3 time repetition question

[ben2600]

Hello all,

This has never happened, so it is theoretical. I asked an NTD, since I did not know, and he had no clue!

Imagine in some random position, where white may castle later, black plays Qa5+. And in this position, all white legal moves are King moves (no way to capture the queen or block). So, in this position, after Qa5+, one could argue (like me) that white cannot castle (ever), since all legal moves are king moves.

Then, let’s suppose white plays Kf1, and black plays Qb6, white plays Ke1, and black plays Qa5+. So… now white cannot castle, because he moved his king twice. The first time this position “occurred” white could not castle, because every legal move was a king move to stop check.

In my opinion, this position has now occurred twice, because in both positions, through any series of legal moves after Qa5+, white cannot castle. Although I can easily see it argued the other way. White hasn’t moved his king yet, and even though he is forced to, he still hasn’t done so yet! lol

I think a VERY strict interpretation of the rule-book would indicate the positions are different, but I would prefer that they are treated as the same position.

Opinions?

Thanks!

Ben

[Mulfish]

14C2 specifies “ if the possible moves of all the pieces are the same, including the right to castle”. In this case the possible moves in the position are the same, but the long term “right to castle” is different. I can see an argument either way, but I would give more weight to “possible mives@ than to “right to castle”. This non-TD says it’s a valid draw claim.

[wintdoan]

I believe that it is settled law that the first position is different, that is, the “right to castle” isn’t lost until the King actually moves.

[relyea]

Messrs. Mulford and Doan are correct. FIDE explicitly states this.

Alex Relyea

[Mulfish]

@relyea

Messrs. Mulford and Doan are correct. FIDE explicitly states this.

Alex Relyea

Well, Mr Doan and I differed so we can’t both be right. I argued that because in all three cases castling wasn’t legal that the requirement was met, even though in the first case the inability to castle was temporary and in the later cases it was permanent. He disagreed, arguing the right to castle existed in the first case even though it wasn’t legal.

[scottrparker]

There is a difference between the “right” to castle and the “ability” to castle. In the first position given the King has not lost the right to castle. It does not have the ability to do so because it is in check, but it has not yet moved, so it retains the right to castle. When the position repeats the King has moved, and so has lost the right to castle. The positions are therefore enough different that they do not constitute a repetition of position as defined by 14C. Triple Occurrence of Position, which requires that the right to castle be the same in both instances. My two cents.

[Smythe_Dakota]

@ben2600

… In my opinion, this position has now occurred twice, because in both positions, through any series of legal moves after Qa5+, white cannot castle. Although I can easily see it argued the other way. …

I think I posed a similar question several years ago, somewhere on these forums. My opinion was the same as yours, that since all possible future moves are the same in both cases, the two positions should be regarded as the same.

And, last time as this time, others disagreed, and I could understand the other side too.

From the purist standpoint, I stand by my opinion. But from a practical standpoint, the purist standpoint may be impractical. (Yogi Berra must have said something similar once upon a time.)

Suppose you are building a chess engine – not one that plays well, just one that tracks two players playing an online game against each other, and that is capable of ruling on draw claims by the players.

To watch out for repeated positions, the engine would have to maintain a list of all positions reached so far in the game. Not difficult – the number is finite and small. A “position” would have to be defined as consisting not only of a list of which pieces are on which squares, but also a set of six “has this piece ever moved?” flags, one each for the squares a1, e1, h1, a8, e8, h8.

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

Then, to determine if a position has previously appeared, the engine need only compare the current position, including all squares and all flags, against each previous position, square by square and flag by flag.

But this would work only if you use the “practical” definition of repeated position rather than the “purist” definition. It would be extremely difficult, probably impossible in general, for an engine to figure out, in comparing two positions, whether “all possible future moves” are the same in the two positions.

In fact, I could easily see one of this forum’s puzzle experts coming up with a “rulebook puzzle” position requiring 20 moves of difficult retrograde analysis to figure out whether castling will ever be possible.

