Rational Japanese territory via Maas-style encore?

[hzamir]

Honest question:

I'm not specifically interested in the myriad of variant rules.

I would like to understand just how different the outcome would be between:

• Japanese rules
• Replacing life-death precedents / hypothetical play with Maas-style encore played with prisoners

(IOW this is not a question about Lasker-Maas rules generally, just the Maas encore, and applying Japanese territory scoring rules at the end).

I'm very late in paying any attention to Maas, but have read since every single reference to his encore I could find, and am a bit surprised it doesn't get more attention as a serious territory rules procedure independent of its appearance in Lasker-Maas rules.

[jann]

What do you ask about specifically? Compared to normal play, playing from prisoners gives an extra point for board plays (the removed prisoner), which is like taking an extra penalty for pass plays (as with pass stones). Both effectively switch to area for later phases, and pass stones (with equal moves) are the more common way for this.

Some differences with any areafying encore:
- dame play necessary in main phase
- one sided dame and delayed mannenko resolution gives points
- the normal ko rule cannot be used in encore, so all superko anomalies
- things like torazu3 or bent4+seki play out to different results
- unique to LM rules: they use ko-unsafe encore switch, so simple endgame kos can also give 1 extra point

[hzamir]

What do you ask about specifically?

Japanese rules (say 1989) require significant expertise to apply when needed the most, which may not be available. So I was wondering just how close to result of ideally/expertly applied rules, would the Maas encore be. Which specific things would score differently (which, thanks you attempted to answer).

Playing from prisoners gives an extra point for board plays (the removed prisoner), which is like taking an extra penalty for pass plays (as with pass stones). Both effectively switch to area for later phases, and pass stones (with equal moves) are the more common way for this.

Yes, meaning to ask about that also, but didn't--since I was still working it out for myself--so would like to append as a subsidiary question:
How Japanese rules compare to Ikeda territory rules II?, Those appear even more elegant, yet did they ever merit serious consideration for adoption anywhere?

I wonder how AGA went more for territory style counting, yet ultimately came as close to Chinese rules as possible, when I believe their constituents were far more steeped in Japanese practice, and Ikeda Territory II rules could have suited that goal. Is it that just AGA rules would land much closer to a major established ruleset (Chinese) with their choice, than the choice of Ikeda's Territory II rules would be in relation to Japanese rules?

Some differences with any areafying encore:
- dame play necessary in main phase
- one sided dame and delayed mannenko resolution gives points
- the normal ko rule cannot be used in encore, so all superko anomalies
- things like torazu3 or bent4+seki play out to different results
- unique to LM rules: they use ko-unsafe encore switch, so simple endgame kos can also give 1 extra point

I get the first two points, on dame.

Can you elaborate on ko-unsafe encore switch? Is this unique to LM encore?

[jann]

Can you elaborate on ko-unsafe encore switch? Is this unique to LM encore?

If encore switch is allowed during an endgame ko fight (like on two pass stops), a player with excess threats can delay winning the ko into the encore (where it worths more). This creates a rather frequent pointwise difference to territory scoring. In contrast, a ko-safe stopping rule (like in Spight rules) ensures that the territory phase truly completes before switching to pass stones.

[Harleqin]

Can you elaborate on ko-unsafe encore switch? Is this unique to LM encore?

If encore switch is allowed during an endgame ko fight (like on two pass stops), a player with excess threats can delay winning the ko into the encore (where it worths more). This creates a rather frequent pointwise difference to territory scoring. In contrast, a ko-safe stopping rule (like in Spight rules) ensures that the territory phase truly completes before switching to pass stones.

That wording shows where you stand on the debate

However, a single »endgame half point ko« can be seen as just a special case of »do I have to protect my territory (i. e. play a teire) against an attack that results in a ko that the opponent cannot win (get any compensation for) anyway?«. See https://senseis.xmp.net/?Ruledisputesinvolvinggoseigen

If you do area scoring, delaying the half point ko until after all dame are played is natural, and then winning that open ko amounts to the same as if the open point was a point of territory. So, if keeping the effective scoring as similar as possible is a goal, this would be an argument for an open simple ko that the opponent cannot force closed to be a point of territory.

If someone thinks that stones in atari are somehow not really alive, I'd remind them of the existence of hanezeki. If they think that »there are no points in seki« is an axiom (not an observation), it shows they are taking the Nihon Kiin rules as canonical anyway.

[hzamir]

Can you elaborate on ko-unsafe encore switch? Is this unique to LM encore?

