Here is an interesting proposal for an alternative rule that treats the regions directly involved in long-cycle interactions as seki (vacating the hot ko points and forbidding further play on them):
I haven’t seen it discussed anywhere else (and the SL page is 10 years old), but this is kind of a “third way” alternative to superko vs. long cycle games ending in no result.
I find it interesting because it renders a result based on the score of the unaffected regions (in territory scoring that doesn’t count points in seki) or just on the stones as they stand, with the hot points as dame (area scoring). Seems nicer than both “no result” and “boom, you hit the superko, you lose.”
On the theoretical level, it would be interesting to analyze how this might impact various rules beasts, as it seems to have the effect of incrementally rendering parts of the board as inactive.
On the practical side of things, I’m not sure if this really addresses the common complaint against superko, that is, having to remember the game history well enough to detect immediately when the superko rule should come into effect. Of course, this is mainly a convenience concern for over-the-board play.
To apply such a rule precisely, i.e., figuring out exactly when the players trigger a long cycle seki, they would still need to have a precise grasp of the game history in order to immediately detect when this rule should be applied. This is important due to concerns about who gets sente after a position becomes “settled” due to this rule. If the players have forgotten exactly when the cycle began, there could be a dispute about who owes the last move to finish the cycle and trigger this rule. Each player should want their opponent to be compelled to play the last move, in order to have sente on their own turn.
So, in practice, applying this rule does not seem any more convenient than superko, but also introduces some potential complexities that change the nature of the game.
In fact, suppose that your next move would be the one that completes a long cycle and triggers the permanent seki. If you play it, then the opponent gets sente. You would much prefer that the opponent be the one who completes the cycle, so that you get sente.
What should you do? Well, you can play a ko threat. After the opponent responds, now you can continue the fight. Since most cycles are even length (just like simple kos, which are cycle length 2, longer cycles are most commonly larger even numbers), it will eventually now be your opponent who is in the position of having to either lose sente, or find a ko threat.
I think this means there is still an incentive to play ko fights in many long cycles almost as if the rule weren’t in effect (edit: as if an ordinary superko rule were in place instead). It’s just that the stakes for the ko fight have been lowered from the life/deaths of the groups in the cycle, to instead losing one whole move. (but you still need your threats to be big enough, the life/death of the groups still gets traded away in the case of an ignored threat).
Thanks for inspiring me to give a closer reading to Pauli’s LJRG. I had been meaning to do so, but was still bogged down in contemplating Spight’s “amateur-playable” version of Japanese rules instead. (And Pauli’s rules as published at that link really need a couple of passes of copy-editing and editing for clarity… I found it more difficult to follow than it ought to have been.)
I’ve been thinking this over, and it seems to me the first player to recognize the cycle can take sente by pointing it out immediately, if it’s their turn to move, or as soon as their opponent finishes the turn otherwise. I don’t see why the sixth capture of a triple ko must actually be played to point out that a triple ko is in effect, nor why reading skill shouldn’t be rewarded.
So perhaps some versions of this rule wouldn’t necessarily have all the disadvantages of superko.
Are you thinking of some variant of the rule, where vacating the cycle points happens only when a player first points it out, rather than at the exact time of first full board repetition?
You certainly can’t point it out before the cycle has completed a full repetition or is completing a repetition as of your current move, because you can’t put moves into the opponent’s mouth, so to speak. The opponent could be planning to deviate from the potential cycle on a future move, and they shouldn’t be under any obligation to confirm or deny to you on your turn what their future plans are.
And the players shouldn’t want to delay after the cycle has completed because of the sente issue, right? So the players will still have strong incentives to declare it basically when the exact rule would have declared it (and also fight it as a normal ko rather than repeating the cycle beforehand, again due to the sente issue).
This is in fact the way it works in Pauli’s ruleset that @jannn mentioned. A player can remove a disturbing cycle (basically taking away the active stones in the ko mouths and declaring it seki) at the beginning of any move, before playing their stone.
I’m still trying to figure out what kind of dispute resolution provisions Pauli makes, if any, for when an attempted “disturbing cycle removal” is disputed by the opponent (e. g. for any of the reasons you mention), and what restrictions there might be on the timing of such moves.
The version where you can only point it out after the board state has repeated, but it only happens when you do point it out seems like it slightly reduces the human implementation difficulty in the sense that if the players miss and go over by a move or two, the game is still valid under the rules. As opposed the rules mandating the point and so the game has continued in an illegal way and needs rewinding (or a forfeit). But I think if the players are trying to play well, they still need to track the cycle or potential cycle just as precisely either way.
Pauli does make it pretty clear that cycle removal is part of a move made by one player or the other, and whenever it first becomes allowed to do so, thereafter neither player is required to remove the cycle, but the first who chooses to do so may. So as you surmise, there’s no need for rewinding or forfeit.
And yes, Pauli goes on at great length about the implications for best play, with examples of long cycles from historical games and how they might have been played out differently under the incentives provided by his suggested rules.
I understand you were referring to the post in which I hadn’t referenced Pauli’s rules directly (I was going to edit it to add them, but @hexahedron had already replied, so I just continued in thread).
However, whatever criticisms I’d make of Pauli’s particular implementation of this long-cycle-seki rule, “imprecisely formulated” isn’t one of them.
(The presentation of these rules is unnecessarily complicated, poorly organized, perhaps a little redundant, and badly needs copy-editing, and these are the main reasons I can’t pluck out just this one aspect of his system and give you a precise formulation of it yet.)
Yes, I’m specifically criticizing these words as being imprecise:
Without further clarification and precise definitions, I think that attempted interpretation and application of the above would lead to confusion and unresolvable disputes.
I can’t comment on what Pauli might have done to resolve further clarify these along a similar vein (actually the site isn’t loading for me right now), but if you can eventually extract out the precise definitions from his work, that might be helpful.