In the previous triple-occurrence thread, most posters have now agreed that, in deciding whether two positions are the same for triple-occurrence purposes, it is necessary to examine not only the locations of all the pieces, but also the possible moves of the pieces, including possible future moves.
Let’s say that a player is temporarily unable to castle – for example, because the f1 square is attacked, or because there are pieces between the king and rook. Let’s say, further, that the same physical position comes up again, but now the player is permanently unable to castle, because he has moved the king or rook in the meantime.
These two positions are different. The set of moves currently available is the same in both cases, but the set of possible future moves has changed.
Even if everybody agrees with all of the above, there is still a legitimate question as to how to define the permanent loss of castling privileges. As I see it, there are (at least) two reasonable definitions:
Definition 1. Castling privileges are permanently lost if there does not exist a sequence of legal move-pairs leading to a position where the player may castle.
Definition 2. Castling privileges are permanently lost if the player has moved either the king or rook.
The two definitions are not the same. Consider the following example from the Castlemate thread (white to move):
In this position, even though white has moved neither king nor rook, there is no way white will ever be able to castle, even with cooperation from his opponent. (Please stare at the position long enough to convince yourself of this.) So white’s castling privileges are permanently lost according to Definition 1, but not according to Definition 2.
At first glance, Definition 1 seems more satisfying. It hits the nail on the head, in that it grants equal status to two identical positions if, and only if, the set of possible future moves is the same in both positions. Let’s call Definition 1 the purist definition.
But there could be practical problems. In some situations, it could be difficult for a tournament arbiter, or even a computer, to determine whether there exists a sequence of legal moves leading to castling. In the above-mentioned Castlemate thread, for example, one poster presented a position in which he believed castling would never be possible, though white could play on for 28 moves before being forced to move king or rook. That example, however, turned out to be cooked by a sequence of 16 move-pairs after which white could castle.
So there is a strong case for Definition 2. Let’s call Definition 2 the practical definition. For one thing, it’s algorithmic – to determine if a position fits the definition, just look at the move history. I’d bet dollars to donuts (hmm, that’s only about an even bet nowadays, isn’t it?), that chess engines such as Fritz would employ Definition 2. (Maybe some volunteer could set up a situation which is a triple occurrence by Definition 1 but not by Definition 2, then try to claim a draw, and see what Fritz says.)
According to some posters in previous threads, FIDE has apparently wrestled with this question, and has settled on Definition 2. Perhaps USCF should do the same, to clarify the rule and stay synchronized with FIDE.
Here’s another monkey wrench. In some situations, neither definition “feels” right:
In theory, white could retain castling privileges with Qc3, but what player in his right mind would do so, and even if he did, what opponent would play anything other than …Bxc3+, still ending white’s castling privileges?
Thorny questions, no?
Bill Smythe