I’m going to have a couple posts this week about interesting endgame positions I’ve run into recently. One of them is from a blitz (8-minute) game I played last night against Gjon Feinstein, and the other is from a tournament game between two 1400-level players that a reader of this blog sent to me.
One thing I’ve noticed is that when you go over a game with a computer, very often the last position where one player can still save a draw often turns out to have some very subtle tactics, because it is just balanced on the razor’s edge between a win and draw. Unfortunately, we humans very seldom sense when this last chance for salvation occurs. So in a certain sense the computer analysis is moot… but it nevertheless teaches us some important lessons for next time!
Above is the PGN viewer for the game that I played against Gjon last night. The interesting thing about it was that at no time did either Gjon or I have the sense that Black was busted. It was just a dogfight, with chances for both sides … until the very last move. I played 40. b5, and then Gjon’s flag fell. At that point I said, “But you know, I think I’m winning anyway.” We did some analysis and it looked as if I was right. But then the question arises: What did Black do wrong?
Here, thanks to Rybka, is the answer.
Position after 39. Nxc6. Black to move.
How does Black save a draw?
Here Gjon played the most natural-looking move on the board, 39. … Rxf2?, which loses. After 40. b5! he doesn’t have time to take the f4 pawn, because my pawn would queen. If he tries to get his rook behind the pawn, with 40. … Rf1+ 41. Ka2 Rb1, White plays 42. Nd4!, a tremendous multi-purpose move.
- It defends the b5 pawn.
- It prevents … d4, which would allow Black’s bishop to get into the defense.
- It covers the f5 square, so that White can answer 42. … e5 with 43. Ra6+.
- Finally, it attacks the e6 pawn. Because all of Black’s possible counterplay has been negated, White will simply play 43. Ra6, win the e-pawn, and win the game.
It’s not too often you see one knight do so many different things!
Eventually Gjon conceded that the position after Nd4 looked lost. But there’s another winning variation we didn’t quite appreciate at the time, because it’s too computer-esque. White also wins after Black’s ingenious try 40. b5 Rc2!? 41. b6 Rc1+! (trying for a perpetual) 42. Ka2 Rc2+ 43. Ka3 Rxc3+ 44. Ka4 Bc2+ 45. Kb5 Rb3+ 46. Nb4!
But if the pawn were on b5, instead of b6, this variation wouldn’t work, because White wouldn’t have b5 available for his king!
And that’s the idea behind what Rybka says is the drawing variation. Black should spurn the pawn on move 39 and instead play 39. … Rg1+!! 40. Ka2 Rc1! and now, if 41. b5, Rc2+ draws.
Can humans learn to find such lines? In a speed chess game, with less than 10 seconds left on your clock, probably not. However, you’ve got to admit it makes a certain amount of sense. The move 39. … Rxf2 is basically a wasted tempo. Winning this pawn does Black absolutely no good unless he can win the f4 pawn too, but he doesn’t have time for that. This reinforces a lesson that every player should remember: In the endgame (especially in rook endgames!) a tempo is often worth much more than a pawn.
Black had to realize that his only hope was to set up a pawn race between White’s b-pawn and his c-pawn, and therefore job one is to win the c3 pawn, not the f2 pawn. Also, he had to see (admittedly difficult) that White’s awful king position was the tactical ingredient that would let him carry out his plan.
There were some other interesting moments earlier in the game, but I’ll just mention one of them. What do you think of Gjon’s opening?
Position after 3. … Nc6.
One of the great things about Gjon’s approach to chess is that he is constantly experimenting with new opening ideas. You never know what you’ll see from him next. And he never does it in an impulsive, spur-of-the-moment way. There are always ideas behind his experiments. When you go over the game with him afterwards, he’ll show you a sequence of seven or eight moves outlining his plan.
So here he’s done something the books say Black absolutely shouldn’t do — he’s blocked his c-pawn with his knight. Already I was totally confused. I knew there was something called the Guimard Variation of the Tarrasch, which goes 1. e4 e6 2. d4 d5 3. Nd2 Nc6. And in my confusion (remember, it’s a speed game) I actually thought that this WAS the Guimard Variation! But it’s not! White’s knight is on c3, not d2.
But does it matter? Is the knight so much better on c3 that Black’s 3. … Nc6 is ruled out?
I sure couldn’t see a refutation. And neither can Rybka! In fact, when you put the position after 3. Nc3 on Rybka, its third choice (after the book lines 3. … Nf6 and 3. … de) is 3. … Nc6! It actually rates this as being better than 3. … Bb4, even though that is an ultra-fashionable book variation.
So if anyone can tell me what’s wrong with 3. … Nc6, I’d be interested to hear it. Otherwise, perhaps it is a good way for Black to avoid the highly analyzed main lines of the French.
