Why Kf8 in that Carlsen - Bologan game? A priori it looks like white is likely to get the tempo back when he goes Nd2 - c4 and having the black king on f8 really isn't going to help if/when white plays an e5 sacrifice, or Nxd6, or Bxe7+ etc. Seems quite scary to me

Qd8 does seem a pretty natural move.

In general I do think it's a tiny bit optimistic to rule this out on general principles in via Rfb8, Ne8 - c7 etc, because doing that will give white some ideas. Maybe they'll not work vs some specific move orders, but there's definetly real potential there.

The objective of 12.a4 doesn't have to be to simplistically promote the a-pawn (as Neric said somewhere), I think it's just the best place for it, since the b-pawn does well on b3, countering Nc4, in fact it's the same set-up as in the fianchetto variation with Rb1, but in my opinion Black has less pressure here.

Neric wrote on 08/20/12 at 09:12:40:


In both cases, I am struggling to see Black's compensation if White plays 16.Bc3. Without the bishops and without any direct threats against White's queenside pawns, it does not look like Black has the standard Benko compensation.

Neric wrote on 08/20/12 at 09:12:40:


Here I agree, this looks more like a standard Benko. For this reason, I think 12...Qb6 looks better than 12...Qa5.

Neric wrote on 08/20/12 at 09:12:40:


Again I agree. I was surprised Bologan went for ...Qd8 instead of ...Kf8 in reply to Bg5.

Besides, if Black is ready to play the 12.Re1 Ra6 line, I think 12.a4 Ra6 also looks like a reasonable response for Black (as TN pointed out, this may be why Carlsen chose 12.Qe2 first). Now 13.Re1 takes the game back to 12.Re1 Ra6 territory, while 13.Qe2 Qa8 only seems to serve Black, for several reasons:
- White will find it hard to play Nb5 as a4 will be hanging.
- Moving away from the d-file encourages Black to carry on with the ...e6 plan.
- The queen could end up exposed on the e-file after ...e6, dxe6 fxe6, and a further ...d5, exd5, exd5.

I don't have Kaufman; does he cover 12.a4 Ra6?

I remember reading about the idea of a4 and blocking the queenside long time ago. Because I had trouble playing against the Benko I tried it in a blitz game against an IM who has played it many times. I really got a better position, but blundered in time trouble, which had nothing to do with the opening. Just an anecdote, but I have really liked the plan ever since.

Let us start with the Kaufman suggestion which I think is bad. The problem is, that regardless what I suggest, the computer will always add 1 pawn-unit to the white position and say that white is better. 

Anyways, his mainline goes:
12. ....Qa5
13. Bd2 Qa6
14. Rhe1 Rfc8
15. Nb5 c4
16. Bc3!

This is dubious for multiple reasons.
1. If black plays Qa5, he should follow up with Rfb8. The move Qa6 is not forced and looks like a simple waste of tempo. It also sets up black for the fork on c7 and that cannot be the best solution.

2. c5-c4 should only be played if it leads to something concrete. If you play this move too early, you allow white to plant a knight on c6 (via d4) or open the b-file for future operations.

My suggested improvements over Kaufmans mainline would be: 
13.... Rfb8 14.Nb5 Qd8 (the queen prevents e5 breaks) with Nc7 to follow, not Nb6!
and
15. ... Ne8 with the same idea

His alternative looks dubious to me also:
12.......Qb6
13.Qe2 Rfb8
14. Bd2 Qb3
15. Rhb1 Nb6
16. Ra3 Qc4
17. Qe1 "and white is consolidating his extra pawn".

There are absolutely pointless moves in this line. The whole black play starting with Qb3 looks simply artificial and it discoordinates the black pieces. I am not even talking about his assessment as consolidation of the extra pawn means nothing in the Benko. Like in the Berlin Defence, it doesn't matter to consolidate the advantage, it is all about realizing it and that is impossible if you can't break through.

The "correct" sequence in the Qb6-line would simply be 14...Rfb8 15. Bd2 Ne8 16. Rhb1 Nc7 and Qa6 next. While computers go crazy about the exchange of queens "consolidating the extra pawn", every Benko-player knows that it is thematic. It is the positional "trap" that makes the Benko so attractive.

Going back to the Carlsen game, we notice, that he improved upon Bd2 by playing Nb5 immediately. This doesn't change the black plan as he can go for the exchange by Ne8-c7 and if Bg5 then Kf8. There is still a full game ahead, but black should have the typical Benko compensation as long as he doesn't touch his pawn structure.

Thanks Stigma.

Some of you will recall that the line 1.d4 Nf6 2.c4 c5 3.d5 b5 4.cxb5 a6 5.bxa6 g6 6.Nc3 Bxa6 7.e4 Bxf1 8.Kxf1 d6 9.g3 Bg7 10.Kg2 0-0 11.Nf3 Nbd7 12.a4 was recommended by Kaufman in 'The Kaufman Repertoire for Black and White'. Since then the moves 12.a4 and 12.Qe2 have become virtually the main line of the Benko, but unfortunately none of my other Benko sources (e.g. Pedersen, Pinski, Alterman) even mention these moves. 

Carlsen has played 12.Qe2 twice in the last two months, presumably with the idea of preventing 12.a4 Ra6. Against Bologan he quickly gained an edge:



Instead 12...Qa5 might be a better try; although this was only a blitz game, Black never seemed to be in danger even if his winning chances are minimal in the resulting simplified position.



Returning to 12.a4, the continuation 12...Qb6 13.Qe2 Rfb8 would transpose to Carlsen-Bologan. Instead 12...Qa5 seems more critical, when 13.Ra3 as played in Brunello-Caruana is tempting but I think if Black plays 13...Rfb8 and only plays ...Nb6 when White moves his queen (so that White can't break with e5 so rapidly due to d5 being in need of defence), his position is fine. 

The move recommended by Kaufman is 13.Bd2, when at first I thought 13...Rfb8 might be an improvement over the only move given by Kaufman (13...Qa6), but the following game convinced me otherwise:



So it seems that 12.Qe2 Qa5 is okay for Black, but after 12.a4 White obtains an advantage with fairly obvious moves, as White will achieve the ideal Anti-Benko setup of an unchallenged White knight on b5 and pawns on b3 and a4.

Edit: Does anyone know how to fix the PGN problem?