Jonathan Tait wrote on 01/24/19 at 00:11:17:
try 4...Qh4
Yeah, that might be worth a try, though probably not in corr. games. Black can play it in two ways: play ...Kd8, grab pawn e4 and hope to survive White's quick development or allow the fork Nxc7+ and try to prove sufficient compensation. I remember (if my memory doesn't fail me) GM Paul Motwani discussing a game like that in Schaaknieuws. I can't find it though. It made me buy GM Lev Gutman's book on 4...Qh4 many years ago. Unfortunately exactly in the spectacular lines his analysis contains quite a few flaws. After 5.Nb5
a)5...Bb4+ 6.Bd2 Bc5 doesn't make much sense to me, while Qxe4+ 7.Be2 is not exactly my cup of tea;
b)5...Bc5 6.Qe2 Nf6 7.Be3! Bxe3 8.Qxe3 looks promising for White;
c) To avoid this 5...Nf6 always has interested me. Some lines:
1) 6.Nxc7+ Kd8 7.Nxa8 Bc5 8.Qf3 (8.Qe2 Nxe4! with interesting complications) Re8 (Nd4 is better, but doesn't seem to equalize either to me) 9.Nc3 Nd4 10.Qd3 Nxe4 11.Nxe4 Rxe4+ 12.Kd1 d5 and we have a rare case of a GM producing three bad moves in a row: 13.Be3 (13.Qg3) b6 (Bg4+) 14.g3 (14.b4) Qh5+.
2) 6.N1c3 Bb4 is an improved version of line a).
3) 6.Be3 Bb4+ is normally reached via 5.Be3 Nf6 6.Nb5 Bb4+.
4) 6.Nd2 looks reasonable too.