Two of my most intense and difficult wins of the past year both happened against the same person, who also happens to be a reader of this blog: Praveen Narayanan. By coincidence, I spent the morning studying the game I call Praveen II (though perhaps he would call it Dana II), and on the same day my ChessLecture  on Praveen I was posted. So I’m declaring today Praveen Day.

Praveen I wasn’t the best game for a ChessLecture because there were a few too many mistakes, on both sides. But it was nevertheless extremely entertaining, especially some study-like variations in the endgame.

Praveen II also went down to a very tricky endgame, where I went from almost busted to winning as if by magic. Of course there’s no magic in chess, so let’s see what actually happened.

Position after 29. ... Rxe3+. White to move.

FEN: 8/p7/2pk3p/5pp1/1PP2p2/3Pr2P/2P1K1P1/5R2 w – – 0 30

It was the move before the time control, so I had about 60 seconds to make a very important decision — which way to go with my King? 30. Kd2 or 30. Kf2?

I think I chose the correct way: 30. Kf2. If White plays 30. Kd2 Rg3 31. Rf2 his rook becomes passive, and moreover I think that Black can take advantage right away with 31. … f3! If 32. Rxf3 Rxg2+ 33. Ke3 f4+ 34. Ke4 Rxc2 35. Kf5 Re2 Black snags a pawn and gets his rook back in time to stop White’s king incursion. If 32. gf Rxh3 I can’t say for sure that Black is winning, but he has definitely improved his position, creating the possibility of two connected passed pawns, while White’s pawns are still disorganized.

After 30. Kf2 Black played what seems like an extremely natural move: 30. … g4?! When I saw Praveen at Mike’s chess party last weekend,  he said something like this: “I went over the game and it was pretty good, all except g4. What a bad move!”

Now it seems to me that if Black is playing for a win, he almost has to play 30. … g4. So I was very surprised to hear Praveen say this. But after looking at the endgame carefully, I think that he has a point. The basic problem is that Black shouldn’t be playing for a win. It seems surprising because his king is better placed than White’s king and his rook is way better placed than my rook — at least for the moment.

But the decisive factor is pawn weaknesses. Black has loose pawns all over the board: a7, c6, and after 30. … g4 he will have two more, at f4 and h6. Meanwhile, despite White’s lack of activity, his pawns are at least very compact and difficult for Black to attack.

In a R+P endgame, weak pawns = trouble. And that’s why Praveen’s move 30. … g4 was a little bit too optimistic. Nevertheless, in my opinion this is not the move that lost him the game. I think he had two more chances to draw.

The game continued 31. hg hg 32. Rh1! g3+ 33. Kf1 and now Praveen faced an important decision.

Position after 33. Kf1. Black to move.

FEN: 8/p7/2pk3p/8/1PP2p2/3Pr1p1/2P3P1/5K1R b – – 0 33

Here he played 33. … Ke5! Ironically, during the game I thought this was his losing move, because I had already calculated the following line which ends with him one tempo short of a win. I thought he was guilty of trying too hard to win here. But in fact, 33. … Ke5 is his best chance to save a draw. If he plays 33. … Re6?, the move I thought he had to play, then I am just going to play 34. Rh5. Not only does this stop all of Black’s attacking chances, it also threatens Rf5 followed by picking off Black’s overextended kingside pawns, one by one. This is where we really see Black’s weak pawns in their full ghastly glory.

So in fact Praveen’s best hope is to charge with his king into the soft underbelly of my position. It turns out that by combining this with checkmate threats on the back rank, he has just enough counterplay to draw.

The game continued 34. Rxh6 Kd4 35. Rxc6 Kc3 36. Ra6 and now we hit another key position.

Position after 36. Ra6. Black to move.

FEN: 8/p7/R7/8/1PP2p2/2kPr1p1/2P3P1/5K2 b – – 0 36

As if by Newton’s First Law of Motion (“Every body persists in its state of … moving uniformly straight forward, except insofar as it is compelled to change its state by force impressed”), Praveen continued to move his king forward with 36. … Kd2?!

What could be wrong with threatening checkmate? Well, for one thing, it’s easy to defend. And for a second, it takes the king away from a square where he was doing good things, threatening the pawns on both c2 and b4.

The correct move was the defensive move, 36. … Re7! We’ll see that he played this on the following move, but the big difference is that 36. … Re7 forces me to advance the “wrong pawn” — the b-pawn. After 37. b5 Kxc2 38. d4 Kc3 walks down White’s c- and d- pawns and doesn’t allow connected passers. This looks to me like a pretty easy draw for Black.

Instead, the game went 36. … Kd2?! 37. Ra1 Re7 38. c5! Rh7 39. Kg1, and now the c2 pawn is taboo. If 39. … Kxc2 40. d4, and White is pushing the “right” pawn, the d-pawn. After 40. … Kc3 41. d5! Kxb4 42. Rc1 the connected passed pawns are unstoppable.

Still, Praveen had one more chance to save a draw. He played a good move, 39. … Rb7, and now I think I surprised him with 40. c6!

Position after 40. c6. Black to move.

FEN: 8/pr6/2P5/8/1P3p2/3P2p1/2Pk2P1/R5K1 b – – 0 40

Even during the game I knew that this was just a trap, not an actual winning move, but I couldn’t find a legitimate win. The trap, of course, is 40. … Rxb4?? 41. c7!, and Black’s rook can’t find a square to stop the pawn from queening.

However, 40. … Rb6! holds the game. The back-rank checkmate threats play a huge role. For example, if 41. c7 Rc6 42. Rxa7 Rxc2, the mate threat forces White to retreat with 43. Ra1, and then after 43. … Rxc7 Black is out of trouble.

The closest thing I could find to a win for White (after 40. … Rb6!) is 41. c4 Rxc6 42. Rxa7 Re6! (that back-rank stuff again) 43. Ra1 Kxd3 44. c5 Re4 (see diagram).

Position after 44. Re4 (analysis).

FEN: 8/8/8/2P5/1P2rp2/3k2p1/6P1/R5K1 w – – 0 45

Now 45. Rb1? Kc2 would drop the b-pawn right away. So the best I can do is 45. b5 Re5 46. b6 Rxc5 47. Rb1! Rc8 (47. … Kc2?? 48. b7! Queening with check is the best remedy against mate threats!) 48. b7 Rb8. Finally White has gotten Black’s rook to a passive position. Unfortunately, White has no way to get his king out of jail! After 49. Rd1+ Ke3 50. Rd7 Ke2 51. Rf7 Ke3 (diagram) all that Black has to do is keep his king in the holy quadrangle of e2-e3-d2-d3 and White can never get his king out.

Position after 51. ... Ke3 (analysis). White's king is condemned to lifetime imprisonment.

FEN: 1r6/1P3R2/8/8/5p2/4k1p1/6P1/6K1 w – – 0 52

This was a somewhat long-winded variation, but I wanted to show that Praveen’s idea of locking the White king up on the back rank and using his own king as the jailer was actually a great idea. It just needed to be pursued with a little bit more conviction. (No pun intended. Okay, well, maybe pun intended a little bit.)

Instead, going back to diagram 4, Black caved in and played 40. … Rc7? The rest was easy: 41. b5 Kc3 (if 41. … Kxc2 42. d4 and the d-pawn flies to its destination) 42. Kf1 Kb4 43. Rb1+ Kc5 44. Ke2 Kd4 45. Kf3 resigns.