So I don’t blame FIDE for adopting the “practical” definition.

Bill Smythe

[relyea]

@Mulfish

@relyea

Messrs. Mulford and Doan are correct. FIDE explicitly states this.

Alex Relyea

Well, Mr Doan and I differed so we can’t both be right. I argued that because in all three cases castling wasn’t legal that the requirement was met, even though in the first case the inability to castle was temporary and in the later cases it was permanent. He disagreed, arguing the right to castle existed in the first case even though it wasn’t legal.

Oops. Sorry. I misread. Mr. Doan is correct.

Alex Relyea

[Allen]

@wintdoan

I believe that it is settled law that the first position is different, that is, the “right to castle” isn’t lost until the King actually moves.

Indeed. We have a winner.

[Smythe_Dakota]

This topic has been covered in several earlier threads:

Triple occurrence questions
Another 3-fold question
Triple occurrence -– a graduate course?

And, more whimsically:

Castlemate contest

The third of the above four includes a discussion of the following position, white to move –

–- where, even though neither white’s king nor white’s rook has ever moved, there is no way white will ever be able to castle.

Bill Smythe

[wintdoan]

@Smythe Dakota

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

Actually, you would need to allow for 14 possible e.p.'s for the player on move to fully describe a static position (a given b-g pawn could be e.p.'ed from either direction).

[narceleb]

@Smythe Dakota

Suppose you are building a chess engine – not one that plays well, just one that tracks two players playing an online game against each other, and that is capable of ruling on draw claims by the players.

To watch out for repeated positions, the engine would have to maintain a list of all positions reached so far in the game. Not difficult – the number is finite and small. A “position” would have to be defined as consisting not only of a list of which pieces are on which squares, but also a set of six “has this piece ever moved?” flags, one each for the squares a1, e1, h1, a8, e8, h8.

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

All of that information is in the Forsyth-Edwards Notation (FEN). All you have to do is keep a history of those.

[jwiewel]

@wintdoan

@Smythe Dakota

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

Actually, you would need to allow for 14 possible e.p.'s for the player on move to fully describe a static position (a given b-g pawn could be e.p.'ed from either direction).

If the flag is on the position then one flag. Was the last move a pawn moving two squares and ending adjacent to an opposing pawn? Which pawn doesn’t matter because you can never again have a pawn that had just moved two squares to repeat that position.

[narceleb]

@jwiewel

@wintdoan

@Smythe Dakota

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

Actually, you would need to allow for 14 possible e.p.'s for the player on move to fully describe a static position (a given b-g pawn could be e.p.'ed from either direction).

If the flag is on the position then one flag. Was the last move a pawn moving two squares and ending adjacent to an opposing pawn? Which pawn doesn’t matter because you can never again have a pawn that had just moved two squares to repeat that position.

FEN just has the e.p. “target square” if a double pawn move was the last move made.

[wintdoan]

You’re right—even if there are 14 possible e.p. moves, there are only 8 possible squares on which it could take place.

[jwiewel]

@narceleb

@jwiewel

@wintdoan

@Smythe Dakota

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

Actually, you would need to allow for 14 possible e.p.'s for the player on move to fully describe a static position (a given b-g pawn could be e.p.'ed from either direction).

If the flag is on the position then one flag. Was the last move a pawn moving two squares and ending adjacent to an opposing pawn? Which pawn doesn’t matter because you can never again have a pawn that had just moved two squares to repeat that position.

FEN just has the e.p. “target square” if a double pawn move was the last move made.

I don’t use FEN. Is the target square populated for every two-square pawn move? Or just if en passant is a possibility? If the first then a different e.p. flag (Yes/No) would be needed for three-fold repetition checking. If the second then the target square suffices as the flag.

