What is White's compensation in the Smith-Morra Gambit, Paulsen Formation (a6, e6, Nc6) after an early ...Na5?

Question

The variation is almost forced if White aims for the standard Smith-Morra setup. At the end White is a pawn down and no longer has two bishops, but is well developed. However Black's setup does not seem to have any weaknesses that can be pressured by White's superior pieces.

White's most immediate plan is to strike on the flank with a4, provoking ...b4.

In most of games at 2000+ level, White ends up defending a worse endgame.

 [fen ""] 
 1. e4 c5 2. d4 cxd4 3. c3 dxc3 4. Nxc3 e6
 5. Nf3 Nc6 6. Bc4 a6 7. O-O b5 8. Bb3 Na5
 9. Re1 Nxb3 10. Qxb3

If Black reveals his intention to play ...a6 early, White has alternative plans: Play Bf4, threatening Nd5 with pressure on the dark squares c7, d6 and b6, leading to decent games with White retaining the initiative. So Black should delay ...a6 as much as he can, as Bf4 only makes sense after ...a6.

[fen ""]
1. e4 c5 2. d4 cxd4 3. c3 dxc3 4. Nxc3 Nc6
5. Nf3 a6 6. Bf4 e6 7. Be2

Also, ...Na5 needs to be played immediately, as 8...Bb7 allows 9. a4 with initiative and nasty tricks from White. Example game:

[fen ""] 
[White "Vivante-Sowter, John (2259)"]
[Black "Kavc, Andrej (2270)"]

1. e4 c5 2. d4 cxd4 3. c3 dxc3 4. Nxc3 e6 5. Nf3 a6 { B21 Sicilian Defense: Smith-Morra Gambit Accepted, Kan Formation } 6. Bc4 b5 7. Bb3 Bb7 8. O-O Nc6 9. a4 b4 10. Nd5 Na5 11. Bg5 f6 12. Ne5 Bxd5 13. exd5 Nxb3 14. dxe6 dxe6 15. Qxb3 Qd5 16. Nc4 fxg5 17. Rad1 Qf5 18. Rfe1 Bc5 19. Nd6+ Bxd6 20. Rxd6 Nf6 21. Rdxe6+ Kf8 22. Qxb4+ Kf7 23. Qc4 Kg6 24. g4 Qf3 25. R1e3 Qd1+ 26. Kg2 Rad8 27. Rxf6+ gxf6 28. Qe4+ Kf7 { The game is a draw. } 1/2-1/2

Answer 1

[Event "Morra Gambit Accepted - Paulsen Formation"]
[FEN "RNBQKBNR/PPPPPPPP/8/8/8/8/pppppppp/rnbqkbnr b KQkq - 1"]
[StartPly "14"]

1.e4 c5 2.d4 cxd4 3.c3 dxc3 4.Nxc3 e6 5.Nf3 Nc6 6.Bc4 a6 7.O-O b5

To investigate the position after 7...b5, I setup a round-robin tournament between Stockfish NNUE 14.1, Stockfish Classical 14.1, Komodo Dragon 2.6.1, and Leela Chess Zero (Lc0) 28.2. These 4 engines are arguably the top chess-playing entities on the planet at the moment.

Each engine played W and B against every other engine. The time control was 10/10 and each engine was given 16 logical cores (3080 Ti for Lc0) with 8 GB of RAM.

Based on the results:

• All 12 games were drawn. This is strong evidence for the theory that White has sufficient compensation for the pawn.
• In every game every engine evaluation stayed between +1/-1 pawn. So no dramatic surprises, and more evidence that White has sufficient compensation.
• Most games were fought with a material imbalance as part of complex unbalanced positions.
• Most games were very tactical in nature - so fully fitting the typical Morra Gambit theme.
• Every engine played 8. Bd3, not 8. Bb3.
• Several games ended with a Black fortress - Black down material, but creating an impenetrable barrier to the White queen.
• Several games included a White sac on b5.
• As Black, Dragon always insisted on returning the pawn by allowing a4, axb5, and Bxb5.

Answer 2

To be honest, I see very little there, but Re1 also just looks like it is not in the spirit of the opening since it does little to add any pressure to the center. Of course, if black were to play d5, then it would make sense, but black will probably refrain from playing that for quite a while. Bc2 probably should be played there, and follow it up with the standard Qe2 and Rd1.

Overall, the only compensating factors right now are a slight lead in development, and more space; but as always, in such positions, if the other side is very solid, they can catch up in development, and then the extra pawn will become a major factor.

There is a reason the Morra is only popular at lower levels: It just loses a pawn, and there are not many redeeming factors.