Mate in 3 moves before the FIDE rules changed in 1974

[OliverBr]

I am sure you know the famous "Mate in 3" Problem, that was only solvable before the adaption of the FIDE rules concerning castling:

[d]8/8/4P3/3p4/P1p3p1/1pP1kPPp/1P5P/R3K2R w KQ 0 1

My question is: Were Chess Computer that were programmed before 1974 and followed *exactly* the FIDE rules able to solve it?

Discuss!

[bob]

I am sure you know the famous "Mate in 3" Problem, that was only solvable before the adaption of the FIDE rules concerning castling:

[d]8/8/4P3/3p4/P1p3p1/1pP1kPPp/1P5P/R3K2R w KQ 0 1

My question is: Were Chess Computer that were programmed before 1974 and followed *exactly* the FIDE rules able to solve it?

Discuss!

I am not sure what rule change you are talking about...

[Dirt]

I am sure you know the famous "Mate in 3" Problem, that was only solvable before the adaption of the FIDE rules concerning castling:

[d]8/8/4P3/3p4/P1p3p1/1pP1kPPp/1P5P/R3K2R w KQ 0 1

My question is: Were Chess Computer that were programmed before 1974 and followed *exactly* the FIDE rules able to solve it?

Discuss!

I am not sure what rule change you are talking about...

Checking Wikipedia, it seems that the old rule could be read to allow white to castle vertically after promoting his e pawn to a rook. I'd be very surprise if any program implemented castling this way.

[M ANSARI]

I am sure you know the famous "Mate in 3" Problem, that was only solvable before the adaption of the FIDE rules concerning castling:

[d]8/8/4P3/3p4/P1p3p1/1pP1kPPp/1P5P/R3K2R w KQ 0 1

My question is: Were Chess Computer that were programmed before 1974 and followed *exactly* the FIDE rules able to solve it?

Discuss!

I am not sure what rule change you are talking about...

Checking Wikipedia, it seems that the old rule could be read to allow white to castle vertically after promoting his e pawn to a rook. I'd be very surprise if any program implemented castling this way.

What ???? That certainly is news to me!

[elpapa]

I am sure you know the famous "Mate in 3" Problem, that was only solvable before the adaption of the FIDE rules concerning castling:

[d]8/8/4P3/3p4/P1p3p1/1pP1kPPp/1P5P/R3K2R w KQ 0 1

My question is: Were Chess Computer that were programmed before 1974 and followed *exactly* the FIDE rules able to solve it?

Discuss!

I am not sure what rule change you are talking about...

Checking Wikipedia, it seems that the old rule could be read to allow white to castle vertically after promoting his e pawn to a rook. I'd be very surprise if any program implemented castling this way.

What ???? That certainly is news to me!

I had to look it up. Turns out there was a loophole in the castling definition. So they didn't change the rule per se, just the wording.

[OliverBr]

I am sure you know the famous "Mate in 3" Problem, that was only solvable before the adaption of the FIDE rules concerning castling:

[d]8/8/4P3/3p4/P1p3p1/1pP1kPPp/1P5P/R3K2R w KQ 0 1

My question is: Were Chess Computer that were programmed before 1974 and followed *exactly* the FIDE rules able to solve it?

Discuss!

I am not sure what rule change you are talking about...

About that line:

Code: Select all

1.e7 Kxf3 2. e8=R Kg2 3. O-O-O-O# (or Ke3#)

This is no cheat, it was possible with the origin chess rules that were valid for hundreds of years:

Code: Select all

1. The king must never have moved;
2. The chosen rook must never have moved;
3. There must be no pieces between the king and the chosen rook;
4. The king must not currently be in check.
5. The king must not pass through a square that is under attack by enemy pieces.
6. The king must not end up in check (true of any legal move).

If any chess programm would have implemented this rules in a EXACT way it should find the mate in 3 with that O-O-O-O castling!

To avoid this move they *added* 1974 this rule:

Code: Select all

7. The king and the chosen rook must be on the same rank.

[Dirt]

This is no cheat, it was possible with the origin chess rules that were valid for hundreds of years:

Code: Select all

1. The king must never have moved;
2. The chosen rook must never have moved;
3. There must be no pieces between the king and the chosen rook;
4. The king must not currently be in check.
5. The king must not pass through a square that is under attack by enemy pieces.
6. The king must not end up in check (true of any legal move).

It could be argued that the rook has moved even though it was a pawn at the time. Not that I would say that - I think your interpretation is more natural, but if this had ever been played in a real game I'm sure someone would have.

It's an interesting bit of history, thanks for pointing it out.

[OliverBr]

This is no cheat, it was possible with the origin chess rules that were valid for hundreds of years:

Code: Select all

1. The king must never have moved;
2. The chosen rook must never have moved;
3. There must be no pieces between the king and the chosen rook;
4. The king must not currently be in check.
5. The king must not pass through a square that is under attack by enemy pieces.
6. The king must not end up in check (true of any legal move).

It could be argued that the rook has moved even though it was a pawn at the time. Not that I would say that - I think your interpretation is more natural, but if this had ever been played in a real game I'm sure someone would have.

It's an interesting bit of history, thanks for pointing it out.

I am programming chess engines since 12 years but yesterday first I met this Pam-Krabbé-Rochade.

As I never heard about Rule 7 that means that I was implementing the castling wrong as none of my engine would have allow this castle.

But your are right about the interpretation issue. If you have only rules 1 to 6 you might interprete that the promotion does not give a new piece, it only CHANGES the piece (was a pawn) so it moved already, thus cannot promote.

One of my future projects is to program a complete object orientated chess engine.

There this interpretation will be very important.

If the programmer implements that the promotion will get a new piece (this is the most natural way) with the default attribute "boolean moved=false" in the constructer and implement rules 1 to 6 exactly without rule 7, the Pam-Krabbé-Rochade will be possible without the programmer even having heard about it!

But the programmer interpretes it differently, keeps the instance of the pawn, only change it to be a rook, the boolean attribute "moved" will be true and there is no Pam-Krabbé-Rochade.

That's the mightiness of object orientated programming.

Very interesting!

[hgm]

Rest assured that this kind of 'castling' has _never_ been legal, even before the rule change. I don't even think the original rules were ambiguous in this respect. Surely a Rook that is obtained through promotion must have been moved, as it was not on that square in the opening setup.

[OliverBr]

Rest assured that this kind of 'castling' has _never_ been legal, even before the rule change. I don't even think the original rules were ambiguous in this respect. Surely a Rook that is obtained through promotion must have been moved, as it was not on that square in the opening setup.

If it was that evident as you say why should they bother changing the FIDE rules?

Rest assured that this kind of 'castling' could very well have been legal. A Rook that is obtained through promotion has only moved if you define that a promotion doesn't give a new piece. But it's common sense that you get a new piece by promotion and this piece hasn't moved yet.

"Tim Krabbé composed a joke chess problem containing vertical castling (king on e1, promoted rook on e8). The loophole in the definition of castling upon which this problem was based was removed by the new requirement that the castling rook must occupy the same rank as the king."

Source: http://en.wikipedia.org/wiki/Castling