From Keres-Boleslavsky Zurich 53
Here Boleslavsky came up with an interesting idea. (He had actually been planning it moves earlier.)
9. ... Qxb2
The idea was to meet 10.Nxc7+ Kd8 11.Nxa8
With 11. ... d5!
In this position, Bb4+ is a major threat. One idea would be Bb4+ Nd2 Bxd2+ Qxd2 Qxa1+.
Bronstein's analysis from Zurich 53 continues with
12.Rc1
Now the rook isn't hanging on a1 after the Bb4+/tradeout on d2.
12. ... Bb4+ 13.Nd2
13. ... Nc5
With a threat of Ne4 And making way for the c8 bishop.
14.Rc2
14. ... Qe5 15.e3
And here 15. ... Bf5
Black wins the exchange back and will try to round up the stray knight. Note 16.Rc1 can be met by Ne4 (Rc2 Nc5) or 16. ... d4! (Computer).
I thought that this was a very creative variation by Boleslavsky.
Unfortunately, it has a major flaw in it.
10.Rb1! (as played by Keres) refutes the whole thing!
The rook is no longer awkard, and the crucial check on b4 is no longer availabe. For the continuation, see the rest of the game (below). I just thought this was a very nice idea even if it didn't work.
Bonus Question:
If black had played 5. ... g6, how would white respond?