Life In 19x19

As Martin Stiassny has pointed out elsewhere
http://www.dgob.de/yabbse/index.php?topic=5181.0
"Asian Games" were in 2010 and will again be in 2014 (then probably without go). "Asian Indoor [...] Games", as they occurred in Korea in July 2013, are a different series of events. IOW, this difference must be kept in mind when you see my abbreviating title "Asian Games 2013 Rules". Abbreviations are dangerous;)

asura, as I said, there are examples why ko-pass-for-one-particular-ko does not work as intended. That's why generic-ko-pass is needed. (No time to look up the examples ATM.) EDIT: http://home.snafu.de/jasiek/j1989c.html Example 0001 is such an example, although not the simplest.

Instead of calling your cycles "fighting/disturbing", I suggest "local"/"non-local" or something similar.

J1989 allow infinite cycles regardless of the ko-pass rule! See official example II.24+25: the ko stones of a double ko seki are declared "dead". This makes sense only if - even after an infinite cycle - a new "permanent-stone" is not "enabled". If only one cycle were used and stopped, then there would be new permanent-stones. - My infinite cycles with ko-passes simply combine the J1989 infinite cycles without ko-passes and the availability of ko-passes. Maybe the rules authors did not think of this possibility at all, but it is the consequence of their machwerk.

Of course, J1989 style rules can be formulated using a Cycle End Rule (on occurrence of a cycle, a hypothetical sequence ends). This can lead to different behaviours in arcane examples and different statuses without scoring impact (such as alive ko stones in double-ko sekis).

Japanese style rules do NOT NEED ANY KO-PASS RULE. Without, the following two examples have a different behaviour:

1) direct teire ko + elsewhere double disturbing death

2) triple ko with one external ko

If the J1989 authors could have accepted (2) (they were not aware of the more "frequent" (1), it seems, which Winfried Borchardt pointed out), they could have used the same basic ko rule used during the regular game alternation. (The Koreans fell into the same trap of wishing to have a particular outcome for (2). This causes problems in the Korean 1992 Rules and the Asian Games 2013 Rules.)

RobertJasiek wrote:Japanese style rules do NOT NEED ANY KO-PASS RULE.

Yes, that's true.

It seems to me that it is possible to reach the "favoured" results of most the official commentary's examples without a rule of this type, and even the rules-intrinsic results (not the "favoured" ones) for the examples 16 to 18 (if we forget about the sudden, but unexplained, "collapse of a Seki").

Even an explicite definition of a "locality" is not mandatory, despite the need for some "implicite" usage, but only for the points that are set free by the capture of a string (and of which at least one has to become occupied permanently to achieve some special status of this string).

I assume that it might be a matter of taste (but not of effort, or complexity), how to first identify different status of strings, for the later combination for the assignment of "alive" (may be and "in Seki") and "dead".

In one way or the other, one has to identify

(1) Strings that are the "owner" of at least two forbidden points,
(2) Strings that are not of type (1), but cannot be taken off the board by actual play [[usually related to "Seki"]], and
(3) Strings that are neither of type (1), nor of type (2) [["pending" strings]].

Without, the following two examples have a different behaviour:
1) direct teire ko + elsewhere double disturbing death
2) triple ko with one external ko

Do you refer to some of the official commentary's examples ?

It is not necessary to define seki in Japanese style rules. See, e.g., the Intermediate Step Rules in
http://home.snafu.de/jasiek/j1989c.html
or create a hybrid of the Simplified Japanese / Korean Rules. (Actually, I have created such, but not published it. It is straightforward: just throw in a few "independently alive" phrases in the dead stone removals + territory definitions.)

I do not refer to official examples. For (1), there is none. For (2), you can, of course, look at II.8.

RobertJasiek wrote:It is straightforward: just throw in a few "independently alive" phrases in the dead stone removals + territory definitions.

