Here a new version of "GT territory rule", taking into account the comments already received.
Though I have in my head the concept of "advangeous loop" I think now that I could write the rule without defining explicitly this notion and OC without changing the ideas of the rule I developped in this thread through examples.
My new DRAFT is the following:
Preliminary definitions:
A "two-eye-formation" is a set of one or several strings of the same player and exactly two empty intersections so that each of the strings is adjacent to each of the two intersections, none of the strings is adjacent to another empty intersection, and each of the two intersections is adjacent only to the strings.
The "inside border" of a set of locations is all the locations in the set which are adjacent to a location not in the set
The "outside border" of a set of locations is all the locations outside the set which are adjacent to a location in the set
A set of locations is a "controlled area" for a player if:
1) The inside border contains neither empty locations nor stones of the opponent.
2) The outside border contains neither empty locations nor stones of the player.
3) the set of locations can be entirely covered by a "two-eye formation" through an hypthetical play (see below)
A "player's territory" is all the locations which are empty or occupied by an opponent stone, in a "player's controlled area".
Confirmation phase:
In the confirmation phase any player may claim she "controls an area". This player is called the "attacker" and the other player is called the "defender".
In an hypthetical play the attacker is trying to build a "two-eye formation" covering all the potential "controlled area" while the goal of the defender is the opposite.
Definition:
A position is said to be "critical" if this position is reached by a defender's ko capture followed by an attacker's pass
Hypothetical play
At the beginning of an hypothetical play there are no ko ban
The hypothetical play begins always by a defender move (a play or a pass) and then each player makes moves alternatively.
Normal play is used during hypothetical play except that all the "permanently prohibited kos" created have to be taken into account (the defender is not allowed to capture a ko which have been "permanently prohibited"
Three successive passes ends always an hypothetical play
Procedure to create "permanently prohibited ko":
As soon as a "critical position" is reached the attacker may (it is not mandatory) claim for creating a "permanently prohibited ko":
1) The attacker proves that she can either reach her objective or makes an infinite number of pass
2) The attacker proves that the defender cannot makes an infinite number of pass
3) If the defender agrees to point 1) and 2) then the game continues up to the following "critical position"
4) As soon as a new critical position is reached a "permanently prohibited ko" is automatically created for the ko capture made by the defender before the last pass
5) Then the game continue taking into account the "permanently prohibited ko" created and the attacker may later create another "permanently prohibited ko" using again the procedure above.
Comment 1 : in a critical position it is always the defender to move.
Comment 2 : the points 1) and 2) above ask for an agreement between players. To reach this agreement another board can be used if necessary. If no agreement is possible the referee will have the last word.
GT territory rule
-
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
-
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: GT territory rule
For a better consistency between points 1) 3) and 4) it seems better to change the wording of point 1) toGérard TAILLE wrote:s soon as a "critical position" is reached the attacker may (it is not mandatory) claim for creating a "permanently prohibited ko":
1) The attacker proves that she can either reach her objective or makes an infinite number of pass
2) The attacker proves that the defender cannot makes an infinite number of pass
3) If the defender agrees to point 1) and 2) then the game continues up to the following "critical position"
4) As soon as a new critical position is reached a "permanently prohibited ko" is automatically created for the ko capture made by the defender before the last pass
5) Then the game continue taking into account the "permanently prohibited ko" created and the attacker may later create another "permanently prohibited ko" using again the procedure above.
1) The attacker proves that she can either reach her objective or reach "critical positions" an infinite number of times
-
Mike Novack
- Lives in sente
- Posts: 1045
- Joined: Mon Aug 09, 2010 9:36 am
- GD Posts: 0
- Been thanked: 182 times
Re: GT territory rule
Excuse me please, but why white 4 = pass? Why with move four does white not play the bulky 5 atari? Then black must capture, white plays in the middle, and so the black group to the right dies. Where upon white, now kills the rightmost group of five black stones. So THAT claim was correct. However the other group of five black stones remains alic=ve in seki.Cassandra wrote:Excuse me folks, but drawing diagrams would help you a lot, as well as your audience.
I know that this activity can be very painful at times, especially when you have to put the same sequence on paper for the 153rd time.
