One of those missed opportunities. When Black castles early he invites an immediate assault, and best here is 10.Bd3! g6 (10....h6 11.Bxh6 gxh6 12.Qxh6 gives White the classic Urusov Gambit attack) 11.Rhe1 Be6 12.Ne4 Nxe4 13.Rxe4 Bxg5+ 14.Nxg5 h5 15.g4 (15.Rxe6!? might be stronger) 15....Kg7? 16.Rxe6 1-0 Kokholm--Christensen, Denmark 1993.
This move is generally regarded as inferior to the standard 10....Be6. The idea is to prevent 11.Bd3, when Black can exchange Bishops and blunt White's attack. And in some lines the Bishop might support Ne4. But the Bishop is a bit exposed to attack where it stands, and if it is driven back to g6 it will be exposed to pawn assaults such as f4-f5 or h4-h5. Also, 10....Be6 has the advantage of blocking both the a2-g8 diagonal and the e-file, which White can now use as avenues of attack.
I had already been studying the Urusov Gambit at this time and should have known that the "book move," from the game Keidanski-Lasker, Berlin 1891, is 11.Qf4. But I had forgotten the book and so was left to my own devices. I rather like the move I came up with even though close analysis reveals that it is not as good as Keidanski's. His game with Lasker went 11.Qf4 Bg6 12.g4 Na5 13.Bd3 Qd7 14.Bxf6 Bxf6 15.Nd5 Bd8 16.Bf5 Bxf5 17.gxf5 f6 18.h4 b5 19.Nd4 Nc4 20.Qe4 Rc8 21.Nc6 Ne5 22.Nxa7 Ra8 23.Nb6 cxb6 24.Qxa8 Qxf5 25.Qd5+ Kh8 26.Qxd6 Qf4+ 27.Kb1 1-0 The idea of moving the Queen closer to the center in order to clear the way for a pawn advance with g4 and h4 makes sense, since the Bishop - which ends up at g6 - invites White to attack it with pawns and gain time for the pawn breakthrough. Lasker works hard in the game to defeat White's plan, but Keidanski still gets his open lines. In the end, though, it is White's initiative and his control of the entire board that wins the game.
My idea with 11.Nd4?! was to keep the Queen at h4 and try to advance pawns to f4 and g4 so as to actually threaten to trap Black's Bishop if it reteats to g6. So, for example, if 11....Bg6 (instead of 11....Nxd4 in the game) 12.f4! (threatening to trap the Bishop with 13.f5 since both 13....Nxd4 14.fxg6! and 13....Bh5 14.Bxf6 offer no escape) 12....Qd7 (12....h6 13.Bxh6 gxh6 14.Qxh6 is deadly while 13....Ne4 is met by 14.Qg4 threatening Qxg6 and the pawn at f4 blocks saving checks for Black at g5) 13.g4! and White's pawns are rolling with the threat of 14.f5 and Black cannot play 13....Qxg4 because the Bishop at e7 hangs after 14.Nxc6. Equally bad would be 12....Nxd4 13. Rxd4 Re8 14. g4 h6 15. Bxh6 gxh6 16. Qxh6 Bh7 when 17. Rd5! looks winning.
As these last lines indicate, one
other potential plus of 11.Nd4 is that it indirectly pressures the
Bishop at e7 (which is potentially exposed since its partner went
to f5 rather than e6 to block the e-file). The
downside of the move, though, is that it weakens White's control of
the often critical g5 square and so allows Black to play h6 with success
This is the most natural response and probably best. By exchanging Knights Black does weaken the Bishop at e7 while helping White activate his Rook, which might be able to swing over to the kingside in some lines.
White does not actually threaten 13.Rxe7? Qxe7 14.Nd5 because of the defensive trick 14....Qe1+ 15.Rd1 Qe4! threatening mate at c2 and forcing the exchange of Queens (this is one plus of having the Bishop at f5). But ideas like that are in the air so Black decides to take precautions, depriving White's Knight of the d5 square while preparing the block the Bishop with d5.
But Black has much better.
