Tafl, the difference in the plans in the games you listed from the original question is that Black is playing ...Nf8 before castling kingside.  If Black intends to castle first, how do you think White should play in response 8.Nf3 O-O?  (which is what I am guessing kevinludwig would like to know)

I'm sorry I didn't even notice that Black played Bd6 instead of Be7.  There was some game, I think between Belyavsky and Julio Granda Zuniga, which Robert Byrne analysed, showing ...Bd6 to be bad.  It was published several years ago in the NY Times, does anyone remember what I'm talking about, or was it just another dream?

The 6...Bd6 move is not given much attention in books, but I think it deserves study, since Black is aiming for an ideal attacking setup against kingside castling.  I know an expert that has used this plan to good effect, without knowledge of main line theory.  The interesting thing about this move is that the most direct methods to counter this setup seem to fail in pursuit of the advantage.  I think 11.h3 is a good waiting move, leaving castling to either side in the air.  If Black continues with his plan of ...Nf8-e6, White can castle queenside, when Black's setup is probably not ideal.  I think White has good attacking chances in the scenario Black plays 11...Nf8, and White responds with 12.O-O-O.  Here are some examples of attacking ideas for White in this scenario.

[Event "?"]
[Site "Dresden op"]
[Date "1994.??.??"]
[White "Paehtz,T"]
[Black "Mueller,Mi"]
[Round "?"]
[Result "1-0"]
[ECO "D31"]

1. d4 d5 2. c4 e6 3. Nc3 Be7 4. cxd5 exd5 5. Bf4
Nf6 6. e3 c6 7. Bd3 O-O 8. h3 Bd6 9. Nge2
Re8 10. Bg5 h6 11. Bh4 Nbd7 12. Qc2 Nf8 13. O-O-O
b5 14. Kb1 Be6 15. g4 Rc8 16. Rhg1 a6 17. g5
Nh5 18. f4 Bd7 19. Bf2 Kh8 20. Ng3 Nxg3 21. Rxg3
b4 22. gxh6 gxh6 23. Rdg1 Ne6 24. Qd1 Qf6 25. Na4
Rb8 26. Bc2 Nc7 27. Qh5 Re4 28. Rg6 Qxg6 29. Rxg6
Bf8 30. Rxh6+ Bxh6 31. Qxh6+ Kg8 32. Bxe4 dxe4 33. Qd6 b3 34. Qxc7 bxa2+ 35. Ka1 1-0

[Event "?"]
[Site "Budapest"]
[Date "1995.??.??"]
[White "Thallinger,Harald"]
[Black "Jamrich,Gyorgy"]
[Round "?"]
[Result "1-0"]
[ECO "E00"]

1. d4 Nf6 2. c4 e6 3. Nc3 d5 4. cxd5 exd5 5. Bg5
c6 6. e3 h6 7. Bh4 Bd6 8. Bd3 O-O 9. Nge2
Nbd7 10. Qc2 Re8 11. h3 Nf8 12. O-O-O Be6 13. g4
Be7 14. f4 N6h7 15. Bf2 Bh4 16. Bxh4 Qxh4 17. Rdg1
Bd7 18. Rg3 Rac8 19. Rhg1 g5 20. R1g2 c5 21. Bxh7+
Nxh7 22. f5 h5 23. Ng1 hxg4 24. hxg4 cxd4 25. exd4
Nf6 26. Nf3 Qh6 27. Rh2 Qg7 28. Ne5 b5 29. Rgh3
Re7 30. Qd1 b4 31. Qh1 Kf8 32. Rh8+ Ng8 33. R2h7
Qf6 34. Rh6 Qg7 35. f6 1-0

[Event "?"]
[Site "Slovak League"]
[Date "1995.??.??"]
[White "Lancz,Ondrej"]
[Black "Lukac,Roman"]
[Round "2"]
[Result "1-0"]
[ECO "E00"]

