Pass As Ko Threat
An interesting question is whether passes can serve as ko threats.[1] In other words, can a player recapture in a ko if he did not play a ko threat on the board, just passed his relevant turn instead?
Suppose it's black's turn here and he takes the ko. This is useless by normal go logic: black is still dead. White cannot take back the ko immediately, but can come back anytime later and capture black if he wants.
But some ko rules don't take passes into account. In their case, if white doesn't even want to play a "dummy" threat (because the board is full like here, or because of some scoring issues), the situation can get tricky: white may even get locked out of the ko forever. This is particularly an issue for Superko rules, since their nature is to look for repetitions of the board position only.
AGA rules are an example of a rule set which would not allow white to capture.
Note that when white passes and takes back the ko after, the position does seem to repeat, with black to move again. But the game doesn't enter a perpetual cycle, as the situation is not the same: it is now black who cannot take back the ko immediately, so white will have time to finish the capture.
Daniele: If I understand correctly, here the point is that under simple Ko rules White would be able to capture Black. This is complicated, but it happens: White passes, then Black also passes, then the game is over, then White declares Black's stones as dead, Black does not agree, then play resumes, White captures the Ko again, Black cannot capture because of simple Ko rules, and White connects the Ko and finally kills Black's group. Simple Ko rules don't give an infinite cycle here.
Under Superko rules, instead, the result is different, we have seen above that White cannot capture.
Since we have two different outcomes, we have to decide which one we prefer. By normal go logic: black is still dead (only one eye, or a fake eye after he took the Ko). So it seems we prefer simple Ko here, instead of Superko. Moreover it is strange that Superko and simple Ko give different results in a situation where simple Ko would not produce an infinite play. (Are there other examples of this, btw?)
Could we solve this issue by introducing a three-fold repetition Superko rule (inspired by chess)? I am proposing to introduce a rule that forbids to create a situation (position + side to move) that has already appeared twice during the same game. This rule must be complemented by the usual simple Ko rule (that forbids all cycles of lenght 2).
The rationale is that if the situation has only appeared once, maybe the second time the players will do something different, and we are not necessarily entering in an infinite cycle. If the same situation appeared twice and is about to repeat a third time, then we are probably entering in an infinite cycle, and we must stop it.
This proposed rule would solve this issue, but I see it also has defects. It is a bit ugly to deal with cycles of lenght 2 very differently from what we do about cycles of longer length, this is a defect. Another defect is that it is probably possible to find a bad example where also this rule fails to respect the normal go logic. But there is no perfect Superko rule until now, isn't it?
[1]
Bill: This is a popular way of putting it, but it is a misstatement. Any board play, not just a ko threat, typically lifts a ko or superko ban. The question is rather, does a pass act like a regular play? For Ing, passes lift ko bans, and that is why he defined a pass as a kind of play.
However, if only a pass lifts a ko ban, as with the hypothetical play of the Japanese '89 rules, calling a pass a ko threat is apropos.