Continued from

What's wrong with suicide ?


Harleqin, should we use a new topic ?

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

Here are Cassandra's ideas for Japanese-style territorial rules.

In my opinion, self-consistency is the most important requirement for the development of a rule set. It goes without saying that there will be some (minor) differences especially to what is known as examples to the 1989 Nihon Kiin rule set. But this cannot be a problem, because it is unavoidable anyway.


####################
General
####################

There are three different stages in a game of Go:
Play - Evaluate - Count.

Each of them can be treated independently (e. g. for combining ideas for rules), but has to be kept separately (no mishmash in the order).

I suppose that no deeper discussion will develop on "Count". For the other two stages there are several options, which would result in self-consistence, so I will highlight what my preference is.


####################
Stage 1: Play
####################


########## How to handle "recurrence of situations" ? ##########

Most important question to answer for "Play" is how to handle "recurrence of situations". I will use "situation" as compound for "player to move" & "arrangement of stones on the board" & "difference of captured stones".

I prefer the point of view following the Japanese tradition of "neither victory nor defeat".

It's just a matter of practicability. I think it will be much easier for the overwhelming majority of players to react after a "recurrence of situations" has been realised. This is even more important, as such recurrence will not happen very often.


If you ban "recurrence of situations", every player in every game has to bear every "situation" after every move in mind before placing his next stone. Will the result really be worth the effort ?
In addition, it will be necessary to decide if and how handle "recurrence of situations" during "Evaluate".


########## What about "suicide" ? ##########

The least interesting question concerns "suicide".

I prefer the rule that you must not place a stone, so that is has no lifeline left.

I think that "suicide" is contradictory to Japanese and / or East-Asian thinking. Of course, it would be possible to allow it. But you have to be aware that there will be a backlash to the definition of what a move is. May be that there will be a backlash to the question of "recurrence of situations", too.


####################
Stage 2: Evaluate
####################

The most critical stage (concerning options for rules) is "Evaluate".

Over years I have won the firm conviction that you must decide on two important elements,

A) Is evaluation "local" or "global" ?
B) How to define "territory" ?

before you can discuss

C) How to evaluate ?

Of course, the core answer to B) has to be given in "Count", but there is a backlash to "Evaluate".


########## "Local" or "global" scope ? ##########

My preference for A) is "local".
May be it's a matter of taste, but this preference follows Japanese tradition. Perhaps this tradition has developed from the way of "infighting" in Japan, where long-distance weapons were not known for a long period of time. And where wheels were not in use until the mid of the 19th century.
A decision for "local" has the effect that there is no "long-distance" impact beyond the border set by "living groups".

Please note that I think it possible to have a consistent "global" evaluation (which comes from area-based rule sets) with territorial-oriented rules, too. But in my opinion there will be several additional rules necessary, especially in respect to Ko positions and the need of Teire.

What Harleqin quoted of Honinbo Shusai (one point of territory in a Ko that cannot be won by the opponent) can be true (and consistent) only with global evaluation.
There will be a backlash to B), too. It will be necessary to define territory as something like "unoccupied points within chains of stones of only one colour".
Consistency of the rule set then requires - in my opinion - to allow this point of territory in an open Mannen Ko, too. Generalising, you will have to count territory in Seki, eventually.

If you do not want to do this, you immediately face the need of adding some rules, stating in which cases an open Ko point counts as territory and in which it does not.

To distinguish from the quoted opinion of Honinbo Shuei (moonshine Ko lives) you will have some work to do with C), therein especially with the definition of "living groups".


########## How to do "Evaluate" ? ##########

If your preference for A) is "global", the consistent method for C) is "by actual play".
In my eyes it is impossible to have mix of "global" evaluation (if Ko is concerned) and of a "local" one (if questions of Life & Death of "normal" single groups are affected).


Because I prefer "local" for A), it should be clear that I have another suggestion for C).

Let me give some preliminary notes. You must be aware that there is the great danger to mix up the three phases, especially "Evaluate" and "Count". What may be the reason for the partially unsatisfying treatment of some "examples" in the 1989 Nihon Kiin rule set.

It is very important that taking "dead" opponent's stones off the board take place in "Count", after "Evaluate" has been terminated and the status of each chain of stones has been determined. It will become clear later, what this has to do with what I stated earlier: "no 'long-distance' impact beyond the border set by 'living groups'."


What follows, is my suggestion for processing "Evaluate".

"Evaluate" will take place in two sub-stages:

1) Determine the "status" of each chain of stones.
2) Combine the results of 1).

Determining the status of each chain begins with a move of the opponent (may be a Pass).

There are three "status":

a) "2-eyed"
b) Not a), but cannot be "captured"
c) Not a), and not b)

Remark to a): "2-eyed" is the compound of "already has two eyes (minimum)" or "can get two eyes" or "can be captured, but the complete captured shape can be filled again with stones, which will become part of a 2-eyed group", the latter concerning snap-back stones, for example.
Remark to b): "Captured" means "taken off the board completely, without rest". "Rest" may be a stone played at any of the points of the captured shape, which cannot be taken thereafter (= becomes part of a 2-eyed group). Refers to the "x points without capturing" corner shapes, for example.
Remark to c): That means that the chain can be taken off the board completely (= can be "captured").


After all results have been found, they are combined as follows:

d) Chains in status a), "2-eyed", are called "alive".
e) Chains in status c), which are situated within opponent's "alive" chains, are called "dead".
f) All other chains are called "in Seki".

Perhaps you have realised that there is no "domino-effect". Evaluation in sub-stage 1) takes place with all stones on the board, which later in sub-stage 2) will get the status "dead". So there is no automatism like "After I have taken this dead Bent-Four off the board, I can resolve this Double-Ko-Seki, too, in my favour.", for example. What would mean to enter "Count", return to "Evaluate" and re-enter "Count".
So not everything, which does not have two eyes, but cannot be taken off the board during "Play", will suddenly disappear during "Evaluate" & "Count" as prisoners.


