This question led me to spend two hours of my life looking at a line that I most likely never will get to play. I went into it being quite confident that
Nd5 was a bad move simply because White would get too much development for the cost of a simple pawn. Analysing back and forth with Houdini has nevertheless convinced me that
Nd5 is perfectly playable.
The similarities to the Sokolsky Opening
To get a feel for the
b4 pawn it helps to understand a variation from the Sokolsky Opening. It goes like this:
1. b4 e6 2. Bb2 Nf6 3.b5
Here the pawn on
b5 is slightly annoying. It cannot be easily removed as
a6 can be answered by
a4 (it does not matter if the rooks are swapped off as White intends to castle short anyway) and
c6 can be answered by
e3 when Black must take away from the centre to remove the pawn.
In the game, however, Black can simply play
a6 in the similar position as White can't answer with
1.f4 Nf6 2. b4 e6 3.b5 a6
2...e6 White is only left with
3 a3, which, although reasonable, is another pawn move on the wing. Black will be very solid after playing simple development like
0-0. This is slightly better than one of the main lines of the Sokolsky Opening, which is not that good to start with (but where White often ends up playing
Houdini 3 on
The currently strongest engine on the planet is Houdini 3. When you let it analyse the position before
2...Nd5 to a depth of 25 it likes
2...d6 (claiming a small advantage for Black) the best (with
2...d5 as a strong second).
2...Nd5 does not come up as a top five alternative. Houdini is initially not impressed by grabbing the pawn.
2...Nd5 is still an interesting alternative and certainly gives more entertaining variations to look at. First let us take a look at what happens if White plays
3 e4 after
2...Nd5 and then Black takes the pawn at
1.f4 Nf6 2.b4 Nd5 3.e4 Nxb4 4.c3 N4c6 5.d4 d5 6.Nd2 dxe4 7.Nxe4
To explain this variation it is important to know that Black's hope and main idea against a setup with f4-e4-d4 is to play
d5 and develop
g4. In the variation above it turns out Black can't play either.
Bg4 simply hangs the bishop and
Bf5 runs into
Ng3 chasing it back. Here I (and Houdini) believe White has compensation for the pawn (although perhaps nothing more).
So, let us look at the alternative. Initially I was sure that
3...Nxf4 must be horrible for Black, but it turns out it is good after
4. d4 e5!:
1.f4 Nf6 2.b4 Nd5 3.e4 Nxf4 4.d4 e5 5.dxe5 Ne6
As seen above, if White makes the natural capture on e5, Black just retreats the knight. Material is now even, but Black has two threats; The killing
Qh4+ and just grabbing the
b4 pawn. So Black will end up a pawn up with a much better pawn structure and no imminent attack from White.
Instead White must play:
1.f4 Nf6 2.b4 Nd5 3.e4 Nxf4 4.d4 e5 5.Nf3 Ne6 6.d5 Nd4 7.c3 (7.Ng5) Nxf3+ 8.Qxf3 a5
In the above variation please note that
6 Nxe5 runs into
6...Qxh4+. The same idea works against
7 Nxd4. And despite all the knight moves Black is left clearly better in this variation too. If you want to play this variation with White you need to try the chaotic
7.Ng5!?, hoping to trap the knight at d4. Have fun analysing that.
3 d4 is a strong alternative for White. If Black takes the pawn on
b4 it transposes to a variation we have already looked at:
1.f4 Nf6 2.b4 Nd5 3.d4 Nxb4 4.c3 N4c6 5.e4 d5 6.Nd2 dxe4 7.Nxe4
So, as I mentioned in a comment to the question, perhaps Black can try the amazing
3...Nf6 with the simple plan of
Bf5. In this variation White will have more space, but also more weaknesses (and, no, Black is not behind in development):
1.f4 Nf6 2.b4 Nd5 3.d4 Nf6
Seeing that White will lose one pawn anyway, why not play a development move like and see which pawn Black decides to take? It seems that Black can just grab the
f4 pawn and get away with it. Let us see what happens if White tries to push Black off the board:
1.f4 Nf6 2.b4 Nd5 3.Bb2 Nxf4 4.e3 Ng6 5.h4 h5 6.Be2 d5 7.Bxh5 Qd6
White won the
h5 pawn, but now the threat of
Qg3+ is just killing White. Instead, if White tried some quiet development moves:
1.f4 Nf6 2.b4 Nd5 3.Bb2 Nxf4 4.e3 Ng6 5.Nf3 e5 6.Nxe5 Nxe5 7.Bxe5 Nc6
Black gets a small advantage by the precise
5...e5 which trades the not so optimally placed
Ng6 (Black is still a pawn up). Note that the alternative
3 Nf3 could easily transpose as it does not seem that White can improve much
2...Nd5 works out well for Black, despite Houdini's initial reluctance towards it. It seems White should play
3 d4 to equalise. I would certainly try
1 f4 Nf6 2 b4 be played against me.