If encore switch is allowed during an endgame ko fight (like on two pass stops), a player with excess threats can delay winning the ko into the encore (where it worths more). This creates a rather frequent pointwise difference to territory scoring. In contrast, a ko-safe stopping rule (like in Spight rules) ensures that the territory phase truly completes before switching to pass stones.

That wording shows where you stand on the debate

However, a single »endgame half point ko« can be seen as just a special case of »do I have to protect my territory (i. e. play a teire) against an attack that results in a ko that the opponent cannot win (get any compensation for) anyway?«. See https://senseis.xmp.net/?Ruledisputesinvolvinggoseigen

If you do area scoring, delaying the half point ko until after all dame are played is natural, and then winning that open ko amounts to the same as if the open point was a point of territory. So, if keeping the effective scoring as similar as possible is a goal, this would be an argument for an open simple ko that the opponent cannot force closed to be a point of territory.

If someone thinks that stones in atari are somehow not really alive, I'd remind them of the existence of hanezeki. If they think that »there are no points in seki« is an axiom (not an observation), it shows they are taking the Nihon Kiin rules as canonical anyway.

I'm unsure where you are going with this. The context of the post is whether an actual encore can preserve results very close to traditional Japanese rules, not to debate how seki points should or should not be counted, or whether seki counts as life. It's just a different system. I have no problem with a territory system that defines territory in terms of "immortal" stones (stones that cannot be captured even with no opposition, per Ikeda Territory II). So what if it doesn't agree with Chinese practice, it is still perfectly rational.

[Harleqin]

I'm unsure where you are going with this.

I was responding to the wording in the post I quoted, especially »truly completed«. I think this should be worded as a difference to Nihon Kiin rules, not as a flaw.

[jann]

I don't think this is specific to Nihon Kiin. The example was Spight territory rules, but also in a broader sense:

argument for an open simple ko that the opponent cannot force closed to be a point of territory.

He CAN force it closed, under territory scoring. At most he needs a few passes and (under some rulesets) resumptions, but then the player will need to connect the ko (or lose it). The difference only arise if - in the middle of a ko fight trading real threats for passes - a sudden rule switch (to area) interrupts ongoing territory play.

Btw, the mentioned "no points in seki" isn't specific to Nihon Kiin either, it is simply a stability necessity under pure/LD (ie. without pass stone encore) territory scoring.

[CDavis7M]

Japanese rules (say 1989) require significant expertise to apply when needed the most, which may not be available.

Is that true? I do not think it is publicly available but my understanding is that the Nihon Kiin maintains a list of rulings and precedents. The situation can be recognized and easily applied. Some expertise in Go is needed of course but probably less expertise is needed to apply Japanese precedent than to figure out what Mass-style encore is. It is fair for this claim to be made with as much basis as the rest.

[RobertJasiek]

Precedents were in the J 1949 and WAGC 1979/1980 Rules. For J 1989, see 1989 commentary and 2003 interpretation

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

And before you start arguing precedents might have been less difficult, see

http://home.snafu.de/~jasiek/wagcflaw.html
http://home.snafu.de/~jasiek/wagcmod.html
http://home.snafu.de/~jasiek/wagcinf.html#precedents

Hence, to answer your question: yes, true.

[Harleqin]

argument for an open simple ko that the opponent cannot force closed to be a point of territory.

He CAN force it closed, under territory scoring. At most he needs a few passes and [...]

You assume that a pass lifts a ko ban.

[jann]

You assume that a pass lifts a ko ban.

Sure, I think this is essential under territory scoring (w/o encore). The two scoring methods differ greatly here (that's why existing rulesets only combine each with certain ko/superko rules).

Under area scoring, even if the rules don't have ban lifting passes, the players can usually substitue with plays inside territory (= 0 pt ban lifting moves). So the lack of passes could only cause problems on very crowded and small boards (not in practice).

Under territory scoring, however, only dame could substitute, which are not always available (or enough). The players often have no means to change the board without losing points. So if the rules woudn't have fully functional passes, anomalies related to 1-eye-flaw (and differences to area scoring) could become untolerably common - even on 19*19 boards. Mannenko resolution could already be a problem, let alone things like this (no komi no prisoners, W normally wins regardless of scoring method):

$$B B to move
$$ +---------------+
$$ | O . O . . O . |
$$ | X O O O . O . |
$$ | . X X O O O @ |
$$ | X X X O X X X |
$$ | . . X X X . . |
$$ +---------------+
$$

Click Here To Show Diagram Code

[go]$$B B to move
$$ +---------------+
$$ | O . O . . O . |
$$ | X O O O . O . |
$$ | . X X O O O @ |
$$ | X X X O X X X |
$$ | . . X X X . . |
$$ +---------------+
$$[/go]

[hzamir]

You assume that a pass lifts a ko ban.