####################
Stage 3: Count
####################

My suggestion for "Evaluate" has some consequences for "Count".

First, all "dead" chains of stones are taken off the board and count as prisoners for the opponent, as usual.

Thereafter, each empty point of the board that lies within "alive" chain(s) is counted as "territory".

I'm aware that this definition of territory, with its strict interconnection to "life", differs slightly from the "traditional" one. But I'm sure that it is simpler to apply. There are no difficulties with open Ko points, for example.

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

If you work out your ideas more carefully and separate them from guesses about Japanese rules history, then maybe it might become worthwhile to comment.

Too difficult to understand ?

PLAY

• "Recurrence of situations" will result in "neither victory nor defeat".
• "Suicide" remains forbidden.

EVALUATE

• Scope is "local".
• Status is determined for each chain of stones individually.
• Opponent moves first.
• Policy is: "You must own what you claim to possess."
• So determine whether the chain* is or can become 2-eyed (1) or can be eleminated completely** off the board (2).

COUNT

• Opponent's (2) inside your (1) are taken off the board as your prisoners.
• Territory are the unoccupied points inside your (1).

-------------------
* including its successors on each of the primary occupied point(s)
** no successor on any of the primary occupied point(s)

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

Ko rules / clarifications:

- I assume that you also want to use some basic ko rule.
- Since you do not state it in the summary version, let me add what you state in the long version: recurrence relies on situations.

Ko rules / comments:

- You fail to state in general how "neither victory nor defeat" compares to scores for the sake of making strategic decisions: You make the same mistake as J1989 of creating undecidable strategy in positions like one play before either double ko seki or triple ko.
- The "will result" with the implication "occurs or has occurred" is an elegant tournament rules method for the purpose of implementing no-result-like rules provided intention of that implication has been made clear. From the text alone, it is not clear enough. E.g., it is unclear what happens if both repetition and a score result have occurred; which takes precedence?
- Your rule means that also cycles with unequal numbers of prisoners removed during them lead to "neither victory nor defeat". I.e., a player being behind by potential scoring but with the option of starting a sending-2-returning-1 cycle might do that. Well, who cares when it comes down to your preference - but do you really think that your rules would find lots of friends under such circumstances?
- Suppose a sending-2-returning-1 is on the board and has not been played in during the Play phase. Applying your rule during Evaluation means that evaluation is interrupted by the creation of a "neither victory nor defeat". Presumably this is your intention, is it?

Evaluate rules / first clarifications:

- "2-eyed" you describe as "is the compound of 'already has two eyes (minimum)' or 'can get two eyes' or 'can be captured, but the complete captured shape can be filled again with stones, which will become part of a 2-eyed group'".
- You introduce 3 status types: 2-eyed, completely removable, everything else.

Evaluate rules / first comments:

- What is "local"? Does it even matter?
- "the compound of" might mean "a connected part of the board". It is unclear whether this is the intended meaning.
- "already has two eyes (minimum)" is known as "2-eye-formation".
- "can get two eyes" in the context of "already has two eyes (minimum)" might intended to be a terribly confusing alternative expression for "can be forced to be transformed into a 2-eye-formation". Else define "eye"! (You will notice that eye must be defined via first defining 2-eye-formation.)
- "can be captured, but the complete captured shape can be filled again with stones" creates a new design problem: Suppose we have a shape with 2 unfilled intersections, stones are captured from the shape, then a 2-eye-formation is created with 2 (partly or fully) DIFFERENT unfilled intersections. Why do you create such a new problem?
- Why do you introduce such a complicated life concept as "2-eyed"?! "2-eye-formation" is much simpler: It does not need any subtypes.
- You should study the everything else type carefully. Undesired side effects might occur. In particular, capturable-2 strings should be studied carefully.
- AFAICS, sekis do not have territory. You seem to offer a rather elegant treatment for that.
- Unlike J1989 and J2003, you use a local maximal instead of a local minimal requirement for life. Presumably you find this aesthetically appealing but thus far it is very risky because Chris Dams's proof of the relation between J2003-life and WAGC-life relies on the local minimal requirement. Until you prove the same for the local maximal requirement (if such an equivalence exists at all...), you do not achieve the same powerful universal characterization of life. Yours would be a weaker form of life. This does not make your construction attractive (yet).
- Needless to say, you make the standard mistake of using grammar ("can") instead of contents ("can force").
- Provide examples!

@Robert:

Cassandra defined situation in the earlier post to include "difference in prisoners", so unequal cycles such as "send two return one" do not result in "neither victory nor defeat".

"already has two eyes (minimum)" is not exactly the same as "2-eye-formation", as 2-eye-formation does not allow three eyes.

I see, thanks! I should have trusted Cassandra's summary yet less, it seems.

I would be interested in how you would treat the life and death examples No. 4 and 16 from the Nihon Kiin rules.

_________________
A good system naturally covers all corner cases without further effort.

Dear Robert, thanks for your annotations.


Some general notes in this context.

• "Ideas" is no synonym for "explicit rule text".
• I do not see any sense in discussing the colour of the surface. So I will not give definitions of elements (e. g. "board", "move"), which had not been the source for troubles with the 1989 Nihon Kiin rules, for example, in my understanding. And I suppose that there will be people, who could give better definitions than mine in the sense what you seem to understand as "definition".
• This will be true for "Ko", too. I have not realised yet that there might be any problems to understand that a player must not immediately recapture a stone, which just captured one of his / her stones.

Further quotes of your posting will be written in Blue.

- You fail to state in general how "neither victory nor defeat" compares to scores for the sake of making strategic decisions: You make the same mistake as J1989 of creating undecidable strategy in positions like one play before either double ko seki or triple ko.
"Score" is a term, which will become relevant in "Count".
"Recurrence of positions" occurs in "Play".
So the player can make up his / her mind if he / she wants "neither victory nor defeat" or wants to play on and to stop playing Triple Ko, for example.

