Yes, it's all about making a double-ko inoperative.Matti wrote:Many years ago I discovered a patch for the "pass-for-ko rule. Even with it there wolud be other flaws in the J1989 rules, so I didn't bother to release it. Anyway:
In the confirmation phase:This keep the double ko seki alive. If there is additional stuff attached, a player may rotate the double ko seki n times to gain n tenukies to resolve the thing.
- A player may not recapture a ko before making a pass to recapture that particular ko.
A player may not recapture a ko for the second time before making two passes to recapture that particular ko.
A player may not recapture a ko for the third time before making three passes to recapture that particular ko.
Etc.
J89's pass-for-ko: Misinterpreted in the Western Go World?
-
Cassandra
- Lives in sente
- Posts: 1326
- Joined: Wed Apr 28, 2010 11:33 am
- Rank: German 1 Kyu
- GD Posts: 0
- Has thanked: 14 times
- Been thanked: 153 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)
Igo Hatsuyōron #120 (really solved by KataGo)
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
Sorry, but this is nonsense. There is no way a group is dead if it can only be captured by sacrificing / enabling capture of another independently alive group, this is obvious from the rules text, examples, commentary and spirit.Cassandra wrote:White's group left the board ...
... BEFORE Black's group disappeared.
Thus, Black's group at the left is dead, while White's group at the right is alive.
-
Cassandra
- Lives in sente
- Posts: 1326
- Joined: Wed Apr 28, 2010 11:33 am
- Rank: German 1 Kyu
- GD Posts: 0
- Has thanked: 14 times
- Been thanked: 153 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
The rules text, examples, commentary and spirit say that a group is considered "alive", if a permanent stone of its colour can be positioned on the board AFTER the group has been captured.jann wrote:Sorry, but this is nonsense. There is no way a group is dead if it can only be captured by sacrificing / enabling capture of another independently alive group, this is obvious from the rules text, examples, commentary and spirit.Cassandra wrote:White's group left the board ...
... BEFORE Black's group disappeared.
Thus, Black's group at the left is dead, while White's group at the right is alive.
This is true for White's single stone, as well as for her group, at the right.
If you wanted to keep the original J89's "enable" alive in the current J89: If White had decided to capture the other ko-shape in the double-ko, Black would have captured into the remaining ko at the left, and White would have been helpless (Black would never pass for "his" ko-shape to be recaptured). Thus, her sacrifice at the right enabled her to capture Black's group at the left (with the side effect of having placed several permanent stones of hers).
In contrast, the disappearance of Black's group at the left "enabled" nothing.
Black is always free to fill his own territory. Nobody can prevent him from doing so.
-----------------------------------
The fate of Black's group at the left depends on ko, while everything White around is unconditionally alive. Therefore, Black's group is "dead" outright, according to Japanese understanding of this issue.
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)
Igo Hatsuyōron #120 (really solved by KataGo)
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
You are free to invent new rules, but J89 enable doesn't work this way, even among the official examples.Cassandra wrote:The rules text, examples, commentary and spirit say that a group is considered "alive", if a permanent stone of its colour can be positioned on the board AFTER the group has been captured.
The right side was not B territory, especially the intersections with big W seki string. Even with your incorrect interpretation, B can play new alive stones (even AFTER the capture of his left!) on the right at intersections where he could not have played at originally, if W haven't tried to capture his left. So even with the (nonexistent) timing constraint you logic doesn't work here.In contrast, the disappearance of Black's group at the left "enabled" nothing.
Black is always free to fill his own territory. Nobody can prevent him from doing so.
-
Cassandra
- Lives in sente
- Posts: 1326
- Joined: Wed Apr 28, 2010 11:33 am
- Rank: German 1 Kyu
- GD Posts: 0
- Has thanked: 14 times
- Been thanked: 153 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
jann,jann wrote:You are free to invent new rules, but J89 enable doesn't work this way, even among the official examples.Cassandra wrote:The rules text, examples, commentary and spirit say that a group is considered "alive", if a permanent stone of its colour can be positioned on the board AFTER the group has been captured.
The right side was not B territory, especially the intersections with big W seki string. Even with your incorrect interpretation, B can play new alive stones (even AFTER the capture of his left!) on the right at intersections where he could not have played at originally, if W haven't tried to capture his left. So even with the (nonexistent) timing constraint you logic doesn't work here.In contrast, the disappearance of Black's group at the left "enabled" nothing.
Black is always free to fill his own territory. Nobody can prevent him from doing so.
I did not invent new rules.
But YOU will have do make up your mind:
If you wanted to consider L&D globally, you would have to also accept "enable" globally.
If you wanted to have "enable" only locally, you would have to consider L&D also locally. And in this case, Black's group at the left would be more than dead outright.
And I do not think that it is so very surprising that capturing something would enable you to occupy points that you were not able to occupy before. If you liked this idea of yours, you could turn everything on the board "alive" that could, but yet has not been, taken off the board.
If you are able to occupy the SAME points at the right before AND after Black's group at the left was taken off the board, this occupation is completely independent from what happened at the left. Just filling your own territory, for what reason ever.
