“I woke up with a headache like my head against a board/ Twice as cloudy as I’d been the night before/ And I went in seeking clarity.” — Indigo Girls, “Closer to Fine”
Besides being a line from one of my two favorite songs in the whole world (I’ll leave you guessing about the other one), this is a little bit how I feel this morning. Not so much the headache part, but the “seeking clarity” part.
Gjon Feinstein and I met last night to go over some of the games from my tournament last weekend. Besides my miraculous win in the last game, I didn’t have much to be happy about; in two games I squandered what appeared to be decisive advantages and had to settle for a draw.
Here is one of the positions where I went in seeking clarity.
Position after 30. ... Kd6. White to move.
FEN: 8/p2P3p/3kn2p/1pp1N3/4K3/2P5/PP4PP/8 w – – 0 31
I was playing White in the fifth round against an expert named Scott Mason, and we had just passed the time control on move 30. This means I had lots of time to think (second time control was game/60, and with material so reduced it’s very unlikely that time pressure will become an issue again).
As I looked at the position it seemed to me that I stood at a fork in the road. I can either win a pawn on the kingside with 31. Nf7+ Kxd7 32. Nxh6 or I can win a pawn on the queenside with 31. Kf5 Ke7 (forced — Black cannot allow the king to penetrate to f6) 32. Nc6+ Kxd7 33. Nxa7. Which way should I go?
I started my analysis with the latter variation.
Position after 33. Nxa7 (analysis)
FEN: 8/N2k3p/4n2p/1pp2K2/8/2P5/PP4PP/8 b – – 0 33
First of all, let’s dispel the notion that White’s knight is in any trouble after 33. … Nc7. It’s true that it has no moves, but also Black’s king has no way to approach it. I can just march my king to f6 and g7, gobble Black’s kingside pawns, and win.
A line that concerned me more was 33. … b4 34. a4 bc 35. bc (or 34. … ba 35. ba if you prefer). With my pawns split, I thought that Black might somehow manage to liquidate the queenside pawns and sac his knight for my kingside pawns. Looking at it now I think this is ridiculous. So that was my first analytical failure — I was way too pessimistic about the queenside pawn grab.
Now, what about the kingside pawn grab, 31. Nf7+ Kxd7 32. Nxh6? Here is where things get murky, and I’m not quite sure what is the correct lesson for me to learn.
After a much more cursory analysis than the other variation, I decided this was easily winning for White. I thought I would be able to follow up with Ng8 and Nf6+, winning the second pawn on the kingside, and then I didn’t see any way that he could even hope to stop my connected passed pawns.
But things started going awry right away. My opponent surprised me with the move 32. … c4!
Position after 32. ... c4. White to move.
FEN: 8/p2k3p/4n2N/1p6/2p1K3/2P5/PP4PP/8 w – – 0 33
I had completely failed to analyze this possibility. In fact, in my rather careless analysis I had not really analyzed anything after Nxh6. Failure number two: After spending a long time analyzing line one and not being completely happy with it, I then leaped into line two without sufficient analysis. This is a very common faulty thought pattern.
Here’s how the game continued from diagram 3: 33. Ng8? Kd6 34. Nf6 Nc5+ 35. Kf5 Na4 (Now things are going seriously wrong for White. There isn’t time for me to grab the h-pawn. As it turns out, Black didn’t have to stop my connected passers, he just had to outrace them.) 36. Ne4+ Kd5 37. b3 cb 38. ab Nc5!!
Position after 38. ... Nc5. White to move.
FEN: 8/p6p/8/1pnk1K2/4N3/1PP5/6PP/8 w – – 0 39
Until Black played this move I still thought I was completely winning. It only gradually dawned on me that it’s now a draw. First, if I play the passive 39. Nd2 I might even lose after 39. … Nxb3! (also … a5 is possible). Second, if I go back to Plan A with 39. Nf6+ Kc6 40. b4 Na4 41. Nxh7 Nxc3, Black’s queenside pawns are just as fast at White’s kingside pawns. It’s not a matter of quantity but quality, and also Black’s knight is very strategically placed. For example, 42. g4 a5 43. Ke5? (A doomed attempt to stop the pawn) 43. … a4 44. Kd4 Nd1! 45. Kd3 a3 46. Kc2 a2 and the knight and pawn jointly keep White’s king out.
Finally there is 39. Nxc5, the move I played, which eliminates any losing possibilities but also leaves me one tempo short of a win. The game concluded 39. … Kxc5 40. Ke4 a5 41. g4 a4 42. ba ba 43. Kd3 a3 44. Kc2 a2 45. Kb2 Kc4 46. h4 a1Q+! 47. Kxa1 Kxc3 48. g5 Kd4 49. h5 Ke5 50. g6 Kf6 51. gh Kg7 ½-½.
Now, before going any further, I think we have to give Mason credit for finding a truly inspired defense. But after acknowledging that, we have to ask what went wrong with my thought process. How was I supposed to know that going after the a-pawn wins, but going after the h-pawn only draws? Surely there must be something better than just flipping a coin.
What Gjon stressed last night is that in a position like this that is a clear fork in the road, you have to put maximum effort into your calculations. There is a place and a time for calculation. I often over-calculate in positions where I don’t really need to, and that’s one reason I get into time pressure. But here there was a need and there was plenty of time. Even though I used 15 minutes (!) I did a shockingly poor job of calculating the line after 32. Nxh6. I never even looked at 32. … c4. This is not grandmaster stuff. This is only TWO moves deep.
I think Gjon is right about that. But there’s another thing that both of us missed last night! Let’s go back to the position after 32. … c4. To help you I’ll copy it over again.
Position after 32. ... c4. White to move.
FEN: 8/p2k3p/4n2N/1p6/2p1K3/2P5/PP4PP/8 w – – 0 33
I’m sure you must be wondering, “Why not 33. Kd5?” It’s the move that cries out to be played. It’s the whole reason I didn’t even consider 32. … c4, because I just assumed that Black had to keep my king out.
Well, the problem with 33. Kd5 is that Black plays 33. … Nf4+, forking the king and the g-pawn. But let’s go a little farther! After 34. Kc5 a6 35. Kb6 Nxg7 36. Kxa6 Kc6 we get to a position that somehow Gjon, my opponent, and I all mis-evaluated.
Position after 36. ... Kc6 (analysis). White to move.
FEN: 8/7p/K1k4N/1p6/2p5/2P5/PP4nP/8 w – – 0 37
In the post-mortem, Mason played this move, 36. … Kc6 and we both said, “Oh yeah, draw.” When I showed Gjon the game last night, he reacted the same way. It’s an understandable gut reaction — you see the king trapped on the a-file and you think he has no way of getting out. But hello! There are knights on the board! In fact, after 37. Nf5 Ne1 38. Nd4+ I win the b-pawn and free my king. Black just remains two pawns down, and this should be an easy win.
So, in the end, my blunder was NOT that I chose the kingside over the queenside. Either way should have worked. The real blunder was that I “took my opponent’s word for it” that 33. Kd5 was no good. I had to analyze deeper. That’s how you win chess games, by seeing one move (or in this case, two moves) farther than your opponent.
So, have we reached clarity? I don’t know! After all of this, it seems as if the main conclusion is simply, “Analyze deeper.” (Or perhaps a more nuanced version: “Recognize the critical positions where analysis matters, and look one move deeper than your opponent.”) But that’s so obvious and banal that it’s hard to consider it a bona fide lesson.
“Twice as cloudy as the night before…”
