Datum der Originalpublikation nicht 100% sicher, laut ACM aus der "Jamaica Gleaner Christmas column".

Originalforderung: How has the position been arrived at and who is the winner, and in how many moves?

From the Jamaica Gleaner: "White mates in two moves. The last move made was by Black playing his Bishop and announcing mate. As it can be demonstrated that the Bishop is not a promoted Pawn and that Black's King's Knight's Pawn was captured on its original square by White's Queen's Knight's Pawn the Black Bishop must have been played from Bishop's square (f8) to Q3 (d6). This being an illegal move, White enforces the penalty of compelling Black to retract it and move his King whereupon White plays 1 PXB(Q) ch (1.gxf8=Q+) and mates next move by 2 Q-B4 (Qf4#). The following is a brief but pointed analysis, demonstrating the false move: White's Pawns have made six captures all on black squares. The Q Kt P (Pb2) made five of these and consequently captured the Kt P on the square upon which it now stands (g7). They could not have captured the Q B which is also lost. The White Bishop is the QRP (Pa2) promoted, the original KB having been captured on its own square as the unmoved Pawns show. To allow this promotion Black's QRP (Pa7) made two captures, the QKtP (Pb7) one, and the QBP (Pc7) two. The KRP (Ph7) has also made a capture, which accounts for the seven pieces White has lost. The Black Bishop is not a promoted Pawn, as if the Black KBP (Pf7) had played to the 7 th square (f2) and then captured a White piece on K or Kt square (e1 or g1) the captures by White Pawns cannot be accounted for without including the Black QB or KRP neither of which is available. As it can be demonstrated, then that the Black Bishop is not a promoted one, and that the KKtP was captured by the White Pawn which now stands on that square, in order to reach Q3 the Black Bishop must have an impossible move.

Der Kolumnist des ACM merkt aber zurecht an:
ACM: The above is a very fine piece of analytical work; but there is a slight flaw in connection with the minor condition, 'mate in two'. In a position of this kind we believe only that which can be proved; thus we do not think that White has any right to enact a penalty, as neither the analysis nor the conditions show that the Black Bishop came from Bishop's square on his last move; indeed, that Bishop may have played outside the Pawns on the very first move of the game which, being played, brought about the position.
HBae: White plays 1 PXB(Q) ch (1.gxf8=Q+). Muß der sK nicht auf f5 stehen? (2019-10-22)
Henrik Juel: Last move (supposedly) was Lf8-d6#, which is obviously impossible and hence illegal
The penalty for this is that Black must replace Ld6 on f8 AND instead make an arbitrary move with his king
So the forward play is
0... Kf5,Kf6,Kh6 1.gxf8=D+ Kg5,Ke5 2.Df4# (2019-10-22)
A.Buchanan: One long-standing approach to resolving illegal diagram jokes is to suppose that only the last move was illegal, with all prior play legal. The illegal move is then retracted, and play continues. Of course, the “illegal move” might in principle be from *any* legal position (even the game array!). So for sanity, we say the illegal move is a simple but somehow illegal shift of a single piece.

So here, candidates for the last move include Pe5-a4+, Sf4-e1+, Ke5-g5+ & B?-d6+. For all of these, White has 6 visible pawn captures, all on dark squares, so Black light-squared bishop is excluded. wPa must have promoted to light-squared bishop, so if the three Black pawns on a-file remain, there is only one unaccounted capture. Thus bPh could not promote, and must be bPg6 now. Thus bPg7 was captured at home, and bBf8 was thus locked in.

So Bf8-d6 is certainly a possible illegal move, but so are e.g. Be8-d6 (as the light-squared bishop is otherwise unexplained) and Pe5-a~. This is an example of an "implausible" joke according to Dawson & Hundsdorfer, because there is more than one retraction to the current position, and one just has to arbitrarily pick the one that makes the forward logic work. (2023-04-02)
A.Buchanan: Another issue is that according to the 1883 laws, White cannot force Black to move their king. The 1883 rules stated:
- If a player touches a piece or Pawn of his own he must move it.
- If he touches one of his adversary's he must take can be taken.
- If he touches plurality of pieces or Pawns of the same colour, in either of these instances his adversary may elect which such piece or Pawn he will call upon him to play or to take, as the case may be.
- If the rules governing the moves of pieces do not admit of the adversary exacting penalty as above, the player must move his King, but may not Castle. If the King cannot be moved without exposure to check, no penalty can then be exacted
So according to this, Black must play Bf8xPg7 as the penalty move.
Was there another revision to the rules between 1883 & 1891? (2023-04-02)
A.Buchanan: I don't think this is cooked, I think it's just a very bad joke. Most time when moves are retracted it's because of self-check or illegal castling, occasional illegal en passant. But here we are meant to conclude that the Bl bishop has hopped over Bl pawn. There are problems by Hieronymus Fischer e.g. P1309484 where we are meant to conclude that a unit has escaped its game array cage somehow, but the piece is simply replaced: there is no move to be retracted and supplanted. But as long as such a problem is true to its own (joke) internal logic, it can't be described as cooked. If it fails (e.g. P1246715) it can indeed be cooked and repaired. (2019-10-22, hidden) more ...
comment

