O O O X O O . . | X O O O X O O . . |
$$ | . X O . X X O . . | . X O . X X O . . |
$$ | X . X X . X O . . | X . X X . X O . . |
$$ | X X O X X X O . . | X X O X X X O . . |
$$ | O O O O O O O . . | O O O O O O O . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ -----------------------------------------[/go]

Black made a mistake, and thereafter complains that the rules do not help him to make up for the loss caused by it?

Click Here To Show Diagram Code
[go]$$B
$$ -----------------------------------------
$$ | O O Z . X X O . . | O O . O X X O . . |
$$ | O O O X . X O . . | O O O X 7 X O . . |
$$ | X O O O X O O . . | X O O O X O O . . |
$$ | . X O . X X O . . | . X O . X X O . . |
$$ | X . X X . X O . . | X . X X . X O . . |
$$ | X X O X X X O . . | X X O X X X O . . |
$$ | O O O O O O O . . | O O O O O O O . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ -----------------------------------------[/go]

Black made a mistake, and thereafter complains that the rules do not help him to make up for the loss caused by it?

Click Here To Show Diagram Code
[go]$$B
$$ -----------------------------------------
$$ | O O X . X X O . . | O O X . X X O . . |
$$ | O O O Z . X O . . | O O O . 7 X O . . |
$$ | X O O O X O O . . | X O O O X O O . . |
$$ | . X O . X X O . . | . X O . X X O . . |
$$ | X . X X . X O . . | X . X X . X O . . |
$$ | X X O X X X O . . | X X O X X X O . . |
$$ | O O O O O O O . . | O O O O O O O . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ -----------------------------------------[/go]

Black made a mistake, and thereafter complains that the rules do not help him to make up for the loss caused by it?

Click Here To Show Diagram Code
[go]$$B
$$ -----------------------------------------
$$ | O O X . X X O . . | O O X . X X O . . |
$$ | O O O X . X O . . | O O O X . X O . . |
$$ | X O O O X O O . . | X O O O X O O . . |
$$ | . X O . Z X O . . | . X O 7 . X O . . |
$$ | X . X X . X O . . | X . X X . X O . . |
$$ | X X O X X X O . . | X X O X X X O . . |
$$ | O O O O O O O . . | O O O O O O O . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ | . . . . . . . . . | . . . . . . . . . |
$$ -----------------------------------------[/go]

Black made a mistake, and thereafter complains that the rules do not help him to make up for the loss caused by it?

Author:  Gérard TAILLE [ Wed Sep 22, 2021 12:55 pm ]
Post subject:  Re: Japonese counting

jann wrote:
Earlier I mentioned a useful reference (imaginary) ruleset: confirmation with normal play plus some moonshine ko rule like yours, and global enable. That would solve these examples naturally (the defender plays the double ko, the attacker sooner or later would need to pass, the defender finishes the double ko and takes pass-alive control after).

If confirmation is made with normal play and global enable how do you handled bent-4 and unremovable ko-threat somewhere else on the board?

Author:  jann [ Wed Sep 22, 2021 1:35 pm ]
Post subject:  Re: Japonese counting

Gérard TAILLE wrote:
jann wrote:
Earlier I mentioned a useful reference (imaginary) ruleset: confirmation with normal play plus some moonshine ko rule like yours, and global enable. That would solve these examples naturally (the defender plays the double ko, the attacker sooner or later would need to pass, the defender finishes the double ko and takes pass-alive control after).

If confirmation is made with normal play and global enable how do you handled bent-4 and unremovable ko-threat somewhere else on the board?

If would be alive OC - just like how it often works out in Chinese rules, where it needs to be played out.

There is no perfect solution here, because if the threat is not usable in actual play (would lose too much for too little gain), then forcing the opponent to carry out the bent4 capture in main game unfairly costs him maybe 5+ points on threat removal. On the other hand, if the threat would be usable / a good trade in normal play, calling the bent4 corner dead while it could not have been captured in reality have unfair cost of 10+ points.

This is one of the weakest points of territory scoring.

Author:  gennan [ Fri Sep 24, 2021 3:09 pm ]
Post subject:  Re: Japonese counting

In area scoring, bent 4 in the corner might only live if there are no unremovable ko threats (as far as I understand). That situation is very rare in real games. You need some seki or double ko of such a small size that its sacrifice reverses the game when the bent 4 lives. So in practice, I feel there is not much harm in declaring it dead in all cases under territory scoring. How many games have you played where this mattered? For me, I think exactly zero in the past 34 years (many thousands of games).

Page 18 of 20 All times are UTC - 8 hours [ DST ]
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/