KGS, bent four in japanese rules, need help.

[Bill Spight]

Here is a demonstration of hypothetical play under the Japanese 1989 rules.



Next question. What about the Korean rules?

[RobertJasiek]

Herman, verbal Japanese rules are ambiguous with respect to whether they have a WAGC-style "bent-4-in-the-corner is dead" precedental rule and whether bent-4-in-the-corner refers to shapes with bent-4-in-the-corner adjacent to seki shapes. For German-Japanese Rules, IMO there is a "bent-4-in-the-corner is dead" precedental rule, but it is unclear whether bent-4-in-the-corner refers to shapes with bent-4-in-the-corner adjacent to seki shapes. For other countries' or regions' verbal Japanese rules, there has been too little consensus-finding to know even whether there is a WAGC-style "bent-4-in-the-corner is dead" precedental rule in them.

I am not finding complications where there are none, but the reality is that there is too little consensus!

You are right about the J1989 pass-for-(the-specific-)ko, which, in hypothetical strategy, overrides other considerations about locality.

[daal]

KGS supports several different rule sets, named after either the organization that uses them or the country where they are common. In some cases the rule sets implemented are not quite the official rules for the organization that they are named after; for example, several different rule sets are allowed in an AGA tournament. The following rule sets are used on KGS:

Image of ruleset selection menu


Japanese
In a Japanese game, the score is the sum of captures and territory. Seki does not count as territory. If the board enters a long repeating cycle that neither player wants to break, and which is of advantage to neither player, the game is scored as no-result. Note: The Japanese system of solving disputes that happen at the end of the game is not available. If players cannot agree whether a group is alive or dead, they must accept the judgement of a third neutral player.

The problem with the KGS implementation of the Japanese rules, particularly with regard to the bent-four-in-the-corner-even-when-there-is-a-seki-on-the-board situation, is twofold. First, the system can't automatically implement the Japanese rules- whatever they may be, and second, the KGS rule is not stated explicitly on the page for both players and mediators to refer to. Since there seems to be so much disagreement on what the Japanese rules are, it would make sense for KGS to adopt one policy and write it into their rules.

[HermanHiddema]

@Robert: I see no reason to assume that the presence of seki would have any impact. Certainly neither the J1989 nor the WAGC rules texts give any reason to do so. That assumption is what I am referring to when I use the phrase "finding complications".

[quantumf]

Sorry, Bill, I don't understand the comment "*** Black cannot take the ko back, because the only threat is a pass for that ko."

Black has just played a ko threat...which white answered...so what am I missing?

[HermanHiddema]

Sorry, Bill, I don't understand the comment "*** Black cannot take the ko back, because the only threat is a pass for that ko."

Black has just played a ko threat...which white answered...so what am I missing?

Japanese 1989 rules (current Nihon Kiin official rules) say:

Article 7. Life and death
1. Stones are said to be "alive" if they cannot be captured by the opponent, or if capturing them would enable a new stone to be played that the opponent could not capture. Stones which are not alive are said to be "dead."
2. In the confirmation of life and death after the game stops in Article 9, recapturing in the same ko is prohibited. A player whose stone has been captured in a ko may, however, capture in that ko again after passing once for that particular ko capture.

So after game end, if there is no agreement on the status of a group, there is a procedure whereby life or death of groups is determined. During this procedure, the only valid ko threat for a ko is a pass for that specific ko. (per section 7.2 quoted above)

[shapenaji]

Since there seems to be so much disagreement on what the Japanese rules are, it would make sense for KGS to adopt one policy and write it into their rules.

I'm not even sure if there's much disagreement, it's just awkward to have a ruleset with these non-intuitive consequences.

Once I got to the point in my studies where I was used to leveraging different areas of the board against each other, I developed a profound dislike for the Japanese rules because of situations like this.

They're self-consistent, but just feel so kludgy. There's something lovely about a bent 4 being balanced by an appropriately sized unremovable ko threat.

[Xylol]

So if i understood correctly:

J1989: The game ends and there is no agreement on the status of a group. Now the life or death of groups is determinded, whilst doing so only passing is considered as an allowed ko-threat.

Therefore the bent four is dead because I don't have to fear the unremovable ko treat of the seki shape.

It would be unfair to thank only one person. I thank every single one in this thread for his time and contribution towards finding a solution.

But still: if i have to proof the group is dead i will lose 3 points (see my first answer), how's this handled?

[jts]

So if i understood correctly:

J1989: The game ends and there is no agreement on the status of a group. Now the life or death of groups is determinded, whilst doing so only passing is considered as an allowed ko-threat.