1. d4 Nf6 2. c4 e6 3. Nc3 d5 4. cxd5 exd5 5. Bg5
c6 6. e3 h6 7. Bh4 Bd6 8. Bd3 O-O 9. Nge2
Re8 10. Qc2 Nbd7 11. h3 Nf8 12. g4 Ne6 13. O-O-O
b5 14. Rhg1 Bb7 15. Bf5 Ng5 16. Bxg5 hxg5 17. h4
b4 18. hxg5 bxc3 19. gxf6 cxb2+ 20. Kb1 Qxf6 21. g5
Qd8 22. Rh1 Ba6 23. g6 Qf6 24. gxf7+ Kxf7 25. Rhg1
g5 26. Nc3 Rh8 27. e4 Bf4 28. exd5 Kg7 29. dxc6
Qxc6 30. Be4 Qd6 31. Bxa8 Rxa8 32. Rh1 Qg6 33. Qxg6+
Kxg6 34. Kxb2 Bb7 35. Rhe1 Rb8 36. Kc2 Bc8 37. d5
Bf5+ 38. Ne4 Rc8+ 39. Kd3 Be5 40. Rc1 Rd8 41. Rc6+
Kf7 42. Kc4 Bxe4 43. Rxe4 Bf4 44. Kc5 Bh2 45. d6
Bg1 46. Rc7+ Kf6 47. Kc6 Bxf2 48. d7 Bb6 49. Re8
1-0

If Black switches plans with 11...Qa5 (which is a better move against queenside castling), White can castle kingside and play for e4.  Here is a model game for White by Volkov:

[Event "?"]
[Site "Ekaterinburg RUS Cup"]
[Date "1997.??.??"]
[White "Volkov,Sergey"]
[Black "Kozlov,Oleg"]
[Round "6"]
[Result "1-0"]
[ECO "D35"]

1. d4 d5 2. c4 e6 3. Nc3 Nf6 4. cxd5 exd5 5. Bg5
c6 6. e3 Nbd7 7. Qc2 Bd6 8. Bd3 h6 9. Bh4
O-O 10. Nge2 Re8 11. h3 Qa5 12. Bg3 Bxg3 13. Nxg3
c5 14. O-O c4 15. Bf5 b5 16. e4 b4 17. Nxd5
Nxd5 18. exd5 Nb6 19. Bxc8 Raxc8 20. d6 g6 21. Ne4
Nd7 22. Rfe1 Kg7 23. a3 bxa3 24. Re3 a2 25. Qc3
Qd5 26. Nc5 Rxe3 27. fxe3 Qxd6 28. Rxa2 Qc7 29. Ra5
Kg8 30. Qa3 Ra8 31. Ne4 Nb6 32. Re5 Kg7 33. Re7
Qd8 34. Nd6 1-0

There are many subtleties in this system, so sometimes it is hard to decide which plan to commit to, especially when the opponent is playing such a principled system.  My first intuition was to play for e4 by preparing f3, but as you mentioned this is not so simple.  Looking at the game by Volkov led me to realize that perhaps the best strategy is to meet this direct strategy with a subtle waiting move, keeping options open against the next commital move.

Let me know what you think.  The expert I mentioned has given many class A players and experts a headache here, and I know he has beaten at least one IM with this system as Black, so it certainly has its poison.  I think this line has worked well for him, since many players are out of book and confused when their usual plans do not work as well.  Also I think another factor is that his general understanding of the game is much stronger than his knowledge of theory.  This setup really tests one's understanding of the QGD exchange.  It was delightful to see his games against stronger players, where he invented his own theory and got an excellent position!

Kevin,

I often play the same opening that you do, and the first game I would suggest you look at is Kasparov-Andersson, Belfort 1988.  I prefer the idea of playing 11.f3 here, to leave your options open.  I also get to play Bh4-f2 where it's surprisingly strong.  I think if you do a database check, you will find that Black doesn't usually chase the B in this exchange variation precisely because it can bounce back to a great square on f2.

I don't like g4 in this position because my main idea is to defend the d4 pawn and push e4 with an advantage in the center.  I haven't yet committed my king, but I plan to castle queenside. The move order allows me to transpose into Kasparov-Andersson (he castled kingside) or play the slightly more popular 0-0-0.  

Good luck, and let me know what you think!