What surprised me most was how poorly the commenters seemed to understand each other’s standards for whether an ending was simple or complex. Gee whiz, here I am, all freshly signed up on blogspot so that I can engage in dialog, and I might not be understood! I’m not posting stuff just so that I can be misunderstood!
So I thought maybe I’d post an interesting endgame that veers (especially in the sidelines) between complex, simple, and extremely complex. That way, whenever anyone misunderstands when I call an ending simple or complicated, I can just refer them to this post and my personal standards (whether or not you agree with them) will become perfectly clear.
Bacon – Shernoff, World Open, 2002 (Round 1)
We reached this ending after an exciting but poorly-played middlegame: first I allowed White to sac a bishop for all three pawns in front of my king, obtaining a dead won position. Then, in time trouble, White declined to push the center pawns that you see in the diagram, or to keep kicking my king around; instead he won an exchange and exchanged queens. I assume he was afraid of hanging material, or just thought the ending must be dead won with his material advantage.
White had made the time control with two seconds to spare (the magical precision of digital clocks!) and then gone outside for a smoke to calm down. So at this point he had about 50 minutes left in the sudden death time control, and I had about half an hour more.
I’ll return to the question of what White should play here at the end of the post. In fact, this is one of a few issues that I’ve broken off into its own post below this one, so that we don’t get our threads tangled. Or at least keep them minimally tangled. I want us all to be understood, after all...
45.Kf2 Nb4!
46.Re4 Nd3+
I felt that the b-pawn was the better one to steal, since my knight is nearer the center and it’s easier for it to jump back into action (for example, Nc3 is impossible if I take on a2). There are also some neat coordination things that happen between the Nb2 and the pawn I’m trying to queen. For one thing, the idea of Nd1, momentarily blocking the back rank, becomes thematic.
47.Ke3 Nxb2
White’s original intention here may have been to simplify, but that loses:
48.Rxd4 cxd4+ 49.Kxd4 b4 50.h4 b3 51.axb3 a3 52.Kc3 Nd1+ 53.Kc2 a2 queens.
When I showed this game to my friend Todd Rowland, he suggested 48.Rh4, which just goes to show that when you randomly lose material in an ending it’s not necessarily bad. Black has to take the rook, since if White gets to play Rh8-b8/a8, Black will never make any progress on the queenside:
48.Rh4!? Nf5+ 49.Ke4 Nxh4 50.gxh4
and this has become an extremely complicated ending, since Black can’t queen his pawns and also get his knight back quickly enough to stop White’s pawns: after 50...b4 51.Kf5 Nc4 52.e6+ Kf8 53.Kg6 Ne5+, Black is involved in tricky maneuvers of trying to impede White from queening while pushing his own pawns.
After 50...b4,
White also has the alternative of 51.h5, forcing a new queen. Then best play seems to be something like 51...b3 52.axb3 a3 53.h6 a2 (53...Kg6 54.e6 Kxf6 55.h7 is not an improvement, as White queens first) 54.h7 a1Q 55.h8Q Qh1+ 56.Kf5 Qf3+ 57.Kg5 Qg2+ 58.Kf5 Qg6+ 59.Kf4 Nd3+ 60.Ke3 Nxe5=
This was my analysis shortly after the game. However, looking at it with fresh eyes just now, I decided that there’s actually no need for Black to spend tempi queening his pawns now, since White can’t stop them later on. This means that if you see the right line (or have the right idea) it’s actually a simple endgame after all:
50...Ke6 (stopping Kd5, attacking the queenside pawns) 51.h5 Nc4 52.h6 Nxe5 53.h7 Ng6 and Black wins.
Unfortunately, this line allows White at least a draw with (51.h5 Nc4) 52.f7! Kxf7 53.Kd5. So it’s still a simple ending – just with the wrong evaluation. In fact, White can even play for the win with a-pawn and h-pawn(s) vs knight. So it looks like Black has to embrace the complications in this line to have any winning chances at all. And it certainly looks like White’s best winning try in this ending.
Back to the game:
48. Rg4 Ne6 49.h4 b4 50.h5 b3 51.axb3 a3
51...axb3? 52.h6 c4 53.h7 Nd1+ 54.Kd2 c3+ 55.Kxd1 Nc5 56.h8Q c2+ 57.Ke2
52.Rh4 c4! 53.Rh1
53.bxc4?? Nd1+ and the a-pawn will queen. If White was a GM and won this game, no doubt s/he would draw our attention to the maneuver Re1-e4-g4-h4-h1.
53...cxb3 54.h6 Kg8 55.Ke4
If 55.f7+ Kh7
55...Na4
Again, looking at this position with fresh eyes, at first I thought this was a mistake that gives Black a chance to draw. The new line I saw is 55...a2 56.Kd5 (if 56.Ke3 Nd1+ and ...b2) Nf8/g5 with Nd1 on the next move and Black will queen. But this only works if White doesn’t play 57.e6 (and f7+ followed by h7). Back to the game:
56.Kf5?
White has to play Kd5. As in the famous Reti study shown here, the king has to move diagonally between his two goals, feinting at the kingside in order to reach the queenside in time.
After 56.Kd5 Nac5 57.Kc4 I was anticipating situations where my Nc5 and two pawns get exchanged for his rook, with his king recapturing way down in the corner. This gives my K+N plenty of time to round up White's remaining pawns, drawing. I’ll call this the queenside liquidation strategy.
57...b2 58.Kc3 Nd3 59.f7+ Kh7 60.Kb3 Nc1+ 61.Kc2
The Last Crossroads
and again, similar to the exchange sac above, there are three lines that Black can choose, one of which makes things very clear and two that entail more complexity.
Try 1
The clear one is 61...Nd4+ 62.Kb1 Nce2 (I like this line not only because it’s clear, but also because I get to threaten a nicely geometric perfect mate with Nc3+) 63.Ka2 Nc3+ (63...Nc1+ 64.Kb1 also repeats) 64.Kxa3 b1/Q 65.Rxb1 Nxb1+ 66.Kb2 Ne6 and here we have the simplest implementation of the queenside liquidation strategy.
Try 2
The more complicated line is 61...Nd3 62.Kb1 (to stop the a-pawn; 62.Kb3 Nc1+ repeats) and Black now has a couple ways to try and implement Nd2/c3+ and b1/Q, ending up with the queenside liquidation as above. I’m pretty sure that I was planning 61...Nd3 during the game, since it is the starting point of my analysis immediately afterwards.
Unfortunately, for instance after 62...Nxe5 63.Re1 Nf3 64.Re2 (not 64.Rxe6 Nd2+) White will get to play either Rf2 with the idea of f8/Q, or Re3xa3. Or after 62...Nf2 63.Re1 (covering e4 and d1) and Black seems to be in zugzwang. He can’t really move either knight, and if he moves the king White can force him to take the h-pawn by pushing it. Then he can play Ka2 and capture the a-pawn, as Nd3 (with the idea of Nc1+) can be met by Rh1+ and Kxa3.
If Black loses his pawns, the best he can do is a draw (even if White magically loses his rook) since White’s pawns are so far advanced; so he should prefer the simple drawing line of Try 1.
Try 3
The other factor that makes this continuation extremely complicated is that Black can force a new queen of his own, but only at the cost of allowing White to queen also. This is not a good bargain for Black. While we’ve all heard a million times what a strong attacking force Q + N are in the endgame (so Q + 2Ns should be even better, right?), for hunting down a lone king on the open board, they really kind of suck. Q + R is what you really want for that, and that’s what White has. Since White pretty much has Mate On Tap ™ as soon as he queens, Black has to mate with all checks, but it’s pretty easy for him to just walk away from the knights while staying on the same (correction) opposite colored squares as them, which makes it impossible for them to catch up with check. Here is a sample of the issues:
After 61...Nd4+ 62.Kb1 Ndb3
White can try for a perpetual check on f8 and g6 with 63.f8N+ Kh8 64.Ng6+ Kh7 65.Nf8+ but now black plays for the win with 65...Kg8 66.h7+ Kf7 67.h8/Q a2+ 68.Kxb2 a1Q+ 69.Kc2 Qa2+ 70.Kc3 Qd2+ 71.Kc4 Qd3+ 72.Kb4. Now the White king is confined, and Black wins by bringing up his remaining piece: 72... Na2+
73.Ka4 (73.Ka3 also Nd4+ mates simply) 73...Nc5+ 74.Ka5 Qa6 mate.
But all this means is that White should queen immediately with 63.f8/Q instead of screwing around with perpetual via the underpromotion. Now the mating line in the last diagram doesn’t work because the Qf8 guards c5, and I have been unable to find another way for Black to prevent the White king from just walking away. Try it. I know that most of you can just dump the position into your computer and it will be perfectly happy to spit out a string of 79 checks (if such exist) leading to mate. This is irrelevant to what you can find at the board with 30-40 minutes on your clock, after the fatigue of what I’d had to find already. I’ve put a post about the position after 63.f8/Q after this one. I’m interested to hear people’s thoughts about how this analytical task can be approached sanely and practically at the board.
Back to the game:
Now (after 56.Kf5) we have a simple ending for the first time in the actual game.
56...Nac5 57.g4 a2 58.g5 b2 59.g6 b1Q+ 0–1
The Ne6 forces the white king back, so after 60.Kg4 Qxg6+ 61.Kf3 Qe4+, Black wins the rook in the corner anyway.
If White had played 56.Kd5 he could have also tried this continuation, since queening as in the game is not check and loses to 60.h7+ Kh8 61.g7+. But instead Black could play 59...Nf4+ 60.Kxc5 Nxg6 and b1/Q on the next move.
No comments:
Post a Comment