Therefore the bent four is dead because I don't have to fear the unremovable ko treat of the seki shape.

It would be unfair to thank only one person. I thank every single one in this thread for his time and contribution towards finding a solution.

But still: if i have to proof the group is dead i will lose 3 points (see my first answer), how's this handled?

No. All the stones put down in hypothetical play are removed from the board once the group in question has been proved to be dead or alive. (That's why it's called hypothetical.)

[quantumf]

But still: if i have to proof the group is dead i will lose 3 points (see my first answer), how's this handled?

The life and death determination (if required) is just an evaluation step. Once the life or death has been determined, the board returns to the state after both players passed. In modern times, you can take a photo with a camera. Historically it had to be done on an adjacent board.

[Bill Spight]

Sorry, Bill, I don't understand the comment "*** Black cannot take the ko back, because the only threat is a pass for that ko."

Black has just played a ko threat...which white answered...so what am I missing?

The Japanese 1989 rules have hypothetical play after the end of regular play to determine life and death if the players do not agree. Under hypothetical play the only thing that lifts a ban on taking a ko back is a pass in which the player designates the ko for which he is lifting the ban. In hypothetical play Black would not actually sacrifice the seki to make a ko threat. That was just to show the futility of doing so.

[Bill Spight]

Since there seems to be so much disagreement on what the Japanese rules are, it would make sense for KGS to adopt one policy and write it into their rules.

I'm not even sure if there's much disagreement, it's just awkward to have a ruleset with these non-intuitive consequences.

Once I got to the point in my studies where I was used to leveraging different areas of the board against each other, I developed a profound dislike for the Japanese rules because of situations like this.

They're self-consistent, but just feel so kludgy. There's something lovely about a bent 4 being balanced by an appropriately sized unremovable ko threat.

The Japanese 1949 rules were criticized for having a number of ad hoc rules. The Japanese 1989 rules provided a rationale for nearly all of the ad hoc situations. (I cannot say that the '89 rules are logical, because I do not know of any computer program that applies them correctly. ) However, they introduced new anomalies, and the rules for hypothetical play are significantly different (non-intuitive) from the rules of regular play. I prefer the 1949 rules, myself.

[palapiku]

This weirdness about ignoring non-removable threats is completely separate from the concept of territory scoring, right? You could have a genuinely territory-based ruleset (not like AGA) that would treat bent four as seki when non-removable threats are present. All you need to do is get rid of that extra rule.

[Bill Spight]

This weirdness about ignoring non-removable threats is completely separate from the concept of territory scoring, right? You could have a genuinely territory-based ruleset (not like AGA) that would treat bent four as seki when non-removable threats are present. All you need to do is get rid of that extra rule.

There are a number of territory scoring rules that do not involve hypothetical play to resolve life and death issues. Those that use an actual encore include Ikeda's rules, Lasker-Maas rules, and my rules. In all of these three the seki could be an unremovable ko threat. (Except for my specifically Japanese rules. )

BTW, some people believe that territory scoring is a relatively new aberration, that a form of area scoring came first. The evidence about which came first is unclear. The earliest known description of the rules appears to be for a form of area scoring. However, the earliest know game records that are scored appear to be for a form of territory scoring.

[RobertJasiek]

Herman, Japanese rules texts with precedental rulings or comments gave very specific examples. Two very similar shapes can have very different rulings. Also therefore, I do not overinterpret a bent-4 precedent. When there was an attempt to generalise a precedent, such an attempt was not welcome.

The Japanese 1989 Rules' pass-for-a-specific-ko rule cannot be applied literally because it creates contradictions to intentions. The Japanese 2003 Rules' generic-pass-for-ko rule should be applied for J1989 interpretation because it agrees to intentions whereever the J1989 rules writers were careful enough with their examples. For one bent-4, both ko-pass variants have the same behaviour.

http://home.snafu.de/jasiek/j1989c.html
http://home.snafu.de/jasiek/j2003.html

Xylol, in hypothetical analysis, a "pass" does not releave a [basic] ko ban. It requires the different (and artificial) move type "ko-pass" to releave a ko ban.

Proof-play is only imagined / hypothetical / executed with independent playing material / temporarily executed and then position and prisoners are restored. Therefore one does not lose points by playing to fill territory in proof-play.

Bill, it is not clear whether ko-pass is the only means of lifting a ko ban. It is (in theory) possible to have a ko, destroy it and resurrect it in the same local shape much later. Is it still the same ko? Undefined.

palapiku, territory scoring does not need special ko rules:

http://home.snafu.de/jasiek/sj.html