Sure, I think this is essential under territory scoring (w/o encore). The two scoring methods differ greatly here (that's why existing rulesets only combine each with certain ko/superko rules).

Under area scoring, even if the rules don't have ban lifting passes, the players can usually substitue with plays inside territory (= 0 pt ban lifting moves). So the lack of passes could only cause problems on very crowded and small boards (not in practice).

Under territory scoring, however, only dame could substitute, which are not always available (or enough). The players often have no means to change the board without losing points. So if the rules woudn't have fully functional passes, anomalies related to 1-eye-flaw (and differences to area scoring) could become untolerably common - even on 19*19 boards. Mannenko resolution could already be a problem, let alone things like this (no komi no prisoners, W normally wins regardless of scoring method):

$$B B to move
$$ +---------------+
$$ | O . O . . O . |
$$ | X O O O . O . |
$$ | . X X O O O @ |
$$ | X X X O X X X |
$$ | . . X X X . . |
$$ +---------------+
$$

Click Here To Show Diagram Code

[go]$$B B to move
$$ +---------------+
$$ | O . O . . O . |
$$ | X O O O . O . |
$$ | . X X O O O @ |
$$ | X X X O X X X |
$$ | . . X X X . . |
$$ +---------------+
$$[/go]

I'm confused from my reading about in which (mainstream) rule sets passes can actually lift ko bans. The discussion on https://senseis.xmp.net/?PassAsKoThreat seems incomplete to me, establishing only--it seems to me--that it is a non question for superkos, and that Ing rules say they lift the ko ban, and it is true during a hypothetical okay phase of J1989 rules. Is that it?

Also, my reading of the above under position in Ikeda Territory II rules is that black gains a tie by capturing the ko, by now having a prisoner the score is 0, and not more for black, since empty points not inside immortal stones aren't counted.

[jann]

I'm confused from my reading about in which (mainstream) rule sets passes can actually lift ko bans.

It's hard to be sure since major Asian rulesets tend to be sparsely worded, and their actual practice not necessarily agrees a too literal reading of their text (over common sense).

As you wrote we do know that passes lifting bans is a central idea of Japanese hypothetical play. This is a clear principle, and would be very surprising to reverse for normal play. I'm not sure about Chinese practice, but the original authors of sensei's 1-eye-flaw page claim that Chinese rules do allow recapture there. Without reference this is hard to verify, but it does agree with the few hints I encountered so far.

For a superko rule, not lifting bans sounds natural. Strict superko is only used in Western rules, and to fix prisoner-unbalanced area repetition it needs non lifting passes. But Spight rules still have full passes (made possible by its stop rule).

$$B B to move
$$ +---------------+
$$ | O . O . . O . |
$$ | X O O O . O . |
$$ | . X X O O O @ |
$$ | X X X O X X X |
$$ | . . X X X . . |
$$ +---------------+
$$

Click Here To Show Diagram Code

[go]$$B B to move
$$ +---------------+
$$ | O . O . . O . |
$$ | X O O O . O . |
$$ | . X X O O O @ |
$$ | X X X O X X X |
$$ | . . X X X . . |
$$ +---------------+
$$[/go]

Also, my reading of the above under position in Ikeda Territory II rules is that black gains a tie by capturing the ko, by now having a prisoner the score is 0, and not more for black, since empty points not inside immortal stones aren't counted.

Don't forget that under area scoring (or area encore) W can always play into territory for free and retake the ko afterwards. This position affects pure/LD territory scoring (with hypothetical play), not hybrid territory + area encore like Ikeda.

Overall, I'm fairly sure this is a W win in both Japanese and Chinese rules - as well as most other sensible rulesets like Spight, even AGA/NZ or Maas encore. Frozen ko mess (under territory scoring only) in everyday positions like this would be way too blatant difference to both area scoring and common sense. Territory scoring needs ban lifting passes to work correctly (and stay close to area).

[jann]

A few things I forgot to say at the time - whether to areafy with "pass stone / penalty" or "board play reward" like Maas:

Pass stones do not directly compensate cleanup moves, but via the opponent's pass (if he doesn't respond on board). This needs an arbitrary rule about equal moves in encore. The reward method also needs extra rule though, about prisoner-pair creation (if none remains to play from) or an alternative way of reward.

Giving a prisoner for a pass distorts the concept of passing. Normally a pass is a "do nothing" or "no (score) change" thing, like a 0 point environmental coupon - ensuring players are never forced to decrease their own score. Rewarding board plays is closer to how area actually works and differ from territory (+1 pt for each played stone).

On the other hand, pass stones are also suitable for whole-game areafication, not just for encores.