[Retros] non-FEN fun

[Noam Elkies]

As it happens Chessbase.com just ran this article

http://www.chessbase.com/Home/TabId/211/PostId/4010825/the-infamous-1999-chessbase-christmas-puzzle-160813.aspx

on its first "Christmas puzzle" ever, which asked in 1999 for
a legal came starting 1 e4 and ending 5...NxR#.  [What's the
source for this puzzle?  I don't think it's given in the
Chessbase article but surely somebody on this mailing list knows it.]
This is very hard for most human players, even though the solution
is not quite unique.  But it's not the only such puzzle.  Another,
which *does* have a unique solution, starts with 1 a3 and ends
on move 5 with White giving mate with a Rook.  I don't know
the composer/discoverer of this either, but I first saw it
decades ago in a column by Pal Benko.  Are there other such examples
known, or discoverable by computer now that F.Labelle is active again
(welcome back!)?  There's of course 1 h4 ... 6 PxB=N#, which is unique
even without specifying the first move; and Labelle found a few more
such examples (all shorter) determined completely by their last move.
But I figure that specifying both first and last move should allow for
a greater and hopefully even more interesting selection of puzzles.

NDE