Why is 15.cxd6 better than 15.cxb6 here?

Question

In Radjabov vs. Gelfand (2013):

[FEN ""]
1.Nf3 c5 2.c4 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e6 6.g3 Qb6 7.Nb3 Ne5 8.e4 Bb4 9.Qe2 d6 10.f4 Nc6 11.Be3 Bxc3+ 12.bxc3 Qc7 13.Bg2 e5 14.c5 b6 15.cxd6 Qxd6 16.O-O O-O 17.f5 Rd8 18.Rfd1 Qa3 19.Rxd8+ Nxd8 20.Bg5 Ba6 21.Qd2 Qe7 22.Rd1 Nb7 23.Bf3 Rd8 24.Qc1 Rxd1+ 25.Bxd1 Nd6 26.Bxf6 gxf6 27.Qe3 Qc7 28.a4 Qd7 29.a5 Nxe4 30.Bc2 Qb5 31.Qf3 Ng5 32.Qg2 bxa5 0-1

On move 15, White plays 15.cxd6 instead of 15.cxb6. Why? I can see two reasons to prefer the latter:

1. It preserves a weak pawn on d6 (and by extension a weak square on d5).
2. It makes a new isolated pawn on b6 that's on a semi-open file too.

It seems to me that the obvious plan is 15.cxb6 intending O-O, Rfd1 and Rab1, etc. However, Radjabov didn't play this, and Stockfish doesn't like it either; at depth 24 it says 15.cxd6 is 0.00, but 15.cxb6 is -0.80. What's wrong with taking on b6?

Answer 1

There are a few ways White is at a disadvantage here that I found:

• Black’s dominance in the center.
• The active black knights.
• White is left with 3 pawn islands, which is never considered good.

Look at White’s pieces here-no really good moves are available. The knight can't move up and come into play. The rooks are undeveloped/have no open files to dominate. The bishops don't have room to move freely, and the same goes for the queen.

    [Event "?"]
    [Site "?"]
    [Date "?"]
    [Round "?"]
    [White "?"]
    [Black "?"]
    [Result "?"]

    [FEN ""]
        1.Nf3 c5 2.c4 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e6 6.g3 Qb6 7.Nb3 Ne5 8.e4 Bb4 9.Qe2 d6 10.f4 Nc6 11.Be3 Bxc3+ 12.bxc3 Qc7 13.Bg2 e5 14.c5 b6 15. cxb6 axb6 16. O-O O-O

If you continue with this all hell breaks loose for White:

[Event "?"]
[Site "?"]
[Date "?"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "?"]

[FEN ""]
1.Nf3 c5 2.c4 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e6 6.g3 Qb6 7.Nb3 Ne5 8.e4 Bb4 9.Qe2 d6 10.f4 Nc6 11.Be3 Bxc3+ 12.bxc3 Qc7 13.Bg2 e5 14.c5 b6 15. cxb6 axb6 16. O-O Ba6 17. c4 Nd4

Bxc4 is always coming, Ng4 is possible at some point, and Nc2 is a possibility. You can always bring the Queen to c2 to force an exchange of material. White's light squared bishop is still not helping much. Don't forget about the dominating center pawns. If you play on you'll definitely find Black up by material.

I will not say "always take towards center,” but it does help you most of the time.

Answer 2

[title "Radjabov- Gelfand, London 2013"]

[FEN ""]

1.Nf3 c5 2.c4 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e6 6.g3 Qb6 7.Nb3 Ne5 8.e4 Bb4 9.Qe2 d6 10.f4 Nc6 11.Be3 Bxc3+ 12.bxc3 Qc7 13.Bg2 e5 14.c5 b6!N 

With the novelty 14... b6! Gelfand shows all his class. According to the official website Gefand said

"Mostly great preparation, it's my strong quality. Any computer would show you c5"

The move 14...b6 goes with the move 13...e5. Gelfand likes "the concept because it cuts two pieces." The Bishop in g2 becomes very passive and the Knight in b3 is deprived from its natural square in d4.

From now on white has to come up with concrete ideas and find counterplay to maintain the game balance. Otherwise, it could be quickly worse as in the game.

Taking in b6

At first glance, taking in b6 looks weird. On general principle, we don't want

• to exchange an active pawn (c5) against a passive pawn (a7).
• to open the a line for free
• to give the square a6 for the bishop.

On a more concrete note, taking in b6 create two weaknesses d6 and b6. In particular, White can target the b6 pawn, like in the following variation

[title "After taking in b6"]
[FEN ""]

1. Nf3 c5 2.c4 Nc6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 e6 6. g3 Qb6 7. Nb3 Ne5 8. e4 Bb4 9. Qe2 d6 10. f4 Nc6 11. Be3 Bxc3+ 12. bxc3 Qc7 13. Bg2 e5 14. c5 b6 15. cxb6 axb6 16. O-O Ba6 17. c4 O-O (17... Nd4?? 18. Bxd4 exd4 19. e5! Bxc4 20. Qc2+-) 18. Rab1  

White got some counterplay and the position is very sharp.

Taking in d6

Radjabov reacted slowly and his position collapsed quickly. In the following game, Svane took on d6, and found some counterplay to equalize.

[title "Svane, Rasmus (2542) vs. Ragger, Markus (2695) Bundesliga 2015"]

[FEN ""]

1.Nf3 c5 2.c4 Nc6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 e6 6.g3 Qb6 7.Nb3 Ne5 8.e4 Bb4 9.Qe2 d6 10.f4 Nc6 11.Be3 Bxc3+ 12.bxc3 Qc7 13.Bg2 e5 14.c5 b6 15.cxd6 Qxd6 16.O-O O-O 17.Rfd1 Qe7 18.f5 Qa3 19.Bc1 Qa4 20.Bg5 Ba6 21.Qe3 Rfd8 22.Bxf6 gxf6 23.Qh6 Rxd1+ 24.Rxd1 Qxa2 25.Qxf6 Qxb3 26.Qg5+ Kh8 27.Qf6+ Kg8 28.Qg5+ Kh8 29.Qf6+ Kg8 ½-½

---

In summary, it's really hard to give a definite answer. However, I tend to agree with @Allure that taking in b6 was a good alternative. At least it's probably not worse than taking in d6 since after 14.c5, the move 14..b6! gave Black a very good position.