This is surely one of the "other" ways )


It is impossible (with Japanese style rules) to reduce something like "alive with territory" / "alive without territory" / "not alive" to only two status !

RobertJasiek wrote: asura, as I said, there are examples why ko-pass-for-one-particular-ko does not work as intended. That's why generic-ko-pass is needed. (No time to look up the examples ATM.) EDIT: http://home.snafu.de/jasiek/j1989c.html Example 0001 is such an example, although not the simplest.

It seems we were not meaning the same with "undesired ko-pass cycles".
In my understanding the problem in your example only comes up because w can endless delay the capture of the single w stone on the right by alternately making a ko-capture and a ko-pass.
Instead of lifting all ko-bans with a ko-pass I'm more with Bill's solution and let the ko-pass only lift a special ko-ban and stop the cycle otherwise.

Instead of calling your cycles "fighting/disturbing", I suggest "local"/"non-local" or something similar.

Less confusing, but also less mysterious

J1989 allow infinite cycles regardless of the ko-pass rule!

It's nessecary to prevent or stop the cycle. You have some freedom how to interpretate the result, but treating it the same as ending with two passes wouldn't work correctly.
For making a decision you should consider the comment to the long life (example 6).
For me it's difficult to interpretate all the comments absolutely consistent and it seems to me that they wanted ensure the correct result only but gave simply a reason that comes first to mind to justify it. But in general you cannot transfere the reasoning on other cases.

asura wrote:For me it's difficult to interpretate all the comments absolutely consistent and it seems to me that they wanted ensure the correct result only but gave simply a reason that comes first to mind to justify it. But in general you cannot transfere the reasoning on other cases.

"they wanted ensure the correct result only"

In this case, it would have been much better if they had provided a selection of "special positions", together with the "officially" given status. Nothing more.
By the way: "intended" would be a better term for "correct".

Seen overall, the examples are NOT consistent (the Japanese tried to develop a consistent way of derivation, but in vain).

Especially (in my opinion) the examples with the sudden "collapse of the Seki" are inconsistent, because these can only be explained using a mismatch of "status confirmation", and "scoring".

Confirming strings to be "dead" is part of "status confirmation", but taking these stones off the board without further play, is part of "scoring". However, there is nothing said in the Rules that "scoring" might be interrupted, to do some "status confirmation" again.

Therefore, it is not very surprisingly that the consistent application of other rule-sets gives different results for some examples than the offical Japanese Rules' text.

Cassandra wrote: Especially (in my opinion) the examples with the sudden "collapse of the Seki" are inconsistent, because these can only be explained using a mismatch of "status confirmation", and "scoring".

For me the "collapse of the Seki" is not the worst. If you add a rule that ensures you cannot endless delay the capture (by a cycle with ko-capture and ko-pass) you can prove that the "seki" is capturable in it's own analyse.
For me the most complicated thing was to assimilate all the comments about cycles and there interpretation, especially because some comments indirectly add comments or implications to that topic.

Therefore, it is not very surprisingly that the consistent application of other rule-sets gives different results for some examples than the offical Japanese Rules' text.

Of corse other rules give other results, else they would be the same (equivalent) rules

However for generating a model for j1989 it is required at least:
- the rules are effective decideable in all situations
- the rules give the same result in all examples from the commentary

asura wrote:For me the "collapse of the Seki" is not the worst. If you add a rule that ensures you cannot endless delay the capture (by a cycle with ko-capture and ko-pass) you can prove that the "seki" is capturable in it's own analyse.

The positions that include a merger of "Double-Ko" and "anything else" are treated (according to the intended result) as if their merged components were independent from each other (, but which they are not). Just saying: "First, decide on the 'anything else' alone. Re-arrange the position according to this result. Then, decide on the 'Double-Ko'."

By the way: In my opinion, it does not make any sense to restrict "cycles" during "status confirmation".

For me the most complicated thing was to assimilate all the comments about cycles and there interpretation, especially because some comments indirectly add comments or implications to that topic.

"Cycles" caused no problems for me. So, our viewpoint may be different.

Cassandra wrote:The positions that include a merger of "Double-Ko" and "anything else" are treated (according to the intended result) as if their merged components were independent from each other (, but which they are not).

It depends on how you treat them. I agree that the reasoning of the comment is a bit strange or confusing.
In my view the double-ko and the "anything else" ARE independent from eachother. (Actually all strings are independent from eachother, so there are more than two things.)
To show that the "anything else" is dead I can simply prove it by capturing it and the double-ko doesn't matter IF THERE IS A RULE THAT COVERS THESE CYCLES.
To show that the double-ko is dead requires first to capture the "anything else" (which itself has no meaning for the double-ko in this analyse) and after that you can capture the double-ko. As said above there may be more than only two playouts if there are more than two strings.

"Cycles" caused no problems for me. So, our viewpoint may be different.

Maybe the viewpoint is different or maybe only the starting point. To make rules that works correctly in some situations is quite easy, but when you add more rules to cover more situations it becomes more and more difficult not to change the things that have worked before. There were some situation where I had the choice to make one thing easy at the cost of making another thing complicated or vice versa.

Have you thought about the comments (and implications) to the long-life in the official commentary?
I slightly recall that I found a deeper problem with this but I never was able to create a position that shows it - maybe I just saw a ghost - and atm I cannot remember my thinking about.

asura wrote:Have you thought about the comments (and implications) to the long-life in the official commentary?
I slightly recall that I found a deeper problem with this but I never was able to create a position that shows it - maybe I just saw a ghost - and atm I cannot remember my thinking about.

To be honest, I have not thought about it really hard.

However, in my opinion, the comment (referring to "repetition" >>> "no result") is somewhat incomplete.

Within the given context, "repetition" must refer to the complete "state" of the game. This "state" does not only include the visible board-position, but also the captured stones. To declare a "repetition", the difference of captured stones must be the same with both visible board-positions.

I suppose that such a "repetition" cannot happen within less than four moves. But I have no idea whether it might be preferable (for "guarantee" reasons) to have something like "... before at least four moves ..." explicitely written.

Cassandra wrote:Within the given context, "repetition" must refer to the complete "state" of the game. This "state" does not only include the visible board-position, but also the captured stones. To declare a "repetition", the difference of captured stones must be the same with both visible board-positions.

Do you mean something like sending-2-returning-1 ?
If so then it wouldn't prevent the opponent to play a capture sequence elsewhere. As far I can see the only problem this can cause is that the hypo-sequence doesn't end with two passes.

Cassandra wrote:To declare a "repetition", the difference of captured stones must be the same with both visible board-positions.

You need not count removed stones, but it suffices to count passes and detect if a cycle is positional / situational:

http://senseis.xmp.net/?CycleLaw

According to Martin Stiassny in Deutsche Go-Zeitung 5/2013, the 4th Asian Indoor & Martial Arts Games Incheon 2013 were the first such event with Go. Hence, for earlier such events, one need not clarify which Go rules they might have used: none.

RobertJasiek wrote:[In his commentary]
Matti Siivola has found an extremely rare seki shape with a group without any neutral point. See his webpage for the example.

I've looked at his webpage but couldn't find this example. However I'm not sure if I understand correctly what you mean with "seki shape without neutral points" so maybe I just couldn't identify that shape.
Can you show (a link to) that example or explain the principle, please?

asura wrote:

RobertJasiek wrote:[In his commentary]
Matti Siivola has found an extremely rare seki shape with a group without any neutral point. See his webpage for the example.

I've looked at his webpage but couldn't find this example. However I'm not sure if I understand correctly what you mean with "seki shape without neutral points" so maybe I just couldn't identify that shape.
Can you show (a link to) that example or explain the principle, please?

See http://www.suomigo.net/wiki/KimuranttiAsema.