Henrik Juel: HC+ Popeye 4.61 (2022-11-28, hidden)
A.Buchanan: This is not a retro. There is no “can’t castle”. This is simple application of castling convention to a forward problem. This is not generally taken to make the problem “retro”. Castling convention is certainly used in many retros, but not here (2022-11-30, hidden)
A.Buchanan: Nice - could be done without twinning, just remove bPh7 for 2 solutions? (2022-11-30, hidden) more ...
comment

D. Meinking: Comm., USPB 1995-1997 (2002-07-06)
Moldenhauer: Computerprüfung: C+ Natch 3.3 in 45 Std. 46 Min.
Bei 16.5, 17.0 keine Lösung. BP 18.0 64 Lösungen.
C+ Euclide 1.11 in 44 Std. 07 Min. (2023-01-10)
A.Buchanan: This is a cool problem with tempo loss for both players (bK9 & wPf2). Noam likes to end proof games with checkmate, and in this case I guess it serves to eliminate shorter solutions which would otherwise save half a move. Checking 16.5 is sufficient for soundness. Thanks for taking the time to test. (2023-01-11)
A.Buchanan: Bit baffled how there can be 64 solutions in 18.0 moves since White just moved. 18.5 maybe? (2023-01-11)
ras@bsiag.com: Quickly checked because 45h seemed a lot for only 17.5 moves and only a few captures: Stelvio 0.98 is done in 210 seconds. (2023-01-11)
ras@bsiag.com: Well, Stelvio treats shorter solutions of same parity as cooks, so any shorter solution in 16.5, 15.5 etc would have been found by checking for 17.5. But anyhow, quickly checking 16.5 is done in a few seconds, finding no solution (of course). (2023-01-12)
A.Buchanan: Thanks for this. In order to be sure that no shorter solution exists, I’d also check 16.0. I agree it’s unlikely that there would be a solution, but it’s quick to check, for completeness (2023-01-11, hidden) more ...
comment

AB: First proof of "can castle", without recourse to conventions. (2009-11-22)
A.Buchanan: deadpos doesn't understand about retro logic yet, so does not eliminate R: 1. Ke4xLSBe3? (2022-11-13)
AB: typo in diagram: Be6 is f5. (2001-12-29, hidden) more ...
comment

Henrik Juel: an important keyword is missing: Donati theme
same in (some of) P1011889-P1011929 (2018-12-18)
Henrik Juel: Natch 3.1 found nothing in half an hour (2018-12-18)
Michel Caillaud: Reprint : H71 p.585, FIDE Album 2001-03

Cooked using Jacobi 0.7.5 with constraints :
1.c4 e5 2.c5 Ld6 3.cxd6 g5 4.d3 g4 5.e3 g3 6.Dg4 c5 7.Sc3 c4 8.Sa4 c3 9.Ld2 Sh6 10.0-0-0 c2 11.b4 Tg8 12.Kb2 gxf2 13.g3 Tg6 14.Lg2 f1=L 15.Ka3 c1=L+ 16.Sb2 Le2 17.Le1 Ld2 18.Tc1 Sg8 19.Txc8 Lc1 20.Tc3 Lf3 21.Se2 Dh4 22.Tf1 Sf6 23.Lh1 Lg2 24.Tf3 Lf1 25.Dh3

Correction : sTg6 to g4 (21...Tg4 instead of 21...Tg6 in the solution)
With the set of constraints used to uncork the cook, Jacobi 0.7.5 finds only the intended solution... (2021-09-20)
Michel Caillaud: The correction sTg6 to g4 was published in StrateGems 99 (p.138, sept. 2022)
It is C+ by Stelvio 0.93 (2022-12-18)
comment

See P0003365 (cooked) & P1411659 (fixably cooked but non-standard thematic material).
This new version has an obvious promoted unit (non-thematic), but Joaquim Crusats (a noted constructor) could not find any way to do better
Ladislav Packa: Mr. Bebesi's first name is Gyula (not Gyulia). It is the Hungarian version of the name Julius. (2023-12-05)
A.Buchanan: Thanks - edited (2023-12-06)
Henrik Juel: if last move was b2-b4: 1.axb3ep bxc6+ 2.b5 cxb6ep#
if last move was d2-d4: 1.exd3ep Dxg8+ 2.d5 cxd6ep# (2023-12-04, hidden)
Henrik Juel: HC+ Popeye 4.61 (2023-12-04, hidden) more ...
comment