To be complete, I should probably make a slight revision to the statement that there cannot be a second two-square pawn move that brings about the same position (considering just the pieces/pawns on the board). If you are playing bughouse then there may be a capture of a pawn and then the placement of a new pawn on a starting square that is then moved forward two squares (the pawn does not have the be on the same file as the earlier pawn move, so the two apparently identical positions could have different pawns subject to en passant). That would only be a potential issue if there was an organizer or TD that actually allowed three-fold claims in bughouse.

[wintdoan]

@jwiewel

@narceleb

@jwiewel

@wintdoan

@Smythe Dakota

(There would also have to be sixteen “is this pawn en-passantable?” flags, one for each square on the 4th and 5th ranks, and a “whose move is it?” flag.)

Actually, you would need to allow for 14 possible e.p.'s for the player on move to fully describe a static position (a given b-g pawn could be e.p.'ed from either direction).

If the flag is on the position then one flag. Was the last move a pawn moving two squares and ending adjacent to an opposing pawn? Which pawn doesn’t matter because you can never again have a pawn that had just moved two squares to repeat that position.

FEN just has the e.p. “target square” if a double pawn move was the last move made.

I don’t use FEN. Is the target square populated for every two-square pawn move? Or just if en passant is a possibility? If the first then a different e.p. flag (Yes/No) would be needed for three-fold repetition checking. If the second then the target square suffices as the flag.

I’m assuming that the same logic would apply to a potential e.p. If (say), White plays f2-f4 with a Black pawn at g4, Black would have the “right” to play g4xf3 in this position (i.e. f3 would be considered an e.p. square in the description of the position), even if g4xf3 puts Black in check.

[narceleb]

@jwiewel

Is the target square populated for every two-square pawn move? Or just if en passant is a possibility?

Every two-square pawn move.

@jwiewel

If the first then a different e.p. flag (Yes/No) would be needed for three-fold repetition checking.

There is – that “flag” being whether there is a pawn in position to make the capture. No other flag is necessary.

[jwiewel]

Different but related question with a few preconditions.

1. illegal moves only claimed by the player

2. somewhat inexperienced players

3. TD having a lot of games to watch and not watching the particular game

4. White Kh8, Rh6, Qb5, Black Ka1, Qf7, Be4

Play goes … Qf8+, Kh7 Qf7+, Kh8 Qf8+ , Kh7 Qf7+, Kh8 and black claims three-fold repetition claim to the TD by writing Qf8+ and saying that would be played. Both players had twice missed that Qf8 was mate, not just check, and played on.

Does the TD award the three-fold repetition?
Does the TD say that mate had already ended the game four moves earlier?
Does the TD say that the written move is mate and that ends the game?
Does the TD enforce the illegal move rule for moves within the last 10 moves and return to the position where white made the illegal move to continue from there (and white can’t play on because it is mate)?

[narceleb]

@jwiewel

Different but related question with a few preconditions.

1. illegal moves only claimed by the player

2. somewhat inexperienced players

3. TD having a lot of games to watch and not watching the particular game

4. White Kh8, Rh6, Qb5, Black Ka1, Qf7, Be4

Play goes … Qf8+, Kh7 Qf7+, Kh8 Qf8+ , Kh7 Qf7+, Kh8 and black claims three-fold repetition claim to the TD by writing Qf8+ and saying that would be played. Both players had twice missed that Qf8 was mate, not just check, and played on.

Does the TD award the three-fold repetition?
Does the TD say that mate had already ended the game four moves earlier?
Does the TD say that the written move is mate and that ends the game?
Does the TD enforce the illegal move rule for moves within the last 10 moves and return to the position where white made the illegal move to continue from there (and white can’t play on because it is mate)?

Mate at the first …Qf8 ended the game.

Had that situation with my own son. He play a double-check (Knight and Queen) and mate. Neither player noticed. Kept playing. When his opponent finally noticed he was in check and called me over, I had them back up the game. Then he realized he was mated. Neither had even noticed it was check.