FIDE threefold repetition and possible moves

Question

FIDE rules section 9.2 says:

Positions are considered the same if and only if the same player has the move, pieces of the same kind and colour occupy the same squares and the possible moves of all the pieces of both players are the same. Thus positions are not the same if:

1. at the start of the sequence a pawn could have been captured en passant.
2. a king or rook had castling rights, but forfeited these after moving. The castling rights are lost only after the king or rook is moved.

My question is about the correct interpretation of "possible moves" in regards to castling and en-passant rights. Has FIDE ever officially clarified this rule, or is there a petition out there somewhere for them to clarify this rule, to decide between these alternatives?

1. It matters only what moves can actually be made from the positions in question. (Literally, the "possible moves" for the positions.)
2. Both the moves that can be made, and the retention of rights to castle and capture en passant, must be considered and must match exactly between the positions in question.

I have two examples to illustrate the question:

---

First, castling rights. In this example, assuming both the king and rook have never been moved at the starting position, there is no possible future where White can castle. So is the final position in this example a threefold repetition or not?

[Title "White to Move"]
[FEN "4k1n1/8/8/8/8/8/4p3/2B1K2R w K - 0 1"]

1. Rh2 Nh6 2. Rh1 Ng8 3. Rh2 Nh6 4. Rh1 Ng8

Is this a threefold repetition ?

Formally, White has lost his castling right. However, even in the starting position, there was no possible legal continuation of the game where White could castle (he cannot get rid of bPe2 without moving king or rook), so even if the formal right to castle has been lost, the possibility of a later castle has not been altered by the moves played.

---

Second, capture en passant rights. In this example, the white pawn can theoretically be captured en passant, but the only black pawn that could do it is pinned. So is the final position in this example a threefold repetition or not?

[Title "White to Move"]
[FEN "2k3n1/8/8/8/2p5/8/3P4/2R1K3 w - - 0 1"]

1. d4 Nh6 2. Kd1 Ng8 3. Ke1 Nh6 4. Kd1 Ng8 5. Ke1

---

In both examples, it depends how the phrase "possible moves" in the rule is defined. Has FIDE ever clarified this rule as it concerns cases like these?

Answer 1

Hi I have clarified this over the years with the senior FIDE International Arbiter at the time, Stewart Reuben, who is busy now writing the definitive book on the History of the Laws of Chess.

You focus on the essential point: treatment of possible moves. The TL;DR is that en passant is based on possible moves, castling is based on rights.

At first glance one might think: "Whoa! How inconsistent! Why would they do it different ways?" But actually it makes sense and to my mind is the best way of doing it. Let me see if I can convince you.

In some sense, it might seem "fairest" to look at "possible moves" for both.

However for castling, it might be complex to determine whether a player is going to be able to execute a castling some time in the future. The examples that you give are cool, but relatively simple. The rules would also have to opine whether, in considering the feasibility of actually castling, one can assume that one player is actively trying to prevent this taking place, and how that interacts with the choosing of moves to repeat the position. It's sooo much simpler just to look at the game score and simply observe which kings and rooks have ever moved. The actual executability of castling is irrelevant. The right is lost, as the rule says, after the king or rook moves. A genuine hole in the rule is that the right should also be lost if an unmoved rook is captured on its start square.

On the other hand, en passant is only feasible for one turn. There is no look-ahead required in checking possible moves. If the e.p. can't happen right now because of check or pin, it seems very reasonable that this is indeed a factor in the determination of position. If I try to put a position on the board where the e.p. can't happen due to check or pin, and try to think of it as different from a situation 2.0 moves later, when the e.p. is also too stale to happen, it feels very odd.

After many years of confusion, this is the point of view which FIDE reached under Stewart. The rules could still be clearer, but I think the key step is to understand exactly why "possible moves" should indeed be relevant for one case, but not the other.

Answer 2

TL;DR: Based on what's written in the FIDE LoC, the second position is a draw, while the first isn't.

1. About castling (emphasis mine):

   9.2.2.2 a king had castling rights with a rook that has not been moved, but forfeited these after moving. The castling rights are lost only after the king or rook is moved.

   In your position, the king had castling rights but couldn't castle. Consider for example the following position:

   [Variant "From Position"]
   [FEN "3kb3/8/8/r1n5/4N3/8/8/4KB1R w K - 0 1"]
   
   1. Nxc5 Rxc5 {First time this position appears, White has castling rights} 2. Rg1 Rc8 3. Rh1 Rc5 {Technically, this is the first time this position appears, since White has lost their castling rights.} 4. Rg1 Rc8 5. Rh1 Rc5 {Second time; the game is not drawn}
   

   As you can see, White has the right to castle, but can't because of the Bishop in f1.

2. About en passant (emphasis mine):

   9.2.2.1 at the start of the sequence a pawn could have been captured en passant.

   In the second position, the pawn on c4 couldn't have been captured en passant (that would be an illegal move), so it is indeed a three-fold repetition.