I propose the following opening:

1. d4 Nf6, 2. c4 g6, 3. Nc3 d5 (Gruenfeld), 4. cd N:d5, 5. e4 N:c3, 6.bc

(Main system)

and now , after 6. ... Bg7 or 6. ... c5, I play 7. Bg5  .

This is a novelty, and was first played in a correspondance game, Championship of GDR, White: F. Brock (myself), Black: R. Tischbierek .

The game ended in a draw. (R. Tischbierek (GM) later won the Championship, I ended up as second . )

After this game I played the opening many times, mostly successful. A few times I lost (not because of the opening) . Some lines are very sharp, one mistake of either side canlead to a fast end... 

I would like to discuss this (still unexplored!)  line with you.

I start with a few ideas:

(I.) 7. Bg5 cannot refuted directly: If 6.... c5, 7. Bg5 Bg7,

(or, which is the same, 6. ... Bg7, 7. Bg5 c5),  

then white playes 8. Rc1, and the pawn d4 is artificially defended.

The idea of Bg5 is to advance d4-d5, and hinder the pawn break e7-e6, which is a common motive for counterplay, for instance in the variant with 6. Be3 (the variant preferred by Kramnik). 

Based on the spacial advantage (after d4-d5 ) white wishes, depending on the circumstances, to enter a favorable endgame (see number (II) below), or to start an attack on the king side , which involves such maneuvres as h4-h5-h:g6 , Qd2, and then Bg5-h6, or instead, after Bd3 and 0-0, with f2-f4-f5. In a few cases it is possible to increase the spacial advantage with d4-d5 and c3-c4, and then play on both wings .    

All these ideas occured in my games...

(II) Let me first look at the typical Gruenfeld endgame which arises after 6. ... c5, 7. Bg5 Bg7, 8. Rc1 Qa5 , 9. Qd2 cd, 10. cd Q:d2+, 11. K:d2.

Now Black has for instance the following two options:

(a) 11. ... Nc6, 12. d5 Nd4, 13. Ne2 (also 13. Nf3 is to be checked) and I would slightly prefer white (my score is positive, but this needs to be explored further!).

(b) 11. ... 0-0 , 12. d5 f5, 13. f3 += .

(III) Black can also avoid an early endgame: 6. ... c5, 7. Bg5 Bg7, 8. Rc1 and now

for instance 8. ... h6 , 9. Bh4 (9. Be3 Qa5 , 10. Qd2 cd leads to the afore-mentioned "Kramnik -variant" with Be3, where the additional move h7-h6 is rather good for Black!)

9. ... 0-0, 10. d5 unclear, or  9. ...  cd , 10.  cd  0-0 ,  11.  d5  f5! , also unclear.

In the last two variations, white can easily fall behind in development ... Still my score is balanced here.   

I must say that I never checked the whole system  with a computer program (the last time when I analysed it in depth was 1989 !!!) , nor

did I see a refutation published somewhere ...

If you have any suggestions, please let me know.

Mandelshtam