Black then usually plays 3...Nf6, at which point the commonest move in ChessBase's 2021 Mega database is 4.Bg5, which scores 59%, ie about four percentage points above White's usual score in large databases.
But there is worse to come - White can play 4.cxd5, the Exchange Variation, which scores a phenomenal 65%, the usual continuation being 4...exd5 5.Bg5.
It is instructive to compare this with White's second-most popular continuation at move three, namely 3.Nf3.
After the further moves 3...Nf6 4.cxd5 exd5, White's score plummets to 50%.
In other words, statistically speaking, White has lost the advantage of having the first move.
In practice White more often follows up 3.Nf3 Nf6 with 4.Nc3.
Black then has a large choice of respectable replies, including 4...Be7, 4...c6, 4...Bb4, 4...Nbd7, 4...c5 and 4...dxc4, each of which occurs at least 5,000 times in Mega21.
Most popular is 4...Be7, when White can again enter the Exchange Variation, but by now 5.cxd5 exd5 6.Bg5 scores only 56%, falling to 52% if Black responds with 6...c6.
True, 56% is marginally higher than White's usual average across all openings, but it is a significant drop from the 65% of 3.Nc3 Nf6 4.cxd5.
One of the main reasons the Exchange Variation with 3.Nc3 scores so much better than with an early Nf3 is that the former allows White to play the king's knight to e2.
This opens new possibilities for White, including a central advance with f3 and e4, which can be particularly effective as White in the Exchange Variation has a 2-1 central pawn-majority.
Another reason is that Nf3 played too early lets Black render the Exchange Variation largely harmless by more-or-less forcing off light-square bishops, as the following ICCF correspondence game illustrates.
James Cairney (1983) - Spanton (unrated)
2nd British Webserver Team Tournament 2013
1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.cxd5 exd5 5.Nf3!?
It would be over-the top to call this a mistake, but it scores just 45% in Mega21.
Black can now plan on playing ...Bf5, occupying the diagonal usually taken by the white light-square bishop.
5...c6
An immediate 5...Bf5!? may even be possible, the idea being that 6.Qb3 can be met by 6...Nc6 with sharp play.
6.Qc2
More popular is 6.Bg5, but Black can reply 6...Bf5, when 7.Qb3 Qb6 8.Bxf6 gxf6 is not considered particularly promising for White.
6...g6
My modern analysis engines, Stockfish12 and Komodo11.01, prefer a bishop move, eg 6...Be7, 6...Bd6 or 6...Bg4, but are fine with the text once shown it.
7.Bg5
Black is at least equal, and maybe even a tad better, after 7.e3 Bf5.
The text also sets a mini-positional trap that has snared grandmasters.
7...Bg7
The problem with 7...Bf5?! is that 8.Qb3 is awkward to meet since 8...Qb6?? is a gross blunder due to 9.Bxf6. Black's most-popular reply in Mega21 is 8...b6?, but 9.e4! dxe4 10.Ne5 Be6 11.Bc4 Qe7 12.d5 was catastrophic for Black in Péter Lukács (2460) - Jakab Attila (2396), Budapest GM-B 2002 (1-0, 15 moves). Black probably has nothing better than 8...Bc8, when 9.e4 is again strong.
Massively more popular than the text is 7...Be7, but 7...Bg7, which has been played by Euwe, Ragozin, Dreev and Negi, seems fine.
8.e3 Bf5 9.Bd3
Black is posed no problems by 9.Qb3 Qb6 10.Be2.
9...Bxd3 10.Qxd3 Nbd7 11.0-0
I see from my notes that two of my engines at the time, Stockfish3 and Houdini1.5a, reckoned White is slightly better, but Junior10 called the position equal. Today's engines, Stockfish12 and Komodo11.01, agree with Junior10.
11...0-0 12.Rfc1
White is preparing the dreaded Minority Attack, but, theoretically speaking, Black has nothing to fear here.
12...a5 13.Na4 Qe7 14.Bh4 Rfe8 15.a3 Qe6 16.Bxf6!?
This is Stockfish12's choice.
16...Bxf6 17.Qb3!?
The choice of both Stockfish12 and Komodo11.01, but it looks strange to me. Certainly, obstructing the b2 pawn does not help White's Minority Attack.
17...Ra7!?
