e4 e5 Nf3 Nf6 Nc3 Bb4

Do you think this is a good way to avoid the Four Knights?

What would be the main ideas for Black here?

That's fine. It's more similar to the Petroff. I personally prefer to play d6 d4 Nbd7. But in your line, you have to be willing to lose the bishop pair. You should look into the Spanish Defense, Berlin Defense, Berlin Wall variation. You'll be playing that reversed.

So the idea is usually to double White's pawns like the Nimzo-Indian?

If you do play the line Charetter115 suggested, be aware of the gambit 5.g4!? which gives White a lot of tricks.

Some Petroff books will cover this line, mentioning to the reader that the Four Knights is a stronger option for Black, mentioning that Black's position ends up extremely passive in the line the OP posted, but that it's "playable".

If you are a 2...Nc6 player, I have a simple solution for you, and you can also use this against the Scotch Game.  It's known as the Three Knights Game.  It goes 1.e4 e5 2.Nf3 Nc6 3.Nc3 g6 4.d4 exd4 5.Nxd4 Bg7 and play deviates from here.  You can also get there from the Scotch Game after 1.e4 e5 2.Nf3 Nc6 3.d4 exd4 4.Nxd4 g6 and White has nothing better than 5.Nc3, which transposes.

For Petroff players, if you are looking for something active, I can tell you when I went thru my Petroff stint, a lot of people played the Exchange French (3.Nxe5 d6 4.Nf3 Nxe4 5.d3 Nf6 6.d4 d5) or the Spanish Four Knights (3.Nc3 Nc6 4.Bb5).  The former is fine for Black as White has committed to the 4.Nf3 Nf6 line of the Exchange French which is great for Black.  The latter, if you want something really active, 4...Nd4, the Rubinstein Variation, is a strong line for Black, but he must know what he's doing, where one mistake could cost him, but with correct play, he gets an excellent game!

HAHAHAHHAHAHAHAHAHAHAAHAHA

The OP is a Petroff player

The powerful move 3.Nc3

I have been saying it for years!

You are afraid. I see the fear in your eyes and you should be afraid.

                                    Halloween Gambit                                       

MWWWAHHAHHAHAHAHAHAHAHAH

What? Firstly, the Halloween gambit objectively loses (oops, I fed the troll). Secondly, how does this relate to the OP's question?

The line itself can't be bad, Jussupow has played it. Apparently, after Nxe5 0-0, White can't hang on to the pawn. From what I can see, Black just castles, puts his rook on e8, pawn on d5 and develops normally.

It's more solid. White's queen's knight can be dangerous, black taking with the bishop will kill everything to do with it and white has no good way to attack. 

In the laboratory the gambits all test unfavorably, but the old rule wears well, that all gambits are sound over the board.  -  William Ewart Napier

Halloween gambit objectively loses.

What is your continuation to beat it! If you do not have one than it seems it is objectively winning

Can White basically force a dull drawish game in all e4 e5 lines? This is excluding risky lines like the Lativan gambit or 3...f5 in the Ruy which I find both to be unsound.

It's not more commital though, it's a well-known extremely solid drawing line. If someone sighs at you for playing the petrof, they'll super-sigh if you play this line. As I said, you take the queen's knight and take the teeth out of white's attack.

Edit: I tried to put in 6. Bg5 as the last move in the diagram. 

And what's white going to do? Without the queen's knight and with the doubled pawns on the queenside white has spurious advantage. 

I just put it into houdini and have to admit I don't understand it... houdini says black is winning by -0.15 after white's last move! Well I don't know why, I thought that looked the most provocative move, but white has nothing no matter what he moves. Of course 3. Nc3 is not a very serious try for advantage. 

Then what about Nd5? I don't know how that move would work out, it looks very complicated and isn't part of my repertoire.  

Also b4-b5 is a reason for playing Nbd7 instead.  

I like to play 3...Bb4 because many people especially Beginniers are comfortable with the Four Knights and the Bb4 move will sometimes confuse them but to be honest the Petroff confuses some people too. 

Ok maybe not unsound but anytime you push your f-pawn your king gets weakened which can cause trouble. 

Isn't that the Schlieman... which has mountains of analysis devoted to it and routinely seen even at the highest levels? Something that cannot be said for the Latvian lolgambit.  

I thought it would be more popular if it was really good though. 

The line is pretty much OK, Sakaev advocates it in his Petroff repertoire book as good and simpler than a transposition to the 4Knights.

From here he says that only castling gives Black any real trouble, instead of the more common 8.Nf4 (preventing ...d5) which gives Black an easy game. It seems that even 8.0-0 gives white very little to boast about- white has the bishop pair, but no targets to attack, and Black has very smooth development.

That is what they all say until a player does it to them OTB.

Notice a game I played below.

My opponent resigned in this position. They asked me where they went wrong. I had to tell them 2...Nf6 was an inaccuracy.

2...Nc6 is the best move in the position.

This line Bb4 is the reason why nobody (almost) play Nc3 at high level.

4 knights is theorically speaking ok for black but Bb4 brings more comfortable play and easier than in 4 knights which can be tricky.

1.e4 e5 2.Nf3 Nf6 3.Nc3 Bb4 4.Bc4 d6 is completely fine for black. Later he will castle, maybe Bxc3, Be6 to trade bishops, with an equal game.

Black can also play with active play, like this high level game (ok its blindfold but top GM even in blindfold are great) :

I think the real test is 4.Nxe5 but now black can simply castle, and will play Nxe4, or Re8 next, with simple and easy equality.

This is more simpler than a 4 knigh game, and it brings some imbalanced play, the position is not symmetrical so u have more chances to win this for black than a "classical" 4 knight. Bb4 is a more useful move than Nc6, u speed up your castle, so important in 1.e4 e5.