However, diagram painting enables your audience to easily understand your explanations, as well as the appearance of "Why not here?" moments with yourself.
Let's this be the final position of "play", and White claiming "territory" at any of the triangled points.
=> White claims that she will be able to establish a "two-eye-formation" somewhere on the left part of the board.
=> Black has the right of moving first.
THE END
-
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: GT territory rule
My mistake has been noticed even earlier => viewtopic.php?p=266881#p266881Mike Novack wrote:Excuse me please, but why white 4 = pass?
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)
-
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: GT territory rule
Once White has taken Black's right-hand group off the board, the seki will be resolved completely from right to left.Mike Novack wrote:However the other group of five black stones remains alive in seki.
As I already mentioned earlier several times, painting diagrams would help a lot
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: GT territory rule
Your first question could explain the second. W's left is big to make sure W never want to actually use that in ko threat (small left may worth throwing away in game).Gérard TAILLE wrote:BTW Jann, why not having "simply" use the following position: with the same result?jann wrote: W claims a chunk of territory in the center with middle B string (the one between symmetric W groups) dead. Inner and outer borders are OK and W can, even with B starting, take pass-alive control of that part of the board and capture 5 stones. Right?
Yes Jann I agree with you, white can claim for a territory in the center. This position looks a seki in J2003 and this seki result seems more in the spirit of japonese rule. Agreed.
Let me try to understand the issue.
The position is not seki because of anything wrt J2003 - it is seki by general Japanese (or general go) logic. W cannot do anything in reality since going for the capture would be disadvantageous (cost much higher than the gain). LD analysis should reflect go reality (which is same under Chinese rule).
This allows confirmation results very different from real go outcomes (no "enable" like J89, so W can play lines he would never in reality - since consequences ignored).jann wrote:The problem with the approach of this thread - as opposed to big sister LJRG - seems to be allowing global play without looking at global consequences.
BTW you should always try to look at the logic behind rules (examples come later, nitpicking on wordings lastly / never). Go has centuries of history, the concepts and cohesive logic in real, widely used rules are simple and somewhat robust (even if modern formalizations aren't yet).
For others: W not claimed anything about left B string, only the middle one.
-
Mike Novack
- Lives in sente
- Posts: 1045
- Joined: Mon Aug 09, 2010 9:36 am
- GD Posts: 0
- Been thanked: 182 times
Re: GT territory rule
No, the leftmost string of five black stones remains alive in seki with the white group to its left (there are two dame between these groups)Cassandra wrote:Once White has taken Black's right-hand group off the board, the seki will be resolved completely from right to left.Mike Novack wrote:However the other group of five black stones remains alive in seki.
As I already mentioned earlier several times, painting diagrams would help a lot
-
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: GT territory rule
As I said, painting diagrams would help a lotMike Novack wrote:No, the leftmost string of five black stones remains alive in seki with the white group to its left (there are two dame between these groups)Cassandra wrote:Once White has taken Black's right-hand group off the board, the seki will be resolved completely from right to left.Mike Novack wrote:However the other group of five black stones remains alive in seki.
As I already mentioned earlier several times, painting diagrams would help a lot
And / or explaining a little, little bit move in detail.
Please excuse my blind spot, but sometimes I am blind in all eyes. Especially when other topics are circling around in the container that was made for these.
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: GT territory rule
How works GT territory rule in the now well known above example.
White claim for controlling the marked area above.
Borders are OK => An hypothetical play will begin with black to play first and white trying to build a two-eye formation covering all the marked area
note that the move
is here allowed because normal play is used in confirmation phase.
and it is clear white will mange to build a two-eye formation on all the claimed area => white controlled the claimed are => white has a territory in it
White claim for controlling the marked area above.
Borders are OK => An hypothetical play will begin with black to play first and white trying to build a two-eye formation covering all the marked area
note that the move
is here allowed because normal play is used in confirmation phase.and it is clear white will mange to build a two-eye formation on all the claimed area => white controlled the claimed are => white has a territory in it
-
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: GT territory rule
Your comments sound quite interesting but I have difficulties to understand what is really your understanding of this general Japanese (or general go) logic.jann wrote:Your first question could explain the second. W's left is big to make sure W never want to actually use that in ko threat (small left may worth throwing away in game).
