In the Italian Game (1.e4 e5 2.Nf3 Nc6 3.Bc4), the move 3...Nd4?! wastes time. White should get the advantage by playing 4.Nxd4 or 4.0-0. However, as the game below shows, the move also sets a diabolical trap that may lead to a quick win for Black. It is worth trying in speed chess, albeit not in regular tournament chess unless one is the gambling type. It is tempting for White to try to punish the "silly" 3...Nxd4 by taking the pawn that Black has left unguarded. However, the attractive-looking 4.Nxe5!? actually loses material to 4...Qg5!, forking White's knight and g-pawn. Then 5.Ng4, trying to hold on to the extra pawn, would run into 5...d5! with another double attack, this time on White's bishop and knight. The natural 5.Nxf7 is even worse, as Kostic demonstrates.
White's best after 4...Qg5 is 5.Bxf7+! Ke7 6.0-0!, sacrificing a piece, counting on his two pawns, rolling pawn center and Black's exposed king to provide compensation. Graham Burgess notes G. Chandler-NN, Stockbridge 1983, which concluded 6...Qxe5 7.Bxg8 (7.Bc4!?) Rxg8 8.c3 Nc6 (8...Ne6 9.d4! Qxe4 10.d5 Nf4?? 11.Re1 wins) 9.d4! Qa5? (9...Qf6 10.e5 Qf7 may be best) 10.d5 Ne5? 11.Qh5! Nf7? (11...d6 12.Bg5+ Kd7 13.Qxh7 also wins for White) 12.d6+! 1-0 (in light of 13.Qxa5).
Steinitz mentioned 3...Nd4?! in his Modern Chess Instructor (Part II, 1895, p. 63). The name "Blackburne Shilling Gambit" is sometimes given to the line, on the basis that the great English player Joseph Henry Blackburne, a/k/a "The Black Death" (1841-1924), supposedly used it to win shillings from amateurs. No games have been found to support this claim, but it makes a nice story.