I always liked the gambit 1.Nf3 d5 2.g3 Nf6 3.Bg2 c6 4.c4 dxc4 5.O-O but since 1996 there is a problem in the line 5...Bf5 6.Na3 e5 7.Nxc4 e4. When Valeri Salov was confronted with this he chose the dubious piece sacrifice 8.Ng5 h6 9.Nxf7 Kxf7 10.Qb3 (Salov - Piket, 1996) because both players thought that Black gets a fine position after 8.Nfe5 Bc5 9.Qb3 Qe7.

I agreed to this and abandoned the line.

But yesterday a stumbled upon following game where White got a very nice position after 10.d4 exd3 11.Nxd3 Bxd3 12.Qxd3 and maybe 12.exd3 is an improvement.

[Event "Seville Villa de Coria op 9th"]
[Site "Seville"]
[Date "2003.02.13"]
[Round "8"]
[White "Matamoros Franco,Carlos S"]
[Black "Vasquez,Rodrigo Rafael"]
[Result "1/2"]
[Eco "A11"]
1.c4 Nf6 2.g3 c6 3.Bg2 d5 4.Nf3 dxc4 5.0-0 Bf5 6.Na3 e5 7.Nxc4 e4 8.Nfe5 Bc5
9.Qb3 Qe7 10.d4 exd3 11.Nxd3 Bxd3 12.Qxd3 0-0 13.b3 Na6 14.a3 Rad8 15.Qc2 Bd4 16.Rb1 Nc7
17.e3 Bb6 18.Bb2 Rfe8 19.Rbd1 Ncd5 20.Nxb6 axb6 21.Rd4 b5 22.Rfd1 Nc7 23.Rxd8 Rxd8 24.Rxd8+ Qxd8
25.Bf3 Ne6 26.h4 h6 27.Qd1 Qc7 1/2

I was quite exited for some minutes until I asked myself how I'm going to refute 10...Bxd4. Of course an unprepared opponent might shy away from the complications after 11.Nd6+, but what if he doesn't? I analysed the position several hours but could not find anything really convincing. After 11.Nd6+ Qxd6 both 12.Nxf7 Qe7 13.Nxh8 Na6 and 12.Qf7+ Kd8 13.Qxg7 Qxe5 are at least slightly better for Black so I finally came up with the Idea 11.Rd1!? Bxe5 12.Nxe5 O-O 14.Nc4 to seek compensation for the pawn on the dark squares. I believe it's good enough for equality after for instance 14...Be6 15.Bf4 b5 16.Qa3 Qxa3 17.Nxa3, but I don't feel attracted by this.

Can anyone help me out with a refutation of 10...Bxd4 or is 10.d4 only a bluff?

You now that this is a job for Rybka, rigth?
11. Nxf7 doesnt seem to work
I dont like 11.Rd1 either (White may have compensation there but nothing more.), so I think you should look in the 11.Nd6+line.
After 11-Qxd6, does 12.Qxf7+ work?

I also asked slightly outdated versions of Rybka and Fritz. After 11.Nd6+ Qxd6 12.Qxf7+ Kd8 I analysed 13.Qxg7 Qxe5 14.Qxb7, 13.Qxg7 Qxe5 14.Qxh8+ and the immediate 13.Qxb7. White always ends up with rook + 1 or 2 pawns versus 2 pieces and Black is well coordinated. The machines think it's only a little bit better for Black but I think they might underestimate the difficulties.

Heron,

I've also found a game where White played 9.d4 instead of 9.Qb3. In the cited game White possibly didn't continue with the best (9...Bxd4 10.Qb3 transposing into the realms of your line).

However, the silicon friend says 10.Be3 when White will have compensation on the dark squares, viz.

10...Bxe3 11.Nd6+ Ke8/f8 12.Nexf7 (+3.00)

10...c5 11.Bxd4 cxd4 12.Qa4+ Kf8 13.Qb4+ Qe7 14.Nd6 (+1.00) where White will regain his pawn and be much better.

10...Bxe5! 11.Qxd8+ Kxd8 12.Nxe5 Be6 (=) where the machine sees enough compensation. I'm not going deeper here, but this assessment might be correct, because Black is not well coordinated and has to take care of e4.

