< He actually became a stronger player when he retired!>

But now -- after analyzing it, and seeing the direct attacking logic of the move -- I've changed my mind. I think most strong players, IM or better, would find it in a puzzle setting. As would a few weaker ones.

Over the board, mid-game, it's a bit harder. And harder still when you're trying to pick yourself up after having your 'forced win' snatched away. Shock is a big factor.

I suppose engines do it, to a degree, although hashtables store past analysis. But they're not burdened by emotional attachment to a plan, or thrown by the shock when the game takes an unexpected turn.

It's one reason I prefer to play OTB chess -- the margin of human fallibility, the poker bluff, the crazy swindles. Of course I've thrown away plenty of won games, but I've rescued a few lost ones too.

Ideally, one should be able to see the board afresh each move. It's often the case in tournaments that a passing kibitzer will see, at a glance, a winning move which neither player has seen. The same goes for my comment about ...Qh7 being easier to find if set as a puzzle.

But pretty much all combinations are easier in puzzle form. And maybe seeing the board afresh is dangerously narrow or low-ply ... maybe the longterm view is also important.

So now we want to be able to see the game afresh, without forgetting our plans and evaluations? Maybe the two are pulling in different directions. Maybe they're mutually exclusive.

Or maybe Kasparov knows the answer, but reveals it only during the 2nd year of intensive and expensive lessons, and neither Carlsen nor Naka lasted the course...

But I think the *very* best human players find these moves. I saw a video recently where Kasparov annotated a win against Karpov: in the middle of a sacrificial attack he pauses to make a quiet King move. Finding it seemed to give him extra pleasure (and the main reason was so that he could offer to sac another Rook without being checked).

Yes, they can make mistakes. Even the best won't find a move like ...Qh7 every time. But they sometimes do.

Speaking of engine saves, I dimly remember a nice game Nigel Short played in Gibraltar about five years ago. It was shortlisted for the best game prize, but didn't win.

The opponent was a strong woman player, possibly from Ukraine, and the game ended with an unusual sacrificial mating combination.

Later, with an engine, Short found a hole in his plan -- his opponent had what he called a 'ridiculous' saving move, the sort that only an engine finds. I think it might have been ...Rh7, ironically.

Must try to dig it out. I remember it as being more counter-intuitive than ...Qh7 is here, where much of the difficulty is recovering from the shock of still having to play on.

< Black has spent the last few moves throwing his pieces at the enemy king and he now has to adapt to a very different situation where the nature of play is changed> I reckon he *over*-adapted. As I said earlier, the combination is still alive and flourishing after 24.Rf4 ... he needs to play ...g5 while maintaining the mate threat, and only ...Qh7 fits.

Complex and clever, but not a patch on this in terms of drama.

A few moves were suggested in the kibitzing as saves for Zhukova. I think the 'computer move' was probably 22...Qh6.

click for larger view

The actual finish was 22...Qe5 23.Rh4+ Kg8 24.Bf4 d3 25.Bxe5 dxe2 26.Nf6+ 1-0. Mate is forced.

I knew it had some vague similarity to this. So the Queen is just one square out. I suppose it's counter-intuitive because it seems to permit a discovered attack on the Queen - but nothing violent works for White, so he has to retreat with 23.Bd2 or Bc1.

But Black should have won anyway. The key is 24...Qh7.

If you scroll back to the start of the page, skip over four short posts, then two by me ... you reach an analysis starting "It gets better".

It contains all you'll ever need to know.

The kibbitzing on the game was especially good. <Then White finds and plays the 'miracle save' 23.Qh5!! ... although two exclams is a tad generous for a move that is (a) forced and (b) losing.> I love it!

2) -4.31 (33 ply) 24...f6 25.Nd1 Qh2+ 26.Kf1 Qh1+ 27.Ke2 Qxg2+ 28.N1f2 b6 29.Rf3 gxf2 30.Rxf2 Qh3 31.Kd2 g5 32.a4 Kb7 33.Ree2 Rh8 34.Re1 Qg4 35.Ree2 Kc7 36.Rh2 Rh3 37.a5 Qh4 38.Rxh3 Qxh3 39.Nf2 Qh4 40.Nd3 Kd6 41.c3 Qe4 42.axb6 axb6 43.Rf2