[Retros] The problemist May 2008

[Noam Elkies]

Alain> Concerning the following problem, i have a proof game leading to
Alain> the position where castlings rights allow two mates in two.

Alain> 4k2r/1p2ppsb/1p2p2Q/bP2S1S1/2P2Pp1/2P2Pp1/2PR2P1/R3K3

Joost> White can't castle either if b6 came from c7.

Note too that the forward play here is not as transparent as is usual
for such problems.  The try 1 Rad1? (2 Rd8#) fails not because of
1...0-0? (2 Qxh7#) but because 1...Bxc3 pins Rd2; thus White's reason
for preferring 1.0-0-0 is to get out of the prospective pin, not
to prove 1...0-0 is impossible as one typically expects in such problems.
The try 1 Sexf7? (same threat and answer to 1 0-0) fails to 1...Bd3!.
1 Sgxf7 (same threat) deals with both Bishop defenses (1...Bxc3/Bd3
2 Ra8/Qxh8#) but not with 1...0-0.  So the choice between 1 0-0-0
and 1 Sgxf7 depends on the history of the diagram.

NDE