Single Comment

I still have some doubts with regard to the application of this rule. In particular, when a (checking) position occurs from which it is possible to deliver perpetual checking, but instead of continuing the checking you play something else. And then later, that position occurs a second time. Now you do continue the checking, and the opponent has no way to escape it, so you get a 3rd and 4th repetition of the position.

Does this now count as perpetual checking? Not all moves since the first occurrence of the position were checks. But all moves since the 2nd and 3rd occurrence are. In my engines I would only consider the moves since the previous occurrence. (Especially because it already terminates the line at the 2nd occurrence; if it cannot break from the loop the first time it goes through it, nothing will change by trying it the second time, except that you have less search depth left, which will degrade the accuracy of the evaluation.)

I would think that only considering the moves since the previous occurrence would be more in line with the spirit of the rule; you try to force a repetition by perpetual checking, and this should not be allowed. It is just that you started the checking a bit late, but now that you are at it, and can keep it up perpetually, why should that matter? When I asked it to a Japanese in connection with mini-Shogi, however, he told me this would not be considered perpetual checking, and thus produce a sente loss. (Which is a special mini-Shogi rule; in regular Shogi that would be a draw.)