The Slav Dilemma

3...dxc4 is not the main line...

AFAIK most Slav players go for a Chebaenko setup against 3.Nc3:

Play_e5: The Botvinnik Semi-Slav isn't a very good option for players until they reach very high level (I would say somewhere around E2250-2300). Just like the Sicilian Dragon (not talking about the Accelerated varation) there's far too much rote-learning involved. You have to know certain lines 30+ moves deep or more, and if the opponent deviates, you have to know those sidelines 30+ moves deep, too. It's great fun playing over games in the Botvinnik Varation (and the Dragon) because the games are completely crazy, but let's remember these variations have been developed by the likes Tal, Kasparov, and Shirov - they're able to work out deviations over the board. Lesser mortals, not so much ...

Up to E2200, we're better off focusing on other aspects of the game. General tactics, calculation and strategy, middle- and endgame play, the theoretical endgames which are great to know in a tournament situation etc.

"It's the engine's best move" so? where's your point? You can't blindly trust engine evaluations in the opening, especially when GM practