Singin’ the time pressure blues

In my last post I reported on the results of the Silicon Valley Challenge. Although I was disappointed in my score (2.5/4), it was not all bad. I had two really fascinating endgames. Unfortunately, I botched them both in time trouble, drawing one and losing the other. Sudden-death time controls make it almost impossible to play subtle endgames correctly.

In round three, I lost to Kyle Shin, who went on to tie for first place with 3.5/4.  I played a really great game for 40 moves, had him on the ropes, and then lost with a beginner’s blunder. It felt like working on a painting and then spilling a can of paint all over it just when it is almost finished. Still, there are no excuses. I mismanaged my clock, and therefore I deserved to lose. (PGN is available here.)

Dana Mackenzie — Kyle Shin 

King’s Gambit

1. e4 e5 2. f4 d5 3. ed ef

One point of this move order for Black is that he can adopt the Modern Variation setup no matter what White does. With the move order 1. e4 e5 2. f4 ef, the move 3. Bc4 would keep Black out of the Modern Variation (although 3. … d5 is, of course, still playable).

4. Nf3 Nf6 5. Bb5+ Bd7 6. Bxd7+ Nxd7 7. O-O! …

When you’re playing White in the King’s Gambit, it’s important not to be bothered by temporary losses of a pawn.

7. … Nxd5 8. Re1+ Be7 9. c4 …

This move comes before 9. d4 because White does not want to allow Black the possibility of … Ne3.

9. … Nb6 10. d4 O-O

Here is where I lost the game! For no good reason, I spent EIGHTEEN minutes on my next move, which is a luxury one cannot afford in game/60. (Actually, we were playing game/55 with a five-second time delay.) After this move I had only 17 minutes for the rest of the game! Meanwhile, my opponent had something like 45 minutes. It is suicidal to give oneself such a time handicap, especially against an obviously gifted young player.

The move I chose was actually the correct one. So I will give it an “!” for being the right move and a “?” because of the time usage.

11. Bxf4!/? Nxc4?!

My opponent is showing a somewhat unhealthy appetite for pawns. The computer’s suggestion here is 11. … c5.

12. Bxc7? …

After this move,  White is scrambling to equalize. Better is 12. Qe2, which I didn’t play because I was worried about 12. … Bb4. But I forgot that my rook on e1 was defended! So in fact I can play 13. Qxc4 Bxe1 14. Nxe1, with two pieces for a rook and pawn and a comfortable advantage for White.

12. … Qxc7 13. Rxe7 Rae8 14. Qe2!? …

Again, the general modus operandi for the King’s Gambit: when in doubt, sac a pawn.

14. … Rxe7 15. Qxe7 Nd6

For the first time (unless you count move 2) Shin decides not to accept my pawn sac. The computer says that Black can in fact get away with playing 15. … Nxb2. However, you have to be a computer to play a move like this and survive.

16. Nc3 Re8 17. Qh4 Nf6 18. Re1 Rxe1+ 19. Qxe1 …

Basically a dead even position. Black’s next move is technically not a blunder, but I feel certain that he did not anticipate my response.

19. … Nf5?!

Now the previously boring position gets spicy!

20. Nd5! Qc2

Of course, 20. …. Nxd5?? would fall into an immediate checkmate with 21. Qe8 mate. I actually won my second-round game against a 1500 player in a similar fashion. But Shin keeps his cool, and realizes that Black still has a lot of activity.

21. Nxf6+ gf 22. Qc3 Qb1+ 23. Kf2 …

In spite of the time situation, I was feeling very optimistic here. White’s pieces are coordinating very nicely. Notice that 23. … Qxa2? would lose a piece after 24. Qc8+.

23. … Nd6 24. Nd2! …

Threatening yet another fork with Qg3+, so Black still can’t grab the a-pawn. This was the only time in the entire game that I saw even the tiniest hint of frustration in Shin’s body language. He made his next move with a sort of disgusted gesture, as if to say, “Enough with the forks already.”

Conversely, I was feeling great. Even though the position is still objectively even, White has the initiative, and you know me — there’s nothing I like better than the initiative. The pieces were singing to me.

24. … Ne4+ 25. Nxe4 Qxe4 26. Qg3+ Kh8 27. Qd6 …

And now an interesting thing happened: Shin offered a draw. Of course, by the rules of chess he has to play a move first and then make the offer, and so I told him this. But anyway, I wasn’t ready to settle for a draw just yet. I just wanted to see him play a couple more moves. And then after that, I just wanted to see a couple more. The more we kept playing, the better my position looked!

Regardless of whether his offer was legal, I could have accepted it. Was it hubris on my part to keep on playing? I don’t know. I have seen grandmasters and IMs play on many times in seemingly drawn positions against lower-rated players, and they almost always manage to win. Because I do aspire to being a master, I think that turning down draws in such positions is absolutely the correct decision.

Except… I now had about 4 minutes left, while my opponent had about 30. With such a huge time differential, playing on is a huge risk. I vividly remember a game I played 17 years ago in the Columbus (Ohio) Invitational, where I had a dead-drawn endgame against the tournament favorite, Chuck Diebert. The second time control (after move 40) was game/60. I used up lots of time trying to squeeze something out of the position, while he played quickly and maybe a bit too superficially. Miraculously, I managed to get to a won position — but I had only one minute left on my clock, and I blundered a rook and lost.

There’s an old saying: “Those who don’t learn from history are doomed to repeat it.” That saying is very apropos for this game.

Let’s fast-forward now to the critical position, 14 moves later. (You can play over the PGN if you want to see how we got here.)

White to play and win. 

I’m now down to 2 minutes, but on the board things are looking very good. I sacrificed the two kingside pawns in order to march my king up the board. This was completely an intuitive decision. In queen and pawn endgames, a far-advanced passed pawn is worth lots of “ordinary” pawns. The big issue in this position, for White, is how to advance the pawn and how to keep his king out of perpetual check.

The move I played here was 41. Kc8?? That’s right, I simply forgot that there was nothing protecting my pawn on d6! Of course he played 41. … Qxd6, and went on to win.

What would you have done here? The right move — and the one I certainly would have played if I had not been so low on time — was 41. Qc6!, defending the d-pawn and shielding the king from checks on the c-file. This move demands a certain degree of faith from White, because if offers up yet another pawn. Black has nothing better than 41. … Qxb2 42. d7 Qe5+ 43. Kc8 Qf5, which leads to another key position.

Position after 43. … Qf5 (analysis).

Can White win this? If so, what is the winning plan?

By the way, this is a great example of an endgame you cannot understand if you only analyze it on the computer. If you put the position in Fritz, it gives White a 6-pawn advantage, and so you think, “Ho-hum, easy win.” But it’s not easy at all! Even knowing the move that Fritz recommends here, I still couldn’t work out the win, and the solution I came up was different from the computer’s.

There are two key questions White needs to answer:

1. Where can White’s king find shelter from the queen checks?
2. Where is the best place for White’s queen?

The answer to #1 is that White has to use the a-pawns (both of them!) for shelter. It’s a pretty scrappy shelter — kind of like weathering a storm under a piece of plywood — but it turns out that it’s just enough to work. If the a-pawns were not on the board, I believe the position would be a draw.

The answer to #2 is that the best squares for White’s queen are either at c7 or behind the passed pawn. Those are the squares where White’s queen both protects the passed pawn and controls the queening square (indirectly, in the second case).

Therefore, the move that suggests itself is 44. Qc7. But this is a little bit too slow. The problem is that White is not really threatening to extricate his king yet with Kb7 or Kb8, and therefore Black can start his pawns running on the kingside with 44. … h5. This is a little bit scary, because it means Black is ahead in any potential pawn race with the a-pawn.

The good news is that White can do better. White can actually get his queen to c7 and still be on move! The idea is 44. Qd6!, stationing the queen behind the pawn. Now White really is threatening 45. Kb8, and Black does not have time to start running his h-pawn. Here’s the proof:

(a) 44. … h5 45. Kb8!

(a1) 45. … Qb4+ 46. Ka8! and Black is out of checks! This is a nice example of how White uses Black’s a7 pawn for shelter.

(a2) A cleverer idea for Black is 45. … Qb1+ 46. Ka8! Qe4+ (If 46. … Qh1+ 47. Kxa7 Qg1+ 48. Qb6! and Black is out of checks and out of ways to stop the d-pawn. Notice that it is now the a2-pawn that provides shelter for White’s king.) 47. Kxa7. White has now been forced to take the a7-pawn, which means his king is now out in the open, but amazingly Black is out of checks after either 47. … Qa4+ 48. Kb6! or 47. … Qe3+ 48. Kb8! The way that White’s queen and that pawn on a2 collaborate to control all of Black’s checking squares is absolutely exquisite.

So after 44. Qd6 we conclude that Black has to do something to prevent 45. Kb8. Neither 44. … Qg4 nor 44. … Qh3 work, again because White’s pieces control b3 and b4. So the only thing left for Black to do is:

(b) 44. … Qc2+ 45. Qc7 Qf5. Just as I promised, White was able to get his queen to c7 and still have it his turn to move! He has gained a precious tempo.

Okay, so now what? 46. Kb8 doesn’t do anything now, because Black can play 46. … Qb5+. So White again makes use of his a-pawn, and plays 46. a4! This again establishes Kb8 as a threat, so Black still cannot begin pushing his h-pawn. If 46. … h5, then 47. Kb8 wins very much as before after 47. … Qb1+ 48. Kxa7. (White doesn’t even need to do the little a8-a7 dance any more.) Therefore, Black has nothing better to do than improve the position of his queen with 46. … Qg4!

Once again, I don’t see a win with 47. Kb7 or 47. Kb8, so the only way for White to make progress is to keep pushing the pawn with 47. a5! Now all of a sudden the pawn changes roles, from merely a defender to an offensive weapon. White is now going to win the game by queening the a-pawn, not the d-pawn! After 47. … h5 (finally Black can play this move, but it’s too late) 48. Kb7 White wins by taking on a7, then marching his king back to c8. At that point Black will run out of checks, and will have to put his queen on the h3-c8 diagonal to pin the d-pawn. When he does so, White plays a5-a6-a7-a8, and there is nothing Black can do about it.

So, what can we learn from this mind-boggling endgame?

1. Queen-and-pawn endgames are never easy.
2. It’s impossible to play a queen-and-pawn endgame correctly with 2 minutes left on your clock.
3. Passed pawns that are within a square or two of queening are hugely valuable; they are often worth 2 or 3 “regular” pawns.
4. If you are trying to win, finding shelter for your king is often the trickiest part. You can use both your own pawns and your opponent’s pawns for this purpose. The closer your pawns are to your king, the more effective they are as shelter.
5. If you are the defender, close-up checks (with the queen separated from the king by only one or two squares) are usually better than long-distance checks.
6. Because queen endgames will often come down to a pawn race (either before or after a trade of queens), every tempo counts! Resist the urge to play meaningless checks or meaningless captures that don’t actually improve your position in some way. If you can find a way to “steal” a tempo, as White did in this example, it may make the difference between a win and a draw.
7. In spite of comments #3 and #6, there are times when pawns do matter. It’s easier to win with queen plus two pawns than queen plus one pawn. In line (b) above, White really needed his second pawn to win.

One might wonder whether it’s worth spending hours analyzing an endgame like this, when I had only a couple minutes over the board. I think it is important, for two reasons. I wanted to make sure that my intuition was correct, because that will help me play queen-and-pawn endgames correctly in the future. But at a certain point intuition breaks down, and you have to back it up with hard analysis. If you have seen similar motifs before, it will help you analyze correctly. Some people can learn by studying books and going over master games, but I always learn best by going over my own games. If I encountered the position after move 40 in a book, I would never have had enough passion and desire to really get to the bottom of it.