I wasn't really familiar with the idea ...Ne4 in combination with ...cxd4 Looking through the database, Serper is one of the champions of this plan.

This seems to be a new move, but probably doesn't have much point. 9.Qe3 gives White the option of Nf5 against b6. 9.Qd2 keeping an eye on the weak d6 square has also been played. Kramnik once played 9.Qf3 against Serper looking to stop ...b6.

I felt pretty good here. 12...Qc8 13.Nf5 with a strong attack or 12...d5 13.cxd5 Qxd5 14.Qxd5 with a small edge to White.

I found out later that Serper had played this pawn sacrifice against Glek in 1990.

13.Ba3 deserves attention

Glek, with his queen on f3 instead of e4, elected to castle here and the game eventually ended in a draw. (Glek had played 9.Qe3 and 12.Qf3).

ECO concludes that Black has compensation for the material. White's extra pawn means very little since it is on a square that can be easily attacked more times than it can be defended.

So that Black will have to swap a pair of rooks when he regains his pawn. With only one set of rooks on the board, the White king can breathe a bit easier, although Black probably doesn't have serious threats to the White king.

The game Blagojevic-Serper Croatia 1990 continued 19.Kc2 Rac8 20.c5 Nxc5 21.Bxc5 Rxc5 with a small advantage to Black which Serper managed to convert into a full point. That line of play doesn't make much sense to me. White's only trump in the position is the long-term superiority of the bishop over the knight.

20...Rxc4

Black would probably like to keep all the rooks on the board and capture c4 with his knight. It looks like 20...Na5 doesn't work because of 21.c5 Then 21...bxc5 drops a pawn to 22.Ra4. However, Black could try 21...Nb7!? looking to transpose to an ending similar to Blagojevic-Serper. White can probably get away with 22.cxb6 Rxc3+ 23.Kb2 with approximate equality.

Getting the pawn on the opposite color of the Bishop.

I didn't mind letting him fix the f4 pawn on Black, since I easily dissolve this with e4.