Been studying Eddie D's fine work on the Rb1 Exchange and something is puzzling me.

1 d4 Nf6 2 c4 g6 3 Nc3 d5 4 cd Nc5 5 e4 Nc3 6 bc c5 7 Nf3 Bg7 8 Rb1 00 9 Be2 Nc6 10 d5 Bxc3+ 11 Bd2 Bxd2+ 12 Qxd2 Na5 13 h4 Bg4 14 Ng5 Bxe2 15 Kxe2. All trendy theory of course.

Now 15...h6 16 Nf3 Kh7 17 Qc3 b6 18 Ng5+ Kg8 19 h5 hxg5 20 hxg6 fxg6 21 Rh8+ Kf7 22 Rh7+ Ke8 23 Qg7 Kd7 24 d6.

This is all following a famous game Chernin-Stohl, recently reprised in Vaisser-Vachier Lagrave. I know we're a long way from the opening, but really this sacrifice was the whole idea of 14 Ng5 and White is meant to be winning here, hence Black avoiding this line with 15...e6 as a rule.

But isn't Black just winning here? In these games he's always chosen to block the e7 pawn with his queen, which doesn't work because he has to play the rook to g8, and then the seventh rank is mined, so he is always falling for Rh8!, or Qxg8/a8, followed by d8=Q+, or some such trick. However if he goes 24...Re8, followed by developing with ..Qc7, ..Kc6-b7 and ..Nc6, then he just seems to be winning. That's what my computer says anyway, and I haven't managed to prove any different. But computers aren't always right - does White have some tactical idea that revives his attack? Anyone cleverer than me find anything - Glenn, you like a speculative piece offer, I noticed the other day. How about it?