After CraigB's move

10.Qxc5 Bxe6 11.Qf8+ Kh7 12.Nxe6 Ra2+ 13.Kb6 Rb2+ 14.Kc6 Rc2+ 15.Kd6 Rd2+ 16.Ke7 Rd7+ 17.Kxd7 Kg6 18.Qe8+ Kh7 19.Qf7 f4 20.Qxg7#

Craig, you are correct- I only briefy scanned Alena's V3- the line I found is 5.....Rd4+, not 5.....Ra2. I just scanned down to see 11....Ra7 and assumed it came from a4 and not a2. In that line, white cannot take at c5 after the check since black takes the knight at f4 and drawing with the e-pawn.

The main idea after Q:c5 is ...B:e6 11. Qf8+ Kh7 12. N:e6 Kg6 13. Qe8+ Kh7 14. Qf7 with spite checks interspersed at Black's discretion.

Yancey_Ward said: Yes, V3 is the key variation, and black's 11th move in your version of it is the move I think Krug overlooked. After a couple of days trying to find a winning line for white, I gave up and put that position before 11.....Ra7 into several different on-line chess engines, and none of them could find more than a pawn edge for white. It might be a winning position, but I just don't see how.

v3 can be improved with 10. Q:c5.

Yes, V3 is the key variation, and black's 11th move in your version of it is the move I think Krug overlooked. After a couple of days trying to find a winning line for white, I gave up and put that position before 11.....Ra7 into several different on-line chess engines, and none of them could find more than a pawn edge for white. It might be a winning position, but I just don't see how.

Yancey, I couldn't post the solution earlier because the site wasn't working properly. I think you are right that the puzzle has got a flaw. I worked it out and got three variations.

V-1

1.Qf7+ Rxf7 2.gxf7+ Kf8 3.Nfe6+ Ke7 4.f8=Q+ Kd7 5.Qf7+ Kd6 6.Qxc7+ Kd5 7.Qc6+ Ke5 8.Nf7#

V-2

1.Qf7+ Kh8 2.Qxf8+ Bg8 3.Nf7#

V-3

1.Qf7+ Rxf7 2.gxf7+ Kh8 3.f8=Q+ Bg8 4.Nde6 Bd6+ 5.Ka4 Ra2+ 6.Kb5 Rb2+ 7.Ka6 Ra2+ 8.Kb7 Rb2+ 9.Kxa7 Bc5+ 10.Ka8 Ra2+ 11.Kb8 Ra7 12.Qxc5 Bxe6 13.Nxe2 Ra2 14.Nf4 Bf7 15.Qc8+ Kh7 16.Qxf5+ Kg8 17.Qc8+ Kh7 18.Qd7 Kg8 19.Qd8+ Kh7 20.Qf8 Ra4 21.Qxf7 Rxf4 It's a draw.