Renaissance rules

[jann]

On further thought, the above new superko variants could also give renaissance to the "three passes stop" idea that various authors (Yasunaga and others) considered in the past, but which is hard to do normally (because of double ko seki). The motivation is to get rid of technical/unintended stops (when first pass was not real, only forced by ko ban).

Three passes are also how Spight rules work in practice. Passes lifting bans (and resumptions) can only partly substitute this, as sometimes it can be important to reach the correct outcome within one phase, without resumption.

One such case is hypothetical play. Another is hybrid rules that switch territory scoring to pass stone playout. This is really hard to do correctly without three passes because the first phase needs to be ko-wise complete. See Lasker-Maas rules for an unsuccessful attempt (one point error for endgame ko left open and connected in dispute phase).

[jann]

So territory mode needs three passes (for pass stone switch), and three passes cannot be used with normal superko (neither with or without passes lifting bans). The minimal variant also needs rewording in case of three pass stops (casting new light on it):

• minimal-superko: a board play must not recreate the position of the last successive two passes

I compiled a first draft, supporting both area and territory, and edited into the first post of the topic.
Anybody with spare time is welcome to skim through and share whatever came to mind.

Because of the above issues with superko, I went for the simplest approach and avoided the whole mess with ko "options", just used the correct one as single universal superko solution. The essence is: two passes block further repetition of the given position, three passes stop play completely.

[jann]

So territory mode needs three passes (for pass stone switch)

I just realized there is an actual alternative. One could keep two-pass stops, and instead switch only when a dispute starts with a pass (ie. the third pass). Simpler and familiar stops - for complicated and counterintuitive pass-stone switches. Could this worth it?