My favourite online chess resource is
ChessPublishing, Tony Kosten's opening-theory site with regular monthly updates covering the entire spectrum of chess openings, and updated ebooks containing and summarising the section authors' previous work. It's far from perfect, and some sections are more useful than others, but I like it.
I subscribe to the section on 1.e4 e5, the one on 1.d4 d5, and the one on Flank Openings - which may not seem entirely consistent but I take a different view of classical openings when I'm on the White side of the board. The 1.d4 d5 section is written by Ruslan Scherbakov: I have a lot of time for Scherbakov, who seems to put a lot of work in when considering and evaluating both the openings and new games which develop the theory. All right, that sounds like a blurb from a mediocre openings book, but trust me. If I didn't think it was good I wouldn't pay to read it.
To my mind Scherbakov not only puts the work in, but gives us the benefit of his own study in having relied on, and therefore studied, certain lines during his own professional career. So, for instance, there is a surprisingly long section - which, I'm afraid, you can't read unless you're a subscriber - on the Marshall Gambit, which he has defended, from the Black side, for a number of years. Given the absence of any specialist books on the opening, I imagine this is the most substantial contemporary source for Marshall Gambit theory, in English anyway.
I'm not referring to the famous line in the Spanish, but White's gambit response to a Black triangle pawn formation.
1.d5 d5 2.d4 e6 3.Nc3 c6 and now, rather than play 4.e3 or 4.Nf3, White can try
4.e4.
Black plays this way in order to avoid the Exchange Variation of the Slav, on the one hand, and on the other, the necessity to learn a second opening, the Nimzo-Indian, which arises if Black begins instead with
1.d4 Nf6 2.c4 e6 and then White plays
3.Nc3 - making the desired Semi-Slav unlikely. Mind you, Sadler (
Queen's Gambit Declined, Everyman, 2000) warns "do you want to spend a lifetime learning the 4.e4 Marshall Gambit?" indicating that not a lot of work will necessarily be saved by making this choice.
Still, in
The Complete Semi-Slav (Batsford, 1994) Peter Wells says that "personally, I enjoy playing both sides of this fascinating gambit" and it was recommended, for instance, in Larry Kaufman's well-received repertoire book
The Chess Openings In Black And White (Random House, 2004). It's one of those openings that you feel you'd like to avoid when playing on the other side, but seem somehow not quite entirely trustworthy as a part of one's own repertoire: Kaufman considers his opening recommendations to be "fully respectable among strong grandmasters" but it certainly doesn't seem to crop up very much above 2650 level. Never mind, that's a higher rating than any opponent I've ever played (or so I believe - what was Tony Miles' rating when I played him in Wolverhampton about twenty years ago?) but even so, I do get the feeling, as with the Benko Gambit or the Chigorin, that one day somebody, or something, is going to find a clear-cut refutation.
Often, when White goes
4.e4 and thus takes up the gauntlet, play continues
4...dxe4 5.Nxe4 Bb4+ 6.Bd2 Qxd4 7.Bxb4 Qxe4+ 8.Be2 Na6 9.Bc3 and now it used to be standard to play
9...Ne7 which was considered absolutely fine for Black until a 1993 last-round interzonal game Lautier v M.Gurevich, given below, in which Black was dramatically demolished. (It's still played, but
9...f6 has attracted some attention given that Lautier's plan still seems strong.)
Lautier-Gurevich now continued
10.Bxg7 Rg8 11.Bf6 and now
11...Qf4, which used to be thought good, was burned up after
12.Bc3! Rxg2 13.Nf3 f6 14.Qd2! when the exchange benefitted White as the queen's absence made it hard for Black to defend his many weaknesses.
I'm not going to swamp this post with theory - Wells is a good place to look it up and if you do, you'll see he doesn't rate the alternative
11..Rg6 very highly. But Scherbakov thinks it's playable and follows a 1995 game he played against Berg which went
12.Bc3 Qxg2 13.Qd2 Qxh1 14.O-O-O Nd5 (not my normal style, this line)
15.Nf3 Qg2 16.cxd5 cxd5 17.Ne5 Bd7 18.Qf4 Rg7 19.Nxd7
and now instead of
19...Kxd7, which he played, he recommends
19...Qg5 20.Nf6+ Ke7 21.Qxg5 Rxg5 and claims that
"Black's position is quite acceptable..."
"...since after 22.Nxh7 Rf5 23.Bd4 f6..."
"...White should be careful about his knight."
As I say, I like Scherbakov's work, but in this instance I wonder how far he's gone to check his conclusion: if there was an Informator symbol for "Jimmy Hill" I'd insert one here. I don't believe it and neither does Rybka, which likes
24.Rg1! and likes it a lot.
The point is that if the king is driven away from the f6 pawn then the knight can take it with impunity and doesn't have to be careful at all. Black's obvious response is
24...e5 which shields the pawn from the bishop and therefore allows the rook to defend it, but after
25.Rg7+ Kd6 it turns out that White's bishops drive the rook away:
26.Bg4 Rf4 27.Be3 and it has to move from the f-file. So
27...Rc4+ 28.Kd2:
and the best I can produce on Black's behalf is "fighting for a draw". Those bishops are really pretty good.
That's not a forced line but I've been unable to find any deviations which help the Black cause - Black can take a queenside pawn, or play an early ...Rc8+, or try ...Rh8 attempting to tie up the White pieces a little, but in every case it looks like Black's position is under too much pressure to be held.
Now it's true that this is Rybka-led analysis and that computers aren't always the best judges of positions in the Marshall - as you'll see if you use one to play through the earlier moves and watch the evaluation swing about. But still: in the absence of any contradictory evidence the position after
23...f6 (though not the Gambit as such, of course) looks like a busted line to me. And I do wonder - given that the computer comes up with
24.Rg1+ immediately, was the original conclusion, that "Black's position is quite acceptable", put to any real test?
