I was going over a game (#16) on Leisebein's site in which he employs his favorite Harikiri-Variante (Seidal-Hall Attack). 

Leisebein - Rohde 2000
1.d4 Nf6 2.Nc3 d5 3.e4 dxe4 4.f3 exf3 5.Nxf3 Bg4 6.h3 Bxf3 7.Qxf3 c6 8.g4 Qxd4 9.Be3 Qd6

Leisebein gives 9... Qd6 an exclamation mark and comments on it. My German is terrible despite having spent 4 years there as a young soldier some 30+ years ago. I think he says that 9... Qd6! is the one move that inhibits White's initiative. If someone wants to look it up and provide an exact translation, that would be great. Anyway, 9... Qd6 stops White from playing an immediate 0-0-0 and continuing on with his rapid-fire development.

10.g5 Nfd7

Again, Leisebein gives an exclamation mark because 10... Nd5 would allow 11.0-0-0 and 10... Nfd7 doesn't.

11.Bd2

When I saw this move I cringed and thought, 'This can't be good. If White has to move this bishop again to get castled, then he has done something wrong or the whole line may not be any good.'

I believe I have found something better. It's the tactical shot 11.Nb5!

a) 11... cxb5 12.Qxb7 e6 (12... Qg3+ 13.Kd2 e6 14.Qxa8 Qe5 with a clear advantage for White) 13.Bxb5! (Forsaking the rook in favor of a bind.) 13... Qd5 14.Qxd5 exd5 15.0-0-0 a6 16.Ba4 and now

 a1) 16... d4 17.Bxd4 Be7 18.Rhe1 Kd8 19.h4 f6 20.Re6 fxg5 21.Be5 Kc8 22.Rxe7 Nxe5 23.Rxe5 gxh4

 a2) 16... Kd8 17.Rxd5 Kc7 18.Rf1 f6 19.Bf4+ Kc8 20.Rf3 Nc5 21.Rc3 Kb7 22.gxf6 g5 23.Rdxc5 Bxc5 24.Bxb8 Raxb8 25.Rxc5

 a3) 16... h6 17.Rhe1 Kd8 17.Rxd5 hxg5 18.Bxd7 Nxd7 20.Bb6+ Kc8 21.Re8+ Kb7 22.Rxd7+ Kc6 23.Rxa8 Kxd7 24.Kd1 Rh6 25.Rxf8 Rxb6 26.Rxf7+ Ke8 27.Rxg7 Rxb2 28.Rxg5 Rxa2

In each case White comes out with a clear edge.

b) 11... Qb4+ 12.c3 Qxb2 13.Rd1 cxb5 14.Qxb7 f5 15.Qxa8 Qxc3+ 16.Kf2 f4 17.Bxf4 e5 18.Qe4 Bc5+ 19.Kg2 0-0 20.Bg3 a6 21.Bd3 g6 22.Qd5+ Kg7 23.Rhe1 Bd4 24.Bb5! and White is winning.

The engine I used is Rybka 3.0. If 9... Qd6 is truely busted then Black shouldn't grab the second pawn with 8... Qxd4 because 9... Qb4 allows White to work up a fierce attack.

Hello Stefan, What do you think of this line I used against Sawyer? We were testing an idea I had. Good Chess! Keith

[Event "Training Game #1"]
[Site "Email"]
[Date "2007.03.26"]
[White "Hayward, Keith"]
[Black "Sawyer, Tim"]
[Result "1/2-1/2"]
[ECO "D00"]
1. d4 d5 2. e4 dxe4 3. Nc3 Nf6 4. f3 exf3 5. Nxf3 c6 6. Bc4 Bf5 7. O-O e6 8. Ng5 Bg6 9. Ne2 Be7 10. Nf4 Bf5 11. Bxe6 Bxe6 12. Nfxe6 fxe6 13. Nxe6 Qd7 14. Nxg7+ Kd8 15. Nf5 Rg8 16. Bf4 Bf8 17. Qd3 Qf7 18. Rae1 Nbd7 19. Re2 a5 20. c4 Ra6 21. Rfe1 a4 22. Re6 Ra5 23. Bg3 Rg4 24. Qf3 Re4 25. R6xe4 Nxe4 26. Bh4+ Kc7 27. c5 Ndxc5 28. dxc5 Bxc5+ 29. Kh1 Bb4 30. Rf1 1/2-1/2

I would really have been surprised if you had known Euwe's recommendation. My source - alas it's lost - is an article from the Losbladige (yes, in Dutch) from the 1950's.

Hmpf, if DF had posted his riddle a few hours earlier I would have been the first to give the answer.  

Challenging White's centre with ...c7-c5 is always an important option. If Black can deny e5 to White's King's Knight the attacking plan is seriously hampered. That knight is worse on d4 as the Sicilian piecesacs don't fit in White's plan. The idea ...c7-c5 can be tried at various stages. It is neglected in quite a few monographs on the BDG.
Fortunately Sawyer does give several examples of 6.Bg5 c5. As always they are unreliable.

1.e4 e6 2.d4 d5 3.Nc3 Nf6 4.Bg5 dxe4 5.f3 exf3 6.Nxf3 c5 7.Bxf6 (Euwe's recommendation 7.d5 deserves attention) Qxf6 (gxf6 8.d5 and e5 9.Bc4 Nbd7 10.d6 and here we have again one of those wacky Diemer games; White seems to be OK here) 8.Bb5+ Bd7 (Nd7 9.0-0) 9.0-0 cxd4 10.Ne4 Qd8 (better than Qf4 11.Bxd7+ Nxd7 12.Nfg5 Diemer) 11.Ne5 Bxb5 12.Nxf7 Qd5 (Qb6!? but Qd7 13.Qh5) and unless I miss something White does not obtain full compensation.

Is it 8.Bb5 c6 9.0-0, with the idea of 9...cb5 10.Ne5 Qh4 11.Nb5 Na6 12.Rf7?

Yes MNb, to take the b2 pawn is very risky, but it's critical to prove that White is at least ok, and I managed to show that it's clearly more than ok. You give 11..0-0 but here too, it's castling in the fire. 12.Ne4! is even better than the immediate 12.Qh4 here, with enough pressure for the pawn. Same with 13.Ne4 Nxe4 14.Bxe7 Qxe7 15.Qxe4 Nf6, Black may be ok but White has enough play for the pawn. So 11..0-0 or 13..Nxe4 may be playable, but that's at least = for White.

It's happened to my big schliemann's file in chesspositiontrainer too (both files about the same month quite recently), but that was not the same error. The file was very big too, with originals and very deep analysis So much days of work lost...