- The "will result" with the implication "occurs or has occurred" is an elegant tournament rules method for the purpose of implementing no-result-like rules provided intention of that implication has been made clear. From the text alone, it is not clear enough. E.g., it is unclear what happens if both repetition and a score result have occurred; which takes precedence?
See above. There can be no coexistence of "recurrence of positions" and "score".
What tournament rules will say about "recurrence of positions", is a different kettle of fish.

- Your rule means that also cycles with unequal numbers of prisoners removed during them lead to "neither victory nor defeat". I.e., a player being behind by potential scoring but with the option of starting a sending-2-returning-1 cycle might do that. Well, who cares when it comes down to your preference - but do you really think that your rules would find lots of friends under such circumstances?
See

So "sending-2-returning-1" does not create a "recurrence of situations". Perhaps you mixed it up with area-oriented rules.

- Suppose a sending-2-returning-1 is on the board and has not been played in during the Play phase. Applying your rule during Evaluation means that evaluation is interrupted by the creation of a "neither victory nor defeat". Presumably this is your intention, is it?
No, it is not. There is no "interruption" in determining the status of a chain of stones.
If you encounter a repetitive sequence during "Evaluate", you will neither be able to have the status "2-eyed" for the chain nor can you prevent this chain from being captured.

- "2-eyed" you describe as "is the compound of 'already has two eyes (minimum)' or 'can get two eyes' or 'can be captured, but the complete captured shape can be filled again with stones, which will become part of a 2-eyed group'".
- You introduce 3 status types: 2-eyed, completely removable, everything else.
I think it should be clear that "2-eyed" means "has 2 eyes or more". You can reduce every multi-eyed group to a 2-eyed one by closing the remaining eye-points by your own. It's just a matter of practicability to realize that you can give the status "2-eyed" to every chain that can be connected to an already "2-eyed" one.
May be that it is more difficult to make a computer program realising a "secure" connection than to make it reducing something to only 2 single eye-points.

In my understanding you have to think about 2 "status" only. "Everything else" can be described verbally, but is no explicit status on its own.
May be that I have to sharpen the wording.

The status "2-eyed" is clear, but there are two options for the second status. In the moment, I'm absolutely not sure which one might be better for general application (or understanding).

• You can look for chains, which cannot be captured (and can be found in what is known as Seki, for example) or can be captured, but the opponent is unable to occupy all of the chain's primary points (what can be found in "x points without capture", for example).
• You can look for chains, which can be captured and the opponent is able to occupy all of the chain's primary points.

Both options result in the other one being "everything else". May be that the first one will be more appropriate when Ko shapes are involved.

In my opinion one has to be aware about the equivalence of (single) "stone" and "chain". It must make a difference, whether the primary points of a chain, after it has been captured, will be occupied again by stones of only one colour or by stones of both sides. The latter cannot happen with a single stone, so both cases must not be mixed up.

- What is "local"? Does it even matter?
In my opinion there are two fundamental different approaches for designing a rule set. "Local" is the approach I prefer. May be I'm too old to become fascinated by the other one, "global". The "local" approach usually can be found in territory-oriented rule sets, the "global" approach is typical for area-oriented rule sets. To stress it a second time: It is possible to generate a consistent rule set with each of both approaches.

"Local" will result in a kind of "borders" generated by what usually is understood as "living groups". None of these separated parts of the board will be influenced by what happens in another one.
If we put "basic Ko" aside, this will be true even during "Play".

- "the compound of" might mean "a connected part of the board". It is unclear whether this is the intended meaning.
"compound of A & B & C" = "A or B or C".

- "already has two eyes (minimum)" is known as "2-eye-formation".
- "can get two eyes" in the context of "already has two eyes (minimum)" might intended to be a terribly confusing alternative expression for "can be forced to be transformed into a 2-eye-formation". Else define "eye"! (You will notice that eye must be defined via first defining 2-eye-formation.)
As written above, I will not open a theatre of war about definitions and / or wording.

- "can be captured, but the complete captured shape can be filled again with stones" creates a new design problem: Suppose we have a shape with 2 unfilled intersections, stones are captured from the shape, then a 2-eye-formation is created with 2 (partly or fully) DIFFERENT unfilled intersections. Why do you create such a new problem?
Perhaps you can show an example ?
It seems important to me that all of the chain's primary points will become part of something "2-eyed". Its 2 eye points must be somewhere else, so I think that it will be possible to fill the chain's primary points completely.

- Why do you introduce such a complicated life concept as "2-eyed"?! "2-eye-formation" is much simpler: It does not need any subtypes.
If "2-eye-formation" is the meta-type of what I defined and is understood widely by the public, there will be no problem.

- You should study the everything else type carefully. Undesired side effects might occur. In particular, capturable-2 strings should be studied carefully.
Perhaps you can show an example ?

- AFAICS, sekis do not have territory. You seem to offer a rather elegant treatment for that.
"Territory" is only within what has been determined as "alive".
But perhaps you will find territory, where you did not expect it.

The process of "Evaluate" is "determine the status of all chains" (ONCE !!!), thereafter combine the results (ONCE !!!).

So chains of stones that have got the status "alive" during "Evaluate", will not suddenly become "dead", after some other "dead" stones have been taken off the board (what can happen only during "Count"). Perhaps the result of handling some very special positions will resemble what could be expected after "resolving by actual play" (during "Evaluate", typical for area-oriented rules) much more than with the 1989 Nihon Kiin rules.

I think, some displeasure with the 1989 Nihon Kiin rules handling of positions, which cannot be resolved by actual play (during "Play"), results from a certain mixture of "Evaluate" and "Count", despite some inexactness in the rules' text.

Harleqin's example of Honinbo Shusai's open Ko would have a result as follows.