The position is not seki because of anything wrt J2003 - it is seki by general Japanese (or general go) logic. W cannot do anything in reality since going for the capture would be disadvantageous (cost much higher than the gain). LD analysis should reflect go reality (which is same under Chinese rule).
This allows confirmation results very different from real go outcomes (no "enable" like J89, so W can play lines he would never in reality - since consequences ignored).jann wrote:The problem with the approach of this thread - as opposed to big sister LJRG - seems to be allowing global play without looking at global consequences.
BTW you should always try to look at the logic behind rules (examples come later, nitpicking on wordings lastly / never). Go has centuries of history, the concepts and cohesive logic in real, widely used rules are simple and somewhat robust (even if modern formalizations aren't yet).
For others: W not claimed anything about left B string, only the middle one.
In chinese GO the logic is clear, and because there are no confirmation phase at all, the same logic applies till the very end of the game. In that sense chinese GO is perfect.
In japanese GO the logic is not as clear, though japanese GO has been played for centuries
. The issue is that when game stops by agreement after "normal play" it is difficult to resolve the dispute by just using normal play.In order for me to really understand your comment, can you show how you apply this japanese GO logic in the three following very common examples:
Re: GT territory rule
Yes. But we know Japanese go and Chinese go are closely tied to each other. So there must not be any outrageously large differences in outcomes. And even in Japanese go you can imagine what would happen if the game would continue (ignoring the small potential score drift caused by unanswered cleanup moves - or using pass stones). In my above example, there is no doubt about the correct ruling - Japanese and Chinese go agree (as is usually the case).Gérard TAILLE wrote:Your comments sound quite interesting but I have difficulties to understand what is really your understanding of this general Japanese (or general go) logic.
In chinese GO the logic is clear, and because there are no confirmation phase at all, the same logic applies till the very end of the game. In that sense chinese GO is perfect.
In japanese GO the logic is not as clear, though japanese GO has been played for centuries
Moonshine life is one particular case (maybe the only one) where the basic rules logic may not be sufficient. But note this is NOT specific to Japanese go but also the same in Chinese go!In order for me to really understand your comment, can you show how you apply this japanese GO logic in the three following very common examples
The question is, of course, the logical difference between real triple ko (which is draw in both Japanese and Chinese) and moonshine kos (which treatment varied a bit in history but the current consensus is that it is dead). Several interesting ideas appeared about this in the past, you can find reference material on usenet and elsewhere if interested.
-
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: GT territory rule
OK let's put aside triple ko and moonshine and their loop handling.jann wrote:Yes. But we know Japanese go and Chinese go are closely tied to each other. So there must not be any outrageously large differences in outcomes. And even in Japanese go you can imagine what would happen if the game would continue (ignoring the small potential score drift caused by unanswered cleanup moves - or using pass stones). In my above example, there is no doubt about the correct ruling - Japanese and Chinese go agree (as is usually the case).Gérard TAILLE wrote:Your comments sound quite interesting but I have difficulties to understand what is really your understanding of this general Japanese (or general go) logic.
In chinese GO the logic is clear, and because there are no confirmation phase at all, the same logic applies till the very end of the game. In that sense chinese GO is perfect.
In japanese GO the logic is not as clear, though japanese GO has been played for centuries
Moonshine life is one particular case (maybe the only one) where the basic rules logic may not be sufficient. But note this is NOT specific to Japanese go but also the same in Chinese go!In order for me to really understand your comment, can you show how you apply this japanese GO logic in the three following very common examples
The question is, of course, the logical difference between real triple ko (which is draw in both Japanese and Chinese) and moonshine kos (which treatment varied a bit in history but the current consensus is that it is dead). Several interesting ideas appeared about this in the past, you can find reference material on usenet and elsewhere if interested.
How do you apply the japanese logic to bent four in the corner beside an unremovable ko threat in a seki?
