"I am not sure what you mean here."
If White can prove a clear edge after ...h6 and ...b6 with Bd3 and Rc1 it's likely that White also can after ...b6 without ...h6 with Bd3 and Rc1. If Marin thinks Black's position playable with the pawn on h6 perhaps it's also playable with the pawn on h7. Perhaps.
And perhaps Black can even try to transpose by answering Rc1 with h6. That depends on how strong Bf4 and Bxf6 are.

I suppose that Marin against Ionescu after 8.Qc2 (iso 8.Rc1) would have played ...c5 immediately. The other move order surprises me more.

1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Nf3 Nbd7 5.cxd5 exd5 6.Bg5 h6 7.Bh4 Be7 8.e3 0-0
How would Marin have met 6.Bf4 ? A queenside fianchetto looks highly suspicious to me.

Do Marin and Stoica think Black's setup playable after 11.Rc1 (iso Sokolov's 11.Qc2 c5 12.Rad1) ? That would be encouraging. See, the first Kasparov-Beljavsky game from the 1983 match is an example of Black rather wanting the pawn on h7. After 14.Ng4 (iso 15.Ng4) Black could have played Bh4 15.g3 f5. The same for the topical 5th game. With the bishop on g5 iso h4 Black could have played 12...Ne4 iso 13...Ne8. Not that I claim this will save Black's case (for not playing ...h6). Right now I lack the time and energy for the necessary systematical research. Maybe a few months later.

LeeRoth wrote on 07/16/16 at 23:42:04:


Cox (2011) noted that after 8...ed 9. Bd3 Bb7 10. 0-0 Nbd7, Ivan Sokolov (2009) considered 11. Qc2 c5 12. Rad1 as "the way to make White's advantage clear."  I'm aware of a Yearbook article by Marin and Stoica (2015) which evidently focused on the version with 11. Rc1 and the possibility of reaching it by a move order with early ...Nbd7.

MNb wrote on 07/15/16 at 12:27:25:


Agreed.  Without ..h6 in, this 7.cxd5 Nxd5 line isn't much fun for Black.   

After 8.Bxe7 Qxe7 9.Nxd5 exd5, I might prefer 10.Bd3 Be6 11.Qc2 when 11..h6 12.Rc1 Rc8 13.O-O c5 14.dxc5 bxc5 15.Nd4 has to be good for White.   

10.Rc1 is good too and after 10..Be6 11.Bd3 Rc8 12.O-O c5 then I looked at 13.e4 right away and think White is better there too.   

Interestingly, the only book recommendation I found was in Play 1.d4, where Richard Palliser gives 11.Qa4 c5 12.Qa3 Rc8 13.Bb5 a6 14.Bd3 Ra7 15.dxc5 bxc5 16.b3 a5 17.O-O Na6 18.Bxa6 Rxa6 19.Rc3 Kf8 20.Rfc1 c4 21.Qb2 Rac6 22.Nd4, following T.Petrosian-Trikaliotis, Siegen ol 1970.       

A good rule of thumb in the opening is 'When they pin you with Bg5/...Bg4, ...h6/h3 is almost always handy, because if they retreat the bishop, they lose the option of moving back along the diagonal, and you gain the option of ...g5/g4 later'. Such an argument works well in the QGD, after of course you play ...Be7 or ...c6 to make sure you aren't losing the d5-pawn. An exception is the Cambridge Springs, where you want their bishop on g5 for certain ...Ne4 tricks. 

The other exception is when they can meet ...h6 with Bf4 to move order Black into a worse version of the Exchange QGD. That only tends to be a problem when Black has already played ...c6 though. 

This is all purely from memory but it serves me well in OTB games.

Quote:


Even further back, Edmar Mednis wrote in a book from the 1990s,  "As recently as ten years ago, Blacks tended to play it on move 6, i.e. after 5...0-0 6. e3.  Lately the overwhelming number of GMs play it immediately."

I was also curious if it makes much difference if you play ...h6 before castling in the Bxf6 line. This line looks a bit scary:

1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Be7 5. Nf3 h6 6. Bxf6 Bxf6 7. Qd2 dxc4 8. e4 c5 9. d5 exd5 10. e5

But if Black castles first, this is a possible line:

1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. Bg5 Be7 5. Nf3 O-O 6. e3 h6 7. Bxf6 Bxf6 8.Qd2 Nc6 (or 8...dxc4 or 8...a6), and White has played e3 instead of e4 in one go as in the previous line. 

Is 6.e3 best in this line though? In a 1995 game between Kramnik and Short, Kramnik played 6.Qc2 h6 7.Bxf6 (Sadler gives this as !) 7...Bxf6 8.Rd1 c5, and he seems to think Black is fine here. 

I was just wondering if currently it matters if Black castles before playing ...h6.

5...h6 allows Black to play 6.Bh4 O-O 7.e3 b6 8.cxd5 Nxd5 as 5...O-O 6.e3 b6 7.cxd5 Nxd5 8.Bxe7 Qxe7 9.Nxd5 exd5 10.Rc1 Be6 11.Bd3 c5 loses a pawn due to 12.dxc5 bxc5 13.Qc2. So Black should play 7...exd5 instead, which is much less solid.
However I always have been intrigued by three remarks of Sadler and Cox.

Cox: "it's hard to say that the pawn recapture (8...exd5 with the pawn on h6 - MNb) is bad. Everyone quotes one of the Kasparov-Beliavsky 1983 candidates games ..... it might be well suit a tough-minded player who prefers to avoid theory. It was, after all, good enough for Petrosian on several occasions."

Sadler: "The light-squared bishop is not on the c8-h3 diagonal anymore. This means that the light squares around the black king are not covered b this bishop. Thus f5 is available to a white piece, while f7 is less protected than if the bishop were on e6. Thus consider the situation after White plays a plan with Bb1 and Qd3 threatening mate on h7. The natural, indeed only, defence is g7-g6. If the h-pawn were still on h7 then the softening move h2-h4-h5 would have no effect."

Sadler: "The move h7-h6, while giving Black some room on the kingside, does weaken the kingside light squares and the g6-squares in particular. Consequently, for example, a knight that comes into e5 cannot be driven away by ...f7-6 as the knight can then simply hop into g6.

Finally there is the line 5...h6 6.Bh4 O-O 7.e3 b6 8.Bd3 Bb7 9.O-O Nbd7 10.Bg3!? which is not possible after 5...b6 etc.

Apparently in several lines Black rather would have the pawn on h7.

Words are one thing, concrete lines something else. I'm not sure if White can punish Black for not playing ...h6. The critical line might be 5...O-O 6.e3 b6 7.cxd5 (7.Bd3 allows Nbd7, which does reasonably well in practice) exd5 8.Bd3 Bb7 (Be6 is said to be better, but results are disastrous) 9.O-O. Comparing with the variation with ...h6 and Bh4 included might yield some interesting insights.
Still being able to play ...Nxd5 is the decisive factor.