• The open point is no territory (in the "local" context, the opponent can capture and connect during "Evaluate").
• If there is a cutting point "behind" the Ko (at which the connection to an already "2-eyed" group would be possible at any time), there will be no need to fill the Teire. This Teire will count as point of territory.

- Unlike J1989 and J2003, you use a local maximal instead of a local minimal requirement for life. Presumably you find this aesthetically appealing but thus far it is very risky because Chris Dams's proof of the relation between J2003-life and WAGC-life relies on the local minimal requirement. Until you prove the same for the local maximal requirement (if such an equivalence exists at all...), you do not achieve the same powerful universal characterization of life. Yours would be a weaker form of life. This does not make your construction attractive (yet).
Do not know what you want to say.
I think, the way to get to know what is "alive", is clearer than in the 1989 Nihon Kiin rules.

- Needless to say, you make the standard mistake of using grammar ("can") instead of contents ("can force").
See above about "wording".

- Provide examples!
Perhaps you can name some of your favourites among the 1989 Nihon Kiin examples ?

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

#4

Let's start with determining the status of White's upper side chain (left side one is identical).

Click Here To Show Diagram Code

[go]$$
$$ -------------------
$$ | . X 1 2 X X X W X
$$ | X X W W W W W W X
$$ | . O X X X X X X X
$$ | . O X , . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

Black begins.

Click Here To Show Diagram Code

[go]$$
$$ -------------------
$$ | . X X O . 3 . O X
$$ | X X O O O O O O X
$$ | . O X X X X X X X
$$ | 4 O X , . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

Click Here To Show Diagram Code

[go]$$
$$ -------------------
$$ | . X X O 5 X 7 O X
$$ | X X O O O O O O X
$$ | 6 O X X X X X X X
$$ | O O X , . . . . .
$$ | . O X . . . . . .
$$ | . O X . . . . . .
$$ | . O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

White's chain is eliminated.


Status of Black's upper side chain (left side one is identical).

Click Here To Show Diagram Code

[go]$$
$$ -------------------
$$ | . X . 2 B B B O X
$$ | X X O O O O O O X
$$ | . O X X X X X X X
$$ | . O X , . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

White begins and captures the chain. If she does nothing, we return to the sequence before.

Click Here To Show Diagram Code

[go]$$
$$ -------------------
$$ | . X . O . 3 . O X
$$ | X X O O O O O O X
$$ | . O X X X X X X X
$$ | 4 O X , . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | X O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

Click Here To Show Diagram Code

[go]$$m5
$$ -------------------
$$ | . X 5 O 3 X 7 O X
$$ | X X O O O O O O X
$$ | . O X X X X X X X
$$ | O O X , . . . . .
$$ | . O X . . . . . .
$$ | 1 O X . . . . . .
$$ | . O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

Click Here To Show Diagram Code

[go]$$
$$ -------------------
$$ | . X X . B B B . X
$$ | X X . . . . . . X
$$ | . O X X X X X X X
$$ | O O X , . . . . .
$$ | . O X . . . . . .
$$ | X O X . . . . . .
$$ | . O X . . . . . .
$$ | O O X . . . . . .
$$ | X X X . . . . . .[/go]

All primary occupied points have become part of a Black 2-eyed group.

White cannot do anything with Black's chain in the corner, so this one is "2-eyed", too.

Click Here To Show Diagram Code

[go]$$W
$$ -------------------
$$ | . 2 . . 2 2 2 P X
$$ | 2 2 P P P P P P X
$$ | . P X X X X X X X
$$ | . P X , . . . . .
$$ | 2 P X . . . . . .
$$ | 2 P X . . . . . .
$$ | 2 P X . . . . . .
$$ | P P X . . . . . .
$$ | X X X . . . . . .[/go]

So we have White chains in status c) (let's call them "remouvable" - I have no better short term in the moment) inside Black chains in status "2-eyed".

White's chains are "dead", Black's chains are "alive".

As you can see, both of White's chains are seperated by "living" Black groups. So there is no side-effect from one chain to the other. What I said when explaining "local".


#16

White cannot do anything here, so we will start determining the status of the single White stone on the left.

Click Here To Show Diagram Code

[go]$$
$$ -----------------------------
$$ | 1 O O . O . O X . X O . O X
$$ | W X X O O O O X X O O O O X
$$ | X X . X X X O X O . O X X X
$$ | . . . , X O X X X O O X . .
$$ | . . . . X O X . X O X X . .
$$ | . . . . X O X X X O X . . .
$$ | . . . . X O O O O X X . . .[/go]

Black captures.

Click Here To Show Diagram Code

[go]$$
$$ -----------------------------
$$ | X O O . O . O X 2 X O . O X
$$ | 3 X X O O O O X X O O O O X
$$ | X X . X X X O X O . O X X X
$$ | . . . , X O X X X O O X . .
$$ | . . . . X O X . X O X X . .
$$ | . . . . X O X X X O X . . .
$$ | . . . . X O O O O X X . . .[/go]

Whatever happens on the right, Black will be able to connect.

If we want to determine the status of White's two-stone chain in the upper left, ...

Click Here To Show Diagram Code

[go]$$
$$ -----------------------------
$$ | X W W 3 O . O X O . O . O X
$$ | . X X O O O O X X O O O O X
$$ | X X . X X X O X O . O X X X
$$ | . . . , X O X X X O O X . .
$$ | . . . . X O X . X O X X . .
$$ | . . . . X O X X X O X . . .
$$ | . . . . X O O O O X X . . .[/go]

... Black will capture this chain with 3.

Click Here To Show Diagram Code

[go]$$
$$ -----------------------------
$$ | X . 5 X O . O X O . O . O X
$$ | . X X O O O O X X O O O O X
$$ | X X . X X X O X O . O X X X
$$ | . . . , X O X X X O O X . .
$$ | . . . . X O X 4 X O X X . .
$$ | . . . . X O X X X O X . . .
$$ | . . . . X O O O O X X . . .[/go]

White captures Black's chain in the middle, Black connects on the left. It does not help White to recapture at 5 with 4.

As will have become clear here, Black will not be able to capture White's left-hand 7-stone-group. This chain cannot be turned into "2-eyed" by White, but will remain on the board, so let us call this type of status "stable" in short (again I have not found a better term).

Click Here To Show Diagram Code

[go]$$
$$ -----------------------------
$$ | . P P . @ . @ # . Z @ . @ X
$$ | P X X @ @ @ @ # # @ @ @ @ X
$$ | X X . X X X @ # P . @ X X X
$$ | . . . , X O # # # @ @ X . .
$$ | . . . . X O # . # @ X X . .
$$ | . . . . X O # # # @ X . . .
$$ | . . . . X O O O O X X . . .[/go]

The big chains in the Double-Ko-Seki on the right are stable, too.
Both single stones in Atari cannot reach a "stable" status, so they are "removable".

Combining what we have found, we see chains, which are "removable", but are not situated between "2-eyed" opponent's chains. There are neither "alive" chains nor "dead" ones, so the whole position is a giant Seki.

This does not represent the 1989 Nihon Kiin rules result, I know. But I'm convinced that there must be a clear distinction between "Evalutate" and "Count". The method of taking "dead" stones off the board without capturing them is just a shortcut. Otherwise you would have to capture them in reality and get compensation by your opponent for your moves, either by playing into territory or by presenting a prisoner.

"Killing" White's 7-stone chain would only be possible after eleminating her single stone on the left. But this would mean changing the position at the end of "Play", which sets the foundation for "Evaluate".

Black would have had the one-sided option to turn this position into "neither victory nor defeat" during "Play", using the Triple-Ko.

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

Cassandra, there are two types of basic ko rules: referring to either the previous move (2-move rule) or the previous play (intervening passes possible).

Your answer "So the player can make up his / her mind if he / she wants 'neither victory nor defeat' or wants to play on and to stop playing Triple Ko, for example." is a fake and does not solve the problem for the decision-making player since the player CANNOT MEANINGFULLY make up his mind for a strategic decision since he does not have any comparison between scores and "neither victory nor defeat". He might as well draw a lot to decide whether to choose a result with a score or the result "neither victory nor defeat".

That there is no coexistence of "recurrence of positions" and "score" is clear but this is not the problem. The problem occurs EARLIER: at some moments when to decide whether LATER there will be either "recurrence of positions" or a score.

Concerning sending-2-returning-1, I have meant the recurrence every 4 moves.

"If you encounter a repetitive sequence during "Evaluate", you will neither be able to have the status "2-eyed" for the chain nor can you prevent this chain from being captured.": I do not understand this yet. What exactly happens when a situation recurs during Evaluation?

If you mean "has 2 eyes or more" with "2-eyed", then you should give that thing a different name!

"It's just a matter of practicability to realize that you can give the status '2-eyed' to every chain that can be connected to an already '2-eyed' one.": One must be careful with such statements. E.g., it is easy to construct a double threat example so that A is 2-eyed, either B or C is 2-eyed, but B and C are not both 2-eyed. Would you call such nice Japanese-style rules? I do not.

"The status "2-eyed" is clear": Not it is not. Define "has 2 eyes or more"! I would define it as follows for a group and a considered set of empty intersections: "On the intersection union of the group and the set, we get a 2-eye-formation for each arbitrary, fixed filling by stones of the group's colour of a selection of all but 2 intersections of the set." However, I cannot know whether you mean this or something very different.

"the equivalence of (single) 'stone' and 'chain'": Which "equivalence" do you mean?

Your "local" is a wish. Imagine the game to enter Evaluation with a position we would perceive as middle game. You do not get such nice partition boundaries as you hope for.

If you want to escape clear definitions of essential terms by means of meta-discussion ("theatre"), then you will face never-ending criticism about great unclarity of your rules. Mind you, you can easily apply earlier research and simply write "can force" where necessary or point to, say, "details as in J2003". But not defining even on the user-friendly level "eye" or "eyes (minimum)" makes your own rules attempt ridiculous. (Hint: I would define eye as follows: "An 'eye' is an empty intersection of a two-eye-formation." Then I would make your higher level terms fitting to that if I were you. Currently I cannot do that since I do not know your intended meaning.)

Currently I lack time for trivial examples. Maybe you will have more luck in autumn.

"It seems important to me that all of the chain's primary points will become part of something "2-eyed". Its 2 eye points must be somewhere else, so I think that it will be possible to fill the chain's primary points completely.": Take John Tromp's example of a string consisting of 360 stones. You will see a problem easily: Life is possible, but not on all the 360 initial stone intersections.

"2-eye-formation": Definition see on various of my webpages or at Sensei's. It is not the meta-type of what you defined but you could reduce some of your life by forcing its transformation into a 2-eye-formation.

You can find capturable-2 examples on my webpages.

I do not see as much interaction of Evaluation and Count in J1989 as you fear.

It is essential that you understand what I say about local maximal instead of a local minimal requirement for life. Maximal is what you want: Finally life on all initial intersections of a string. Minimal is what J1989, J2003 and Chris Dams's proof of equivalence between J2003-life and WAGC-life use: Finally life on at least one of all initial intersections of a string. Start reading here: http://home.snafu.de/jasiek/wagcmod.html

I do not care which examples you use but yours should explain and distinguish your terms from each other!

This seems a question for tournament organizers and not necessarily for rules writers. In knockout/elimination formats a rematch would have to be played. In tournaments with a fixed schedule and some fixed number of rounds you'd have to assign half the points to each player (or some such). If rules have "win" "loss" and "no result" as possible outcomes then tournament organizers will have to account for each in how they choose to run the tournament.