After 9...exd3 play continues along your lines. Especially nice is that Black can't castle after 10.Nxd3 Bxd3 11.Qxd3 Qxd3 12.exd3 because of 13.Na5 with a big advantage for White.

Many thanks for the game. I also think that White has sufficient compensation after 9.d4 Bxd4 10.Be3 Bxe5 11.Qxd8+ Kxd8 12.Nxe5 Be6 13.Rfd1+, but 10.Bf4! threatening 11.Nxf7 might be even better, e.g. 10...O-O 12.Nxf7 Rxf7 13.Nd6 and White gets his piece back with advantage (Bd4 is hanging too), or 10...Bxe5 11.Bxe5 O-O 12.Nd6 with much pressure.

So it looks like 9.d4 is really strong and I will certainly play it if I get the opportunity.

This gambit-line is a big problem for white, but the moveorder 1.Nf3 d5 2.g3 has an even bigger theoretical problem and that is 2....Bf5 right away. The reason is 3. c4 e6 4. Qb3 Na6 and black equalizes as all complications turn out to be in black's favor or simply force a repetition (check it with Rybka). Therefore many people play 1. Nf3 d5 2. c4 but here black can simply equalize with d4 and transpose into a Schmidt-Benoni with colors reversed (note: black should play it with an early a5).

Besides that, I doubt that anyone was able to come up with anything convincing against the hedgehog after 1. Nf3 c5 2. c4 Nf6 3. g3 b6 etc. so far.

So my suggestion is: Forget about the "tricky" Reti moveorders and either play 1.d4 or 1.c4 right away or at least go with 1. Nf3 d5 2. d4(!) like Kramnik did for many years.

Is this a general problem like "I have unclear compensation and if I run out of ideas I'll be simply a pawn down" or do you think of something concrete?
This only proves that you can't refute 1...d5 and 2...Bf5 and that's not surprising because both moves are within reason. Instead of 3.c4 White should play 3.Bg2 and 4.O-O first and then choose between c4, d3 and b3 depending on Black's 3rd and 4th moves. The idea is to outplay him later in the game.
White can seek adventures with 3.e3 Nc6 4.b4!? but in general I do agree.
I think a complete player should be able to play both sides of the hedgehog.
Why do you recommend 1.c4 when you have so much respect for the hedgehog? Right now I'm building a repertoire around 1.c4, but I never dreamed of opening advantage, only of interesting positions that I possibly understand better than most of my opponents. A few months ago I spoke about this with a FM who plays 1.c4 (and nothing else) since 30 years and after only a little persuasion he eagerly admitted that White has little reason to hope for an opening advantage after 1...c5, and there are lots of other good defences as well, of course.

I used to play the line with 4. Na3 and analysed it very deeply with Rybka (even testing virtually every legal move). After 4...b5 white at best gets his pawn back with an equal game and it is not even clear. Besides that I couldn't refute 4...Qd5 either. After castling on move 4, I couldn't refute the line with Nbd7 even though it was given as good for white in Informant 78 (latest game: Radjabov-Smeets, Corus 2009).

The slow version is known as extremely drawish ever since Botvinnik-Smyslov 1958. You can find thousands of games in the database and nobody could ever prove something, not even in the tricky moveorder starting with 1.b3 where white is a tempo up in many cases. In my opinion, every time when black manages to play d5, Bf5, e6 and c6 (London system reversed) and doesn't get punished for it, he has an absolutely equal game.

There is one line (pointed out by Larsen) where it doesn't work: 1. Nf3 d5 2. e3!? Bf5 3. c4 e6 4. Qb3 Nc6 5. cxd5 exd5 6. Qxb7 Nb4 7. Bb5!

1.c4 gives more options to deal with 1....c6. It also gives more options after 1...e6 as you can play the QGD exchange variation with Nge2. The hedgehog can be avoided with 2.g3 (there are games by Seirawan), but that doesn't give much of course.

Marin has a book coming out soon (vol. 2 of his series), so that will be interesting.

Very simple:
1. Nf3 Nf6 2. c4 c5 3. d4 cxd4 4. Nxd4 e5!
1. Nf3 Nf6 2. c4 c5 3. Nc3 e6 or d5
1. Nf3 Nf6 2. c4 c5 3. g3 b6 4. Bg2 Bb7 5. Nc3 e6 or g6
1. c4 c5 2. Nc3 Nc6 3. Nf3 e5 or Nd4 or Nf6 and if g3 then d5
1. c4 c5 2. Nc3 Nc6 3. g3 g6 4. Bg2 Bg7 5. Nf3 e6 or e5

In many cases black even has multiple options...and they are all safe! These variations are just the starting points for research of course, but if you follow them up with games on the highest level you will see that there is nothing. Well, the same could be said after 1.e4 e5 and black playing the Marshall Attack, but that is a different story.

My basic philosophy is this: If I want an equal game with white I play 1. d4 and 2. Bf4 and so on. All pieces will be on good squares with lots of play left and I am sure that I cannot run into a gambit or some trap. When I put up the effort to analyze many variations 20 moves deep they better give me at least a tiny advantage in return, otherwise it doesn't make sense. Since 1987 (Informant 48 or so) I am analyzing openings, even with computers (chess machine, Fritz 2 etc.), one of my findings made it to Informant 74 and two others got published by NiC, but I have not made any progress. 22 years later I am still at stage one: White has nothing. Capablanca was right!

It's funny how one sentence can turn a wonderful post into a nuisance (-:

The concept is that black cannot stop white from setting up a rock-solid formation. The other way around it is not possible. If you trust that position or not is another story.

http://www.chessgames.com/perl/chessgame?gid=1040438

Nobody will dispute the solidity of the London but there is the problem of boredom. Over the last 25 years with White I almost exclusively played things like b3, g3 (with or without preliminary Nf3) and Queen's pawn openings like Colle and Zuckertort. By this I'm trying to prove that I'm not bored easily. This spring I bought and worked through "Win with the London System" by Sverre Johnsen And Vlatko Kovacevic, and played it very often since then. The results are OK, mainly because of all those KID and Benoni players, who feel uncomfortable with the structure, but, after only 5 months or so, I simply can't see it any more.

I think 1. d4 2. Bf4 merits a new thread. I'll start one.

That's a concept that's easy enough to understand.

I view the London as more a psychological weapon, since most of White's points rather have their origin in Black's losing patience at some point than in the chances for an advantage offered by the setup... it's almost equivalent to being an über-rock at poker...

I divide openings into two cathegories:
A. Openings that fight for an advantage
B. Openings that are designed to reach a certain structure or set-up

Many players have noticed that type A openings are very complicated to memorize and even if they do it, it is not even clear if it gives them the advantage. For instance, you can play the pawn sac line in the Gruenfeld with Nf3 and Rb1, but it is not clear if you "win by force" with it. Obviously you can't, because otherwise the Gruenfeld would have gotten the stamp "refuted" a long time ago. So it all becomes a question of how complicated a position should be in order to give the opponent the most chances to make mistakes.

Since the "keep it complicated" approach may backfire easily (too complicated for yourself), many players ,especially at the intermediate level, resort to type B openings. The most popular version is the Kings Indian Attack. You play some autopilot moves that black cannot interfere with to reach a certain set-up and "go from there". Now even there is a certain risk level to these setups, but one with the lowest risk potential is certainly 1. d4 2. Bf4. Now don't get me wrong, I am not trying to advertise this opening. I never played like that myself.

Psychology is a totally different question. Of course it is possible to frustrate opponents with your choice of openings and it may cause them to become overaggressive. This doesn't change the theoretical evaluation of the chosen line though.

Btw, there is a type C of openings and that is the attempt to reach an advantage by playing low risk openings while exploiting moveorders. 1.Nf3, 1.g3 or 1.b3 may all lead to the Kings Indian Attack eventually, but every move has pros and cons. Larsen has been viewed as the off-beat player of the past, very much like Morozevich today, but all he did was to experiment with moveorders. Kortchnoi experimented with various KIA set-ups also, as did Bagirov. This branch of openings is very interesting to a researcher, at the same time nothing can be more frustrating than recognizing that it doesn't work.

The big terror or all closed openings is the slav. Now what is so terrible about it? The slav is just another moveorder to reach the London system! Just look how Dreev is playing these setups. If black gets the chance he will develop his bishop to f5. Once you recognize that, you know the goal for white: This set-up must be prevented! So basically we are back at square 1....