The most logical attempt to improve here would be 12....h6! since 13.Bxh6 gxh6 14.Qxh6 Ne4?! 15.Nxe4 Bxe4 threatens Bg5+ and forces 16.Qh5. But after 16....Bh7 (what else?) 17.Rg4+ Kh8 18.Bxf7, White looks to be doing well, with two pawns for the piece and continuing pressure on Black's exposed King. And no better is 13....Ne4?! 14.Qh5!? Bg5+ 15.Bxg5 Qxg5+ 16.Qxg5 Nxg5 when White seems to have a good game with material equality and a continuing initiative after 17.Nd5!
The critical line is really 12....h6! 13.Bxh6 gxh6
14.Qxh6 Nh7! and Black seems to be winning since the threat of Bg5+
forces White back and gives him time to play Be7-f6-g7 and Qf6 organizing
cover for the King. Play might continue 15. h4 (15. f4 Bf6
White might try to interject 14.Qf4!? here, hoping Black will retreat the Bishop and give him a good attack (after 14....Bg6 15.Qxh6 threatening Qxg6+), but Black can play simply 14....Qd7! and White has gained nothing over the other lines.
Unless some improvement can be found for White here, it looks like the whole 11.Nd4?! idea is a bust due to 11....Nxd4 12.Rxd4 h6!
Because Black does not force matters with h6 he gives White time to get his pawn attack rolling!
It is hard to find a better attempt at defense here with the threat of 15.f5 trapping the Bishop. Now, however, we see the secondary purpose of the pawn advance at g4 and f4 since it allows White to cover an attack on his Queen after 15.Bxh6 Ne4 with 16.g5!
Meeting the threat of 17.Qxg6+by unpinning the f7 pawn.
The Knight retreat threatens both 18....Bg5+ winning White's Queen and 18....dxc4 winning the Bishop (since the pawn at d5 is now unpinned). No other Knight move is better: 17....Ne4 18.Nxe4 and 17....Nxg4 18.Rxg4 allow White to pick up a piece while covering the g5 square. There may be nothing better. The simple retreat with 17....Bh7 will not work since 18.g5 Ne8 19.f6 followed by 20.Rh4 is the perfect fulfillment of White's plan.
The Queen retreats to escape from the check at g5, and from h3 it indirectly puts pressure on the Knight at d7 in case of 18....dxc4. But White misses the kill here: 18.Rxe7! (the most direct way of negating Black's threat of 18....Bg5+) 18....Qxe7 19.fxg6 fxg6 (forced since 20.Qh7 mate is threatened and 19....Nf6? fails to 20.g7! and 21.Qh8#) 20.Nxd5! and White wins in all lines. For example: (1) 20....cxd5 21.Bxd5+ Rf7 22.Qxg6+ Kf8 23.Qxf7+ Qxf7 24.Bxf7 Kxf7 25.Rxd7+ with four pawns to the good; (2) 20....Rf1+ 21.Kd2 (not 21.Rd1 Rxd1+ 22.Kxd1 Qd6!) 21....Qe1+ 22.Kd3 Qd1+ (22....Rf3+ 24.Ne3+!) 23.Kc3 and Black will run out of checks; (3) 20....Qe1+ 21.Rd1 Qe6 22.Ne7+ etc.
Despite missing the win, White gets to continue his attack due to the exposed position of Black's King. But things could have been very difficult for White if Black has played most correctly.
Necessary here was 19....Bg5+! to prevent White from advancing the g-pawn. White must then get creative to recover the piece after 20.Kb1 fxg6 21.Ne4! Re8! (21....Be7 22.g5! and other moves allow White to regain the piece by either Nxg5 or Nc5) 22.Red1 Rxe4! 23.Rxe4 Qc7 and Black has blunted White's attack and emerged up two pieces for a Rook.
Now White recovers his piece without any trouble and is able to continue his attack to boot.
Tricky: not now 21.Rxd7 Bxg5+!
Better may be 23....Rd8 24.Rxd8 Bxd8 though White continues to attack with 25.Qc8!
Black must have expected 25.Rd1, but now the Rook remains on the seventh rank with the threat of Qh7+ hanging in the air.
There was nothing to be done about 27.Qxg6+ followed by mate, since 26....g5 allows 27.Qh5! Qe4 28.Qf7+ Kg8 29.Qg7#. But surrendering the Queen is hopeless.
Though play by both players was quite flawed, the game suggests some of the power of the Urusov Gambit in securing a longterm initiative. The analysis also suggests that the book move 11.Qf4! is best here since Black has effective answers to 11.Nd4?!