This appears slightly awkward, but is a reasonably economical way of defending b7 and a5.
18.Nc3 Qf5 19.Rc2 g5!?
The classic black answer to a Minority Attack is to stabilise the queenside and counterattack on the other flank.
20.Re2 g4 21.Nd2 a4! 22.Qd1
22.Nxa4?! Bxd4.
22...Qg6 23.e4!
This central counterattack is also the choice of Stockfish12 and Komodo11.01.
23...Nb6
23...Bxd4 24.exd5 Rxe2 25.Qxe2 Bxc3 26.dxc6 bxc6 27.bxc3 was given by Houdini1.5a as equal, although Komodo11.01 - but not Stockfish12 - reckons White has a small edge.
24.exd5
Black seems fine after 24.e5 Bg5.
24...Rxe2 25.Qxe2 Nxd5
 |
| Not an easy position to assess |
*****
*****
*****
*****
White has an IQP, which is blockaded by a black knight, but otherwise the white position is sound, and the black kingside attack has been stopped in its tracks. Black's pawns are a little loose and the black rook is passive and not coordinated with the other black pieces, but the bishop bears down on the d4 target. My modern engines reckon the position is equal, and I believe that was also the verdict of my engines seven years ago.
26.Rd1 Nxc3!?
This unisolates d4 but gives White another isolani at a3. It also lets Black play ...b5 to protect a4 without worrying about the c pawn being backward on a half-open file, simultaneously solving the difficulty of the out-of-play black rook. However, White gets an initiative.
27.bxc3 b5 28.Qe8+ Kg7 29.Re1
Not 29.Qxc6?? Qc2, but after the text the c6 pawn is threatened.
29...Bg5 30.Nf1 f5 31.Qc8 Re7
Forced, otherwise White has 32.Re6.
32.Rxe7 Bxe7
Queen and knight can make a better team than queen and bishop, but here the bishop has targets and the white queen and knight are widely separated. However, the black king is much the more exposed, and the engines agree the position is still level.
33.Qd7 Qf6 34.Ne3 Kf8
After the natural-looking 34...Kg6!?, White can cause problems with 35.d5, although it is not clear that White gets a substantial advantage.
35.Nxf5
But not 35.Qxf5?? Qxf5 36.Nxf5 Bxa3, and the a4 pawn is too strong.
35...Bxa3 36.Qc8+ Kf7 37.Qd7+ Kg6!?
37...Kf8 repeats the position. The text keeps the game going, with both players walking a tightrope.
38.Ng3 Bf8 39.Qxg4+ Kf7 40.Qd7+ Kg8 41.d5 a3 42.dxc6
White has won two pawns, but the a pawn is a monster that can promote with check. The game is still level, according to the engines.
42...Qd6 43.Qg4+?!
White can force a draw by 43.Nf1 Qxd7 (other moves lose) 44.cxd7 Be7 45.Nd2. The text may also draw, but is less clear, and is dodgy from a practical view as Black keeps his passed pawn while White will lose his, meaning White cannot realistically win.
43...Kh8 44.Qe2 Qxc6 45.Qe5+ Kg8 46.Qg5+ Qg6!?
Black can force a repetition with 46...Kh8, or try for more in another way with 46...Kf7 47.Qh5+ Qg6 48.Qf3+ Qf6. But after the latter it seems both 49.Qh5+ and 49.Qb7+ draw, eg 49.Qb7+ (during the game I thought this was the surer draw for White, but now I am not so certain) Be7 50.Qxb5 a2 51.Qa4 Qe6 52.Ne4! Qxe4 53.Qxa2+, when White has queen and four pawns versus queen, bishop and one pawn. My engines at the time reckoned Black has the upper hand, but I could see no way to create winning chances, and my modern engines agree.
47.Qxb5 a2
White has engineered a new passed pawn (two actually, including the f pawn) and restored his two-pawn advantage, but the black passer has reached seventh heaven. Stockfish12 and Komodo11.01 reckon the game is level.
48.Qc4+ Qf7 49.Qg4+ Kh8 50.Qd4+ Bg7 51.Qd8+ Qg8 52.Qa5 h6 53.h4
It is interesting that both players have time to make quiet pawn moves. 52...h6 creates a square for the black king at h7, while 53.h4 removes the possibility of a back-rank mate and lets the white king move up to h2 if Black threatens a back-rank check to support the a pawn promoting.
53...Qe6
For a long time Stockfish12 shows 53...Qb3 as winning, but after I play through the moves 54.Qa8+ Kh7 55.Qe4+ Kg8 56.Qa8+ it agrees with Komodo11.01 that the game is drawn, eg 56...Bf8 57.Kh2 Qa3 58.Qd5+ Kh8 59.Nf5 a1=Q 60.Qd4+ Kh7 61.Qd7+ Be7 62.Nxe7 Q1xc3 63.Nf5+ and White has a perpetual.
54.Kh2?!
Komodo11.01 reckons this draws, but that is far from clear, as will be seen. However, 54.Ne4!? probably does draw, assuming 54...Qxe4 55.Qxa2 does not give Black winning chances after 55...Qxh4 or 55...Bxc3, which does seem to the case (I presume White's drawing chances are enhanced by the black h pawn's queening square being the opposite colour to the black bishop).
54...Be5 55.Qa3 Kg7 56.Qa5 h5 57.Qa3?
This loses.
Komodo11.01 gives 57.Kg1, and if 57...Kg6, which I have in my notes, it continues 58.Qa4 (I only had 58.Kh2, which it seems is definitely winning for Black), after which Stockfish12 switches between various moves, all of which at one time or another it reckons win for Black, but Komodo11.01 always manages to defend valiantly, sometimes with an assist from Stockfish12. There are several lines in which White can allow Black a second queen but still draw with perpetual check.
57...Bc7 0-1
Best-play seems to be something like 58.Qa8 (58.Qa4 Qd5 transposes) Qg8 59.Qa4 Qd5 60.f4 Kf8 (the king heads to the queenside for shelter!) 61.Qa3+ Ke8 62.Qa4+ Kd8 63.Qa3 Bxf4 (or 63.Ne2 Qa5 64.Qd1+ Kc8 65.Qa1 Qa4 66.c4 Qc2 67.Qh8+ Kb7 etc) 64.Qf8+ Kd7 65.Qg7+ Kc6 66.Qg6+ Kc5 67.Qa6 Bxg3+ 68.Kg3 Qd6+ etc.
White can make it an Exchange Variation as early as move three, but this approach is innocuous too.
Wolfgang Vullhorst (1820) - Spanton (2034)
Bad Wörishofen 20071.d4 d5 2.c4 e6 3.cxd5 exd5 4.Nc3 c6
The immediate 4...Bf5!? may be playable, intending to meet 5.Qb3 with 5...Nc6.
5.Qc2 g6 6.Bf4
This is the normal move in the position, but maybe White should try 6.e4!?, although it involves accepting an IQP that is already under restraint.
6...Bd6
I rejected 6...Bf5 because I wanted to oppose the active white bishop, and 6...Bf5 7.Qb3 more-or-less obliges Black to play ...Qb6, ruling out a quick ...Bd6. Nevertheless, 6...Bf5 is the main line.
7.Bxd6 Qxd6 8.e3
8.e4 dxe4 9.Nxe4 Qe7 is at least equal and may be a tad better for Black, according to Stockfish12 and Komodo11.01.
8...Bf5 9.Bd3 Ne7 10.Bxf5 Nxf5 11.Nf3 Nd7 12.b4
WV offered a draw.
12...0-0 13.Qb3
13.b5 cxb5 14.Nxb5?! Qb4 15.Nc3 Rfc8 is good for Black, although the engines reckon 14.Qb3 keeps the game equal.
13...b5!
This is a common positional ploy in the Exchange Variation when White no longer has a light-square bishop. White's Minority Attack is halted, and Black creates an outpost at c4 that, if needed, can be occupied by a knight, thus covering the newly backward c pawn.
14.0-0 a5 15.a3 Qe7 16.bxa5!? Rxa5 17.Qb4?
17.a4 keeps White's disadvantage to a minimum.
17...Rfa8 18.Rac1!?
The engines reckon 18.Qxe7 Nxe7 19.Nb1 is objectively better, but it is horribly passive.
18...Rxa3
Black is a protected passed pawn up and has the more-active rooks. I went on to win (although only after blundering my advantage away and having to, in effect, win the game a second time).
So what has all this to do with Garry Kasparov? That is what I will come to in part two of this series.
No comments:
Post a Comment