Ah, the Kieler Gambit is an old love of mine. In corr. chess, after the introduction of Fritz and Chessbase, I have 2½/3 and could have won the drawn game too.

Analysis in the sources I know is quite lame. The way Black should go is 9...e5 (iso Qe4+) 10.Qc3 Bxa3 11.bxa3 Bf5 12.f4 Nc2+ 13.Kf2.
But Black has much stronger: 7...Bf5! (iso Nd4) 8.Na3 Qxd5 9.Be3 (9.bxc6 Qe4+ -+ 10.Ne2? Nd3+ 0-1, Fricker-Schneider,1965) Qe4 10.b3 Nd4 11.0-0-0 e5 12.Bxd4 exd4 -+Carrasco-Peric,1998.

But to be honest, with correct play White can preserve some advantage. As is this advantage is not bigger than teh += White can prove after 4...Nb6 and 4...Nf6, I dare to state that the "cheap trap" 4...Nb4 is not inferior.