Dear Robert, I thought that "a player must not immediately recapture a stone, which just captured one of his / her stones" would be clear enough. You cannot (re)capture a stone by a Pass. Usually the next opponent's stone has to be placed elsewhere on the board. If the opponent uses "Pass" as a Ko threat and you "answer" it by placing a stone of your own on the board, your opponent again will be free to capture the Ko (if you surprisingly enough have not connected the Ko or captured some further stones). If you answer "Pass" with "Pass", the situation has not changed and is as before.

Your answer "So the player can make up his / her mind if he / she wants 'neither victory nor defeat' or wants to play on and to stop playing Triple Ko, for example." is a fake and does not solve the problem for the decision-making player since the player CANNOT MEANINGFULLY make up his mind for a strategic decision since he does not have any comparison between scores and "neither victory nor defeat". He might as well draw a lot to decide whether to choose a result with a score or the result "neither victory nor defeat".
I think the player will be able to get a rough idea of the loss he will encounter when not continuing a Triple Ko sequence, for example.

That there is no coexistence of "recurrence of positions" and "score" is clear but this is not the problem. The problem occurs EARLIER: at some moments when to decide whether LATER there will be either "recurrence of positions" or a score.
I do not understand your problem, Robert. For example, a player must also make up his mind if he dares to invade a big Moyo. May be that there will be no "score", because the player resigns during "Play".

"If you encounter a repetitive sequence during "Evaluate", you will neither be able to have the status "2-eyed" for the chain nor can you prevent this chain from being captured.": I do not understand this yet. What exactly happens when a situation recurs during Evaluation?
Both status a) "2-eyed", and b) "stable" (as I have named it in the previous post for Harleqin) require that the evaluation sequence comes to an end. If there is no end, the status will be c) "removable".

If you mean "has 2 eyes or more" with "2-eyed", then you should give that thing a different name!
I think "2-eyed" is suitable when we talk about something that has at least 2 separated spaces inside, which are forbidden for the opponent. Every further separated space inside, which is forbidden for the opponent, does not change the status. It is the 2 (two !) that makes the difference.

"It's just a matter of practicability to realize that you can give the status '2-eyed' to every chain that can be connected to an already '2-eyed' one.": One must be careful with such statements. E.g., it is easy to construct a double threat example so that A is 2-eyed, either B or C is 2-eyed, but B and C are not both 2-eyed. Would you call such nice Japanese-style rules? I do not.
Each chain of stones is evaluated for its own. What you fear cannot happen.
It would be very interesting to see your "easy" construction".

"The status "2-eyed" is clear": Not it is not. Define "has 2 eyes or more"! I would define it as follows for a group and a considered set of empty intersections: "On the intersection union of the group and the set, we get a 2-eye-formation for each arbitrary, fixed filling by stones of the group's colour of a selection of all but 2 intersections of the set." However, I cannot know whether you mean this or something very different.
Nor can I.

"the equivalence of (single) 'stone' and 'chain'": Which "equivalence" do you mean?
"Chain" is something that in general you have to treat in whole like one single stone. If there an effect arises that is only possible with chains, you must keep it separated from rules for stones.

Your "local" is a wish. Imagine the game to enter Evaluation with a position we would perceive as middle game. You do not get such nice partition boundaries as you hope for.
I think it's a problem for every rule set to decide upon incomplete and unfinished games. As it will be for humans.

If you want to escape clear definitions of essential terms by means of meta-discussion ("theatre"), then you will face never-ending criticism about great unclarity of your rules. Mind you, you can easily apply earlier research and simply write "can force" where necessary or point to, say, "details as in J2003". But not defining even on the user-friendly level "eye" or "eyes (minimum)" makes your own rules attempt ridiculous. (Hint: I would define eye as follows: "An 'eye' is an empty intersection of a two-eye-formation." Then I would make your higher level terms fitting to that if I were you. Currently I cannot do that since I do not know your intended meaning.)
"Theatre of war" is "Kriegsschauplatz", Robert. You are free to define "eye" as you like, might it be recursive or not. An idea of mine you can find above.

Currently I lack time for trivial examples. Maybe you will have more luck in autumn.

"It seems important to me that all of the chain's primary points will become part of something "2-eyed". Its 2 eye points must be somewhere else, so I think that it will be possible to fill the chain's primary points completely.": Take John Tromp's example of a string consisting of 360 stones. You will see a problem easily: Life is possible, but not on all the 360 initial stone intersections.
Evaluation of status has nothing to do with "life". Should there be a string of 360 stones on the board at the end of "Play", its status is "removable". So there is no territory on the board. Difference of prisoners and Komi will decide the outcome.

"2-eye-formation": Definition see on various of my webpages or at Sensei's. It is not the meta-type of what you defined but you could reduce some of your life by forcing its transformation into a 2-eye-formation.
Again, Robert, if you evaluate the status of each chain individually, there is no need to have or to discuss something like "life".

You can find capturable-2 examples on my webpages.

I do not see as much interaction of Evaluation and Count in J1989 as you fear.
Look at my answer to Harleqin's question about example #16 and you probably will understand.

It is essential that you understand what I say about local maximal instead of a local minimal requirement for life. Maximal is what you want: Finally life on all initial intersections of a string. Minimal is what J1989, J2003 and Chris Dams's proof of equivalence between J2003-life and WAGC-life use: Finally life on at least one of all initial intersections of a string. Start reading here: http://home.snafu.de/jasiek/wagcmod.html
For the third time, Robert, if you evaluate the status of each chain individually, there is no need to have or to discuss something like "life".

I do not care which examples you use but yours should explain and distinguish your terms from each other!
May be I will find the time to create some examples for you. Not today.

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)

prokofiev, tournament system designers have the ADDITIONAL task how many tournament game points or whatever else to assign for a no result. Writers of rules of play have the task to allow each player to make meaningful strategic choices at all times in all positions.

Cassandra, "a player must not immediately recapture a stone, which just captured one of his stones" is wrong but "a player may not immediately play a single stone to recapture a single stone, which just captured one of his stones" is clear provided you do state a basic ko rule at all. ALA you do not, it would remain unclear which basic ko rule you mean.

Giving a player only a rough idea is insufficient. The rules must always give the player a clear idea of their intention.

Put it on the board: A position one play before either a double ko seki or a triple ko. Read http://home.snafu.de/jasiek/mistakes.html if you cannot construct such a position. Now the player having the turn is faced with the strategic choice between "The game will have the score result X." and "The game will have the result 'no result'.". To make this strategic choice, he must choose whichever is better for him. Since "no result" is uncomparable with any score X, the player CANNOT MEANINGFULLY make the decision at all. He cannot know which of the score result X and the result "no result" is better for him or whether both are equally good. It is impossible to know this for anybody since the rules fail to define it and since "no result" is not part of the ordered set of numbers, of which X is an element. In fact, "no result" is not even a number at all.

So the effect of a repeated situation during evaluation of some particular string is NOT that the no result rule would apply but the effect is that you need YET ANOTHER KO RULE, which says that the no result rule is not applied during Evalution and how to interpret recurrence / infinite sequences during Evaluation. Was it you wishing some sort of simple ko rules for the non-basic-ko cases?!

"2-eyed" is a terrible name when it is "more than 2 [eyes]". The common name is "n-eye-formation", where n>=2.

Of course, since you evaluate each string separately, you are absolutely ignorant about string set transitivity WRT to "2-eyed".

Here is an easy example of what we know as a double threat connection problem:

Click Here To Show Diagram Code

[go]$$B
$$ -------------------------
$$ |. . . . . . X . . . . . .|
$$ |X X X X X X X X X X X X X|
$$ |. O O O O O X O O O O O .|
$$ |X O . O . O X O . O . O X|
$$ |X O O O O X . X O O O O X|
$$ |X X X X X X O X X X X X X|
$$ |O O O O O O O O O O O O O|
$$ |. . . . . . O . . . . . .|
$$ -------------------------[/go]

Now don't tell me that you could not find that yourself.

If you cannot even know ("Nor can I.") your own intentions of what your rules shall achieve, then stop pretending to design rules!

This is a 3-eye-formation:

Click Here To Show Diagram Code

[go]$$B
$$ ---------
$$ |X X X X X|
$$ |. X . X .|
$$ |X X X X X|
$$ ---------[/go]

This is not a 3-eye-formation:

Click Here To Show Diagram Code

[go]$$B
$$ -----------
$$ |X X X X X X|
$$ |. X . X . .|
$$ |X X X X X X|
$$ -----------[/go]

Now you have to know for your own rules design whether the latter shall fall under "has 2 eyes or more". If yes, then your rules do not work (yet) since you have not defined "has 2 eyes or more" yet.

Your statement "'Chain' is something that in general you have to treat in whole like one single stone. If there an effect arises that is only possible with chains, you must keep it separated from rules for stones." does not explain which "equivalence" you might mean. In particular, one new stone under one of an initial string's stones is also "an effect that is only possible with chains".

Since it is a problem for every ruleset to decide upon incomplete and unfinished games, how do yours apply to them? Study some examples, please!

I do not see any definition of "eye" from you. Give it!

For a 360-stones-string, show me HOW YOUR RULES APPLY to yield the status "removable"! Start by removing, then place a stone on 3-3 and then continue with a (rough) explanation.

In this thread, I use "life" for one of two different purposes:

1) As an informal abbreviation of your "already has two eyes (minimum) or can get two eyes or can be captured, but the complete captured shape can be filled again with stones, which will become part of a 2-eyed group".

2) As a property of a string or strings under (especially) Japanese style rules.

Of course, you pretend to be ignorant about the relevance of earlier, general insight about (2) life for your rules... Let me assure you though that I am not that ignorant. Your rules do not invent a new game other than Go but they want to create something more or less familiar to Go players. As such they ought to recreate life, as it is previously known, as well as possible at least for its basic nature. However, your rules appear to fail to model even these basics. What you thus get is not, as you wanted, Japanese style rules any longer but non-Japanese style territory scoring rules. So if your rules continue like that, they are a failure in terms of their intentions. (Note that I do not care at all whether you recreate all J1989 Rules' diagrams.)

Cassandra, concerning your analysis of example #4 and when determining the status of White's upper side string, please add an explanation of HOW YOUR RULES APPLY! You only show one sequence with the comment "White's chain is eliminated." but you fail to explain what that means for how your rules might have to be applied here.

Concerning the upper black string, what is the relevance of "If she does nothing, we return to the sequence before." WRT to HOW YOUR RULES APPLY and in particular how should we combine consideration of two move-sequences for one particular question of rules application?

I do not understand "have become part of a Black 2-eyed group" because you have not defined "2-eyed group" clearly yet. (Besides, if what we see is a 2-eyed group, then why was the initial black group not also already 2-eyed?)

"White cannot do anything with Black's chain in the corner, so this one is '2-eyed'": Why should not being able to do anything be sufficient for an application of your rules?! They do not have a condition "the opponent cannot do anything", right?

Your conclusion "So we have White chains in status c)" too rash. Please explain why, according to your rules, they are not of the status "can be eleminated completely". You assess status string by string and your sequence for the upper white string has shown it being eliminated completely, has it not?!

To give the thread title a proper meaning, here are ideas for Japanese style rules:

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

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

What is the point of all this?

Of all what? Cassandra's rules sketch or the discussion on it?

Cassandra, "a player must not immediately recapture a stone, which just captured one of his stones" is wrong but "a player may not immediately play a single stone to recapture a single stone, which just captured one of his stones" is clear provided you do state a basic ko rule at all. ALA you do not, it would remain unclear which basic ko rule you mean.
Dear Robert, as I wrote several times before, I will not discuss wording any longer with you. I'm aware that you have invented several new terms with a very special meaning, but I thing that you have to accept if someone else uses another consistent terminology.

Giving a player only a rough idea is insufficient. The rules must always give the player a clear idea of their intention.
I'm not aware that your rule texts give a clear intention to the "normal" reader.

Put it on the board: A position one play before either a double ko seki or a triple ko. Read http://home.snafu.de/jasiek/mistakes.html if you cannot construct such a position. Now the player having the turn is faced with the strategic choice between "The game will have the score result X." and "The game will have the result 'no result'.". To make this strategic choice, he must choose whichever is better for him. Since "no result" is uncomparable with any score X, the player CANNOT MEANINGFULLY make the decision at all. He cannot know which of the score result X and the result "no result" is better for him or whether both are equally good. It is impossible to know this for anybody since the rules fail to define it and since "no result" is not part of the ordered set of numbers, of which X is an element. In fact, "no result" is not even a number at all.
I will not discuss this topic any longer, Robert. It seems to me that you cannot accept the clear distinction between "Play", "Evaluate", and "Count". Any further discussion is meaningless, if you continue to mix these stages up.

So the effect of a repeated situation during evaluation of some particular string is NOT that the no result rule would apply but the effect is that you need YET ANOTHER KO RULE, which says that the no result rule is not applied during Evalution and how to interpret recurrence / infinite sequences during Evaluation. Was it you wishing some sort of simple ko rules for the non-basic-ko cases?!
No, Robert. In Japanese, it is very important to listen to what has not been said. If the termination conditions of case A and case B cannot be reached, case C will apply without saying.

"2-eyed" is a terrible name when it is "more than 2 [eyes]". The common name is "n-eye-formation", where n>=2.
I know your problem with this kind of wording. But as I wrote before, all eyes beyond eye number 2 do not have to do anything with the status of a chain.

Of course, since you evaluate each string separately, you are absolutely ignorant about string set transitivity WRT to "2-eyed".
The explaining examples created especially for you have not been designed yet.

Here is an easy example of what we know as a double threat connection problem:
Your example will be processed as follows.

Click Here To Show Diagram Code

[go]$$B
$$ -------------------------
$$ |. . . . . . X . . . . . .|
$$ |X X X X X X X X X X X X X|
$$ |. O O O O O X O O O O O 3|
$$ |X O . O . O X O . O . O B|
$$ |X O O O O X 2 B O O O O B|
$$ |X X X X X X O B B B B B B|
$$ |O O O O O O O O O O O O O|
$$ |. . . . . . O . . . . . .|
$$ -------------------------[/go]

I suppose it will be sufficient to concentrate on Black's right hand chain, marked here. This chain is not "2-eyed" yet, so White probably has a chance. I suppose you have White's cutting move at 2 in mind. Black will connect at 3; now the string is "2-eyed".

Now don't tell me that you could not find that yourself.
As should have become clear, Robert, your example provides no problem. So there is no need to think about it as a very special case.

If you cannot even know ("Nor can I.") your own intentions of what your rules shall achieve, then stop pretending to design rules!
See above.

This is a 3-eye-formation:

This is not a 3-eye-formation:
I know your thinking about this. It's not mine. I think that you refrain from possibilities to simplify the application of what you seem to have in mind.

Now you have to know for your own rules design whether the latter shall fall under "has 2 eyes or more". If yes, then your rules do not work (yet) since you have not defined "has 2 eyes or more" yet.
The examples for you are ongoing, Robert.

Your statement "'Chain' is something that in general you have to treat in whole like one single stone. If there an effect arises that is only possible with chains, you must keep it separated from rules for stones." does not explain which "equivalence" you might mean. In particular, one new stone under one of an initial string's stones is also "an effect that is only possible with chains".
Important is not the process, but what can be seen after the dust has settled.

Since it is a problem for every ruleset to decide upon incomplete and unfinished games, how do yours apply to them? Study some examples, please!
This is a problem of tournament rules, Robert, so I will not discuss it here.

I do not see any definition of "eye" from you. Give it!
I think I have given one. It seems to be of a kind that you do not have on your radar screen.

For a 360-stones-string, show me HOW YOUR RULES APPLY to yield the status "removable"! Start by removing, then place a stone on 3-3 and then continue with a (rough) explanation.
I suppose that you know that it is impossible to prevent at least one living group of each color on a 19x19 board. By the way, it is irrelevant for the outcome of your example game, whether the status of the 360-stones chain is "stable" or "removable". There is nothing "alive" on the board, which could generate territory or force the chain to be taken off the board.

In this thread, I use "life" for one of two different purposes:

1) As an informal abbreviation of your "already has two eyes (minimum) or can get two eyes or can be captured, but the complete captured shape can be filled again with stones, which will become part of a 2-eyed group".

2) As a property of a string or strings under (especially) Japanese style rules.
"Life" is not necessary, Robert. Generally speaking, you can choose of two directions.

• Start with defining status of chains.
• Start with defining "life" as a property of a chain.

The second one will prove much more complicated. As you will have learned over years, I suppose.

Of course, you pretend to be ignorant about the relevance of earlier, general insight about (2) life for your rules... Let me assure you though that I am not that ignorant. Your rules do not invent a new game other than Go but they want to create something more or less familiar to Go players. As such they ought to recreate life, as it is previously known, as well as possible at least for its basic nature. However, your rules appear to fail to model even these basics. What you thus get is not, as you wanted, Japanese style rules any longer but non-Japanese style territory scoring rules. So if your rules continue like that, they are a failure in terms of their intentions. (Note that I do not care at all whether you recreate all J1989 Rules' diagrams.)
I'm looking forward to receiving an example, which would prove an inconsistency in my ideas.

As I wrote in the introduction, it can be no problem, if some of the 1989 Nihon Kiin rules examples will be treated with another result. Because there is the general meaning, that they are inconsistent.

The scoring is Japanese-style. Written in other words, territory is everything inside chains of only one colour, which have not got the status "in Seki".

_________________
The really most difficult Go problem ever: https://igohatsuyoron120.de/index.htm
Igo Hatsuyōron #120 (really solved by KataGo)