------------------------------------
What is the reason that you do not like the result at all?
Would you have wanted anything else but "Black at the left is dead; at the right, there is a double-ko seki"? If so, what?
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)
Igo Hatsuyōron #120 (really solved by KataGo)
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
Japanese rules judge L/D globally, with global enable (and with modified ko rules). No local view was ever mentioned in this case (that would be Korean rules btw).
With your rules and your analysis, B left cannot be captured without enabling new alive B stones at the right - stones that couldn't have been played without the capture attempt of the left.
Capturing his dead stones never enable HIM to play a new uncapturable stone that he couldn't have played originally, without being captured.
With your rules and your analysis, B left cannot be captured without enabling new alive B stones at the right - stones that couldn't have been played without the capture attempt of the left.
Not you, your opponent.And I do not think that it is so very surprising that capturing something would enable you to occupy points that you were not able to occupy before.
Capturing his dead stones never enable HIM to play a new uncapturable stone that he couldn't have played originally, without being captured.
-
Cassandra
- Lives in sente
- Posts: 1326
- Joined: Wed Apr 28, 2010 11:33 am
- Rank: German 1 Kyu
- GD Posts: 0
- Has thanked: 14 times
- Been thanked: 153 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
Your answers?Cassandra wrote:What is the reason that you do not like the result at all?
Would you have wanted anything else but "Black at the left is dead; at the right, there is a double-ko seki"? If so, what?
I can understand that you absolutely did not expect the self-atari in the double-ko, which ruined your intended counter-example.
However, this is no reason to continue to talk around the bush, which includes nothing more than your understanding of what should not work in the sense of your supposed counter-example, but which failed.
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)
Igo Hatsuyōron #120 (really solved by KataGo)
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
I cannot even make sense of your questions now. Both of YOUR analysis lines ended up W losing the right in exchange for the left (enabling new B uncapturable stones). Black on the left is alive in your rules. It doesn't matter if I like this or not.
-
Cassandra
- Lives in sente
- Posts: 1326
- Joined: Wed Apr 28, 2010 11:33 am
- Rank: German 1 Kyu
- GD Posts: 0
- Has thanked: 14 times
- Been thanked: 153 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
I don't think my questions were so terribly misleading.jann wrote:I cannot even make sense of your questions now.
Next attempt:
What was the intended result of your alleged counter-example?
Which status would YOU have assigned to the Black group on the left?
What status do YOU think the groups in the partial position on the right should have?---------------------------
By the way: I have already understood that you were shocked by White's self-atari in the double-ko. And also that you assess your interpretation of "enable" to be the correct one.
However, this all will not hinder you from telling us the result of YOUR analysis of the above mentioned L&D statuses.
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)
Igo Hatsuyōron #120 (really solved by KataGo)
-
Gérard TAILLE
- Gosei
- Posts: 1346
- Joined: Sun Aug 23, 2020 2:47 am
- Rank: 1d
- GD Posts: 0
- Has thanked: 21 times
- Been thanked: 57 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
I like challenge and, for fun, I tried to find a position with different result.jann wrote:Nice position, thank you. Even if harmless here since the result is the same (B can force the reinforcement with the same play either way), this does show a case where the attacker might need to play the same ko twice in confirmation.Gérard TAILLE wrote:1)jann wrote:This is a longstanding question in J89 which has two possible answers:
- Maybe passing for each ko is only required once for each player (see here)
- Or the problem was overlooked by J89 authors, and will be fixed in the future (with some hack like requiring a normal pass of either player before passing for a ko a second time)
If you use "pass once per ko" then black can pass immediatly and wins the game because the two white marked stones are dead in the confirmation phase
2) without using "pass once per ko", in order to win, black must continue at least one move.
Afterthen black passes and wins the game.
I believe the following one works.
Let's call J89-ko-pass the J89 rule with the ko-pass instead of the pass-for-ko
Let's call J89-pass-once-per-ko the J89 rule with the pass-once-per-ko instead of the pass-for-ko
Assume that black will win the game by 0.5 points if white adds three moves in normal play to kill the black stones in the upper right corner.
What is the point? Assume white has added the two marked stones before stopping the game.
In J89-ko-pass the upper right black group is dead => white wins the game => in normal play black must force the NO RESULT
In J89-pass-once-per-ko black group is alive by: and black cannot be killed because black managed te recreate the moonshine life loop wih three ko.
=> in normal play white must add a third move => black wins the game
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
Nice position again! This example shows more than just a difference between pass-once and pass-each-time (which we knew to be possible, at least theoretically).
First, let me refer to this. The supporting theory between the two approaches/interpretations differ: pass-once tries to freeze kos in relation to the stopped position, pass-each-time alters ko play balance permamently.
This is nothing like a moonshine life where the defender uses a double ko to create a faint illusion of life - this is the inverse. It is the attacker (W) who can use the double ko to force a perpetual (unsuccessful) capture attempt (of a locally alive group) and thus a valid triple ko.
