Recent online games in the Göring Gambit

I've been having a spate of games in this gambit (1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.c3, or 2.d4 exd4 3.Nf3 Nc6 4.c3 which is how I more commonly reach it).  The opening is essentially a mirror-image of the Blackmar-Diemer Gambit which I have been discussing recently, but I think that the Göring is the sounder of the two gambits, especially in the version where White meets 4...dxc3 with 5.Nxc3.  However, my recent games with both White and Black have featured the more daring 5.Bc4 offering the second pawn on b2.  Most of them have then continued with 5...d6 6.Nxc3 although none have yet reached the notoriously complicated "tabiya" that arises from 6...Nf6 7.Qb3 Qd7 8.Ng5 Ne5 9.Bb5 c6 10.f4.

Here is one game that I played from the black side.

  I am aware that 5...cxb2 is the most theoretically critical test of 5.Bc4, but having had a few drubbings on the black side of that variation in casual games, I opted instead for 5...d6, with which I have had good practical results- settling for the one-pawn advantage still generates more than enough imbalance in the position to give both sides good winning chances.  In the above game, I think 7.Ng5 is a reasonable alternative to the main line with 7.Qb3 but the problems for White started at move 10, when 10.Qd4 and 10.Nf3 are both superior to 10.Qc2, which I think leaves White struggling to prove sufficient compensation for the pawn.

There was one online game where I played White and my opponent took me on with 5...cxb2 6.Bxb2 d6 (which, alongside 6...Bb4+, is one of the two most critical tests of 5.Bc4).  I then went 7.Qb3 (though I think 7.0-0 and 7.Nc3 may be of theoretically equal value).

But then the opponent blundered with 7...Qf6?? 8.Bxf6- that sort of thing sometimes happens in online games.  My current opinion is that 7...Be6 will usually transpose into 7.0-0 lines (here I suggest looking into defending the c4-bishop with 8.Na3 or 8.Nbd2, rather than the usual immediate bishop exchange on e6) while 7...Qd7 and 7...Nh6 (both suggested by John Watson in his review of Danish Dynamite) are quite testing but White has a few improvements over Watson's analysis which are probably sufficient to give two pawns' worth of compensation.

Finally, I have one online game which has been a convincing demonstration of why after 5...d6 6.Nxc3 Nf6 7.Qb3, Black should close off the a4-e8 diagonal with 7...Qd7, preparing ...Na5, rather than the 7...Qe7 played in the game, which continued: 8.0-0 Na5 9.Qa4+ Nc6 10.Bg5 h6 11.Nd5.

Ouch.  Black could have just about hung on with 11...Qd8 here, but Black is in a very bad way after 12.Bxf6 gxf6 13.Bb5 or 12.e5 dxe5 13.Nxe5.

The point of 7...Qd7 is of course that after 8.0-0? Na5, 9.Qa4 is no longer check and so White then has to accept the exchange of the important c4-bishop for Black's knight on a5, and this is why White normally increases the pressure on f7 with 8.Ng5, typically leading to the aforementioned line 8...Ne5 9.Bb5 c6 10.f4.