For a moment, I thought I was participating in a demonstration of the futilitarian principle: The key move, 33. ♘h8, is not hard to see, since it wins an exchange in short order. What is not so easy to see is the way to put Black away, since the exchange is not necessarily sufficient to win.

(1) After 33. ♘h8!, ♗xf2†; 34. ♔h1, ♔xh8; 35. ♕xf7, White has the advantage, but not a very satisfying one.

I therefore examined other ideas until I'd ruled them out.

(2) The most obvious move, of course, is 33. fxe3?, taking a bishop and bringing the open f-file to bear against Black's pinned rook. Unfortunately, this fails prosaically, thanks to the loose knight on g6: 33. ...♕xe3†; 34. ♔h1/♔g2, ♕e4† , when Black finishes a pawn ahead and with all pieces defended.

(3) Tempting but pointless is the attempt to transfer the knight to a better square with 33. ♘e7†?!, which after the forced 33. ...♔f8 only improves Black's position by relieving the pin on his rook. Since there is no effective follow-up, this was quickly set aside.

(4) No analysis would be complete without a look at the queen sac 33. ♕xf7†?, with the thought of capturing the bishop on e3 with discovered check. But this loses miserably, since Black gets out of check by taking the knight, leaving White with rook against queen.

(5) More intriguing was the try 33. ♕e6?. This looks strong, since it threatens to win the pinned rook with ♕e8† — and in fact does.

However, the move also loses, spectacularly so in one variation: After 33. ...♗xf2†, White must move the king to either of two squares.

(5.1) 34. ♔g2, ♗xg3†!; 35. ♔xg3 (<declining the bishop with ♔h3/♔h1/♔g1 leads to mate in one; declining with ♖f2 leads to mate in two>), ♕d3†; 36. ♔h4/♔g2, ♕xf1/♕xf1† . Black is the exchange ahead and has an unstoppable attack.

(5.2) 34. ♔h1, ♕xd4!; 35. ♕e8†, ♔h7; 36. ♕xf7??, ♕e4†; 37. ♕f3 ▢, ♕xf3#.

There being no other apparently relevant lines, this brings us back full circle to line (1), and demands that we find a way to win with it.

Continuing from above, we find that Black must save his twice-menaced bishop on f2. What I failed to see on the first examination was that 35. ...♗xd4? lost a piece after 36. ♕xh5†, ♔g8; 37. ♖d1!, ♕ moves; 38. ♕d5† with a double attack.

Black must therefore consider 35. ...♗e3. For some time, I studied this position, seeking a way to bring in the rook (say via a1 or b1) and forcing mate. Unfortunately, Black meets this with ...♔f8 and appears to escape.

Finally, I concluded that the most reliable way to bring home the point might be the quiet 36. d5, when White wins slowly, safely and prosaically.

33.Nh8 Bxf2+
34.Kh1 Kxh8
35.Qxf7 Bxd4
36.Qxh5+ Kg8
37.Qf7+ Kh8
38.Qxc7 (taking back the lost pawns and having the exchange for the ending game)

click for larger view

30.Qxe2 is answered by 30...Qxe2 31.Rxe2 Rxe7 regaining the piece. Geller was evidently trying to get more out of the position.

click for larger view

... wouldn't you be tempted to play just one more move? Say 39...Kh8

... and hope and pray that white snaffles your bishop with 40. Rxd4?

... whereupon you slowly reach over the board and nudge your queen forwards one square ....

click for larger view

I think you'll find that's mate. Very sorry.

33.Nh8 Bxf2+
34.Kh1!

Now black has trouble defending the rook, his queen can't do this.

34...Kxh8
35.Qxf7

and white has won the exchange.

Time to check.

----------

Yes, I was right.

<33. Nh8! Kxh8>

[33...Bxf2+ 34. Kh1 Kxh8 35. Qxf7 Bxd4 36. Qxh5+ Qh6 (36...Kg8 37. Rd1! and depending on what Black plays, White wins the B via 38. Qd5+ or 38. Qe8+ Kh7 39. Qe4+) 37. Qe8+ Kh7 38. Qe4+ and 39. Qxd4 ; 33...Qxd4? 34. Qxf7+ and 35. fxe3 ]

<34. Qxf7 Bxd4 35. Qxc7> White has the more advantageous endgame as he is up the exchange.

[35. Qxh5+?! Kg8 (35…Qh6?? 36. Qe8+ Kh7 37. Qe4+ and 38. Qxd4 ) and here Black has survival chances due to his connected passed pawns]

White to play (33.?) "Difficult", even material.

Losing a pawn is 33. fxe3? Qxe3+ 34. Kg2 Qe4+ 35. Kg1 Qxg6.

That reminds me of a game I played about 30 years ago against a older guy (probably not too much older than I am now, gulp) named Ben Greenstein, who was rated far below me. I think I was 2000-something and he was a 1600. A fish. Easy pickings. He had played the Stonewall Attack. I was unimpressed and castled prematurely, whereupon he launched a pawn storm and got a big attack. I spent a lot of time figuring out how to defend, but then he castled queenside and suddenly <I> got a raging attack. Alas, in time pressure I misplayed what should have been a winning attack, playing three consecutive weak moves and allowing him to relaunch his attack.

After we reached the time control, I disgustedly viewed the wreckage of my position. I was three pawns down and completely crushed. I reached across the table intending to shake his hand to resign. I felt sick. How could I blow such a position? And a win would probably have given me the club championship, which I could now forget about. But instead of reaching his hand, my hand, almost as if it had a mind of his own, grabbed my c-pawn and played ...c3 attacking his queen on d2 and the pawn on b2 in front of his king. It was a vain attacking gesture, which he could easily have rebuffed. But he played an inferior response, after which I felt a glimmer of hope. I played another attacking move, and he again responded with an inferior move. He wasn't taking my counterattack seriously, and indeed it wouldn't have been serious if he had taken a few minutes to analyze it and play the best moves. I played a third attacking move, he yet again responded weakly, and suddenly I had a winning attack again!