Re: GT territory rule
Bent4 is less interesting for rule theory. Japanese invented local L/D, partly as an attempt to fix moonshine life. A doubtful hack IMO but is deeply rooted in Japanese minds now. And once they went this route bent4 become dead even with unremovable threats. Not good but in itself not unacceptably bad either. Just for fun you can also try this example in J89 with enable but without pass for ko.Gérard TAILLE wrote:How do you apply the japanese logic to bent four in the corner beside an unremovable ko threat
-
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: GT territory rule
I begin to understand your view Jann.jann wrote:Bent4 is less interesting for rule theory. Japanese invented local L/D, partly as an attempt to fix moonshine life. A doubtful hack IMO but is deeply rooted in Japanese minds now. And once they went this route bent4 become dead even with unremovable threats. Not good but in itself not unacceptably bad either. Just for fun you can also try this example in J89 with enable but without pass for ko.Gérard TAILLE wrote:How do you apply the japanese logic to bent four in the corner beside an unremovable ko threat
Maybe I am wrong but let'try:
Assume I take Bill Spight "encore" and,
Assume I modify this "encore" by replacing the pass-ko rule by my "permanently prohibited ko" in order to handle loops according to japanese traditions, and
Assume finally I manage to take care of unsided dame which cause some problems with "encore" relatively to the seki in japanese rule (it is not that difficult to do indeed)
Isn't it exactly the perfect japonese rule you have in mind?
In this "perfect rule" you can note OC that bent four in the corner is not uncondionnaly dead if unremovable ko threat exists (it will be a good point for you won't it?)
-
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: GT territory rule
For the time being two positions are an issue for my "GT territory rule".
Firstly the following position I analysed in viewtopic.php?p=266754#p266754
I concluded in "GT territory rule" that black has a territory in the upper right corner though a seki seems more appropriate.
And secondly the position found by Jann and showed in viewtopic.php?p=266868#p266868
Here "GT territory rule" concludes to territory though a seki seems more appropriate.
Two positions is already too many isn't?
That means that I missed something. I think the problem is exactly the same for these two positions:
If in the first position you assume that the outside border is independantly alive then black cannot anymore kill the three white stones
Similarly if in this second position you assume that the outside border is independantly alive white cannot anymore kill the five black stones.
IOW to be a territory need one more condition.
I propose the following improvement to my draft rule (see points 3 and 4 under):
A set of locations is a "controlled area" for a player if:
1) The inside border contains neither empty locations nor stones of the opponent.
2) The outside border contains neither empty locations nor stones of the player.
3) the set of locations can be entirely covered by a "two-eye formation" through an hypthetical play (see below) played only on the "controlled area", assuming the outside border independantly alive
4) the set of locations can be entirely covered by a "two-eye formation" through an hypthetical play (see below) played on all the board
Comment1 : in 99% of cases the sequence found in 3) works also for 4) but in rare cases 3) and 4) might be independant.
Comment2 : due to comment1 (!) I did not see any impact on all other examples I presented here
Firstly the following position I analysed in viewtopic.php?p=266754#p266754
I concluded in "GT territory rule" that black has a territory in the upper right corner though a seki seems more appropriate.
And secondly the position found by Jann and showed in viewtopic.php?p=266868#p266868
Here "GT territory rule" concludes to territory though a seki seems more appropriate.
Two positions is already too many isn't?
That means that I missed something. I think the problem is exactly the same for these two positions:
If in the first position you assume that the outside border is independantly alive then black cannot anymore kill the three white stones
Similarly if in this second position you assume that the outside border is independantly alive white cannot anymore kill the five black stones.
IOW to be a territory need one more condition.
I propose the following improvement to my draft rule (see points 3 and 4 under):
A set of locations is a "controlled area" for a player if:
1) The inside border contains neither empty locations nor stones of the opponent.
2) The outside border contains neither empty locations nor stones of the player.
3) the set of locations can be entirely covered by a "two-eye formation" through an hypthetical play (see below) played only on the "controlled area", assuming the outside border independantly alive
4) the set of locations can be entirely covered by a "two-eye formation" through an hypthetical play (see below) played on all the board
Comment1 : in 99% of cases the sequence found in 3) works also for 4) but in rare cases 3) and 4) might be independant.
Comment2 : due to comment1 (!) I did not see any impact on all other examples I presented here