This case seems similar to J89 erroneous reinforcement examples (lightvector's and others) discussed recently, where J89 ko passes change normal play and force extra plays that are incorrect and unnecessary in normal game. Pass-once seems to work better here in retaining the balance if the stopped position. Since black is uncapturable even locally (and on miai, not on infinite threats from other double kos), seems hard to justify calling him dead.
First, let me refer to this. The supporting theory between the two approaches/interpretations differ: pass-once tries to freeze kos in relation to the stopped position, pass-each-time alters ko play balance permamently.
In this position, not only is black right uncapturable, he can make miai of the two kos at the two edges, so is alive/uncapturable even locally. (EDIT: after 2 W reinforcement moves this is not really true anymore since B would need to play first for that now)Gérard TAILLE wrote: Assume white has added the two marked stones before stopping the game.
This is nothing like a moonshine life where the defender uses a double ko to create a faint illusion of life - this is the inverse. It is the attacker (W) who can use the double ko to force a perpetual (unsuccessful) capture attempt (of a locally alive group) and thus a valid triple ko.
This case seems similar to J89 erroneous reinforcement examples (lightvector's and others) discussed recently, where J89 ko passes change normal play and force extra plays that are incorrect and unnecessary in normal game. Pass-once seems to work better here in retaining the balance if the stopped position. Since black is uncapturable even locally (and on miai, not on infinite threats from other double kos), seems hard to justify calling him dead.
Last edited by jann on Thu Sep 16, 2021 2:04 pm, edited 1 time in total.
-
Gérard TAILLE
- Gosei
- Posts: 1346
- Joined: Sun Aug 23, 2020 2:47 am
- Rank: 1d
- GD Posts: 0
- Has thanked: 21 times
- Been thanked: 57 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
I do not see clearly your point Jann because here it seems the contrary. In J89-ko-pass white needs to add only two moves in order to kill black and in J89-pass-once-per-ko white needs to add three moves.jann wrote:This case seems similar to J89 erroneous reinforcement examples (lightvector's and others) discussed recently, where J89 ko passes change normal play and force extra plays that are incorrect and unnecessary in normal game. Pass-once seems to work better here in retaining the balance if the stopped position. Since black is uncapturable even locally (and on miai, not on infinite threats from other double kos), seems hard to justify calling him dead.
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
I meant this example is the same as lightvector's reinforcement example, where in normal play one side is clearly alive on (creatable) miai, but J89 erroneously calls him dead (because of heavily restricted ko play).
It is pass-each-time that fails in both examples (OC, in lightvector's case pass-once wouldn't help either).
It is pass-each-time that fails in both examples (OC, in lightvector's case pass-once wouldn't help either).
-
Gérard TAILLE
- Gosei
- Posts: 1346
- Joined: Sun Aug 23, 2020 2:47 am
- Rank: 1d
- GD Posts: 0
- Has thanked: 21 times
- Been thanked: 57 times
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
You say black is uncapturable?jann wrote:Nice position again! This example shows more than just a difference between pass-once and pass-each-time (which we knew to be possible, at least theoretically).
First, let me refer to this. The supporting theory between the two approaches/interpretations differ: pass-once tries to freeze kos in relation to the stopped position, pass-each-time alters ko play balance permamently.
In this position, not only is black right uncapturable, he can make miai of the two kos at the two edges, so is alive/uncapturable even locally.Gérard TAILLE wrote: Assume white has added the two marked stones before stopping the game.
This is nothing like a moonshine life where the defender uses a double ko to create a faint illusion of life - this is the inverse. It is the attacker (W) who can use the double ko to force a perpetual (unsuccessful) capture attempt (of a locally alive group) and thus a valid triple ko.
This case seems similar to J89 erroneous reinforcement examples (lightvector's and others) discussed recently, where J89 ko passes change normal play and force extra plays that are incorrect and unnecessary in normal game. Pass-once seems to work better here in retaining the balance if the stopped position. Since black is uncapturable even locally (and on miai, not on infinite threats from other double kos), seems hard to justify calling him dead.
In normal play OC I agree.
But in confirmation phase? If you assume neither player is allowed to use the double ko as an infinite number of ko threats then black is always dead due to the independence of ko which exists in confirmation phase.
Re: J89's pass-for-ko: Misinterpreted in the Western Go Worl
Compare your example to lightvector's. In both cases one side has a decisive play that makes miai of two local kos and creates life on double ko, thus safe in normal play.
J89 doesn't recognize such life, but this is clearly a flaw/anomaly, at least in lightvector's case (revealing this defect is the very purpose of that position). In your example pass-once may actually helped.
You can also test these examples in Korean-style rules with explit local view (Korean, LJRG, or even yours). In both examples the miai life works even locally, independent of double ko sekis.
J89 doesn't recognize such life, but this is clearly a flaw/anomaly, at least in lightvector's case (revealing this defect is the very purpose of that position). In your example pass-once may actually helped.
You can also test these examples in Korean-style rules with explit local view (Korean, LJRG, or even yours). In both examples the miai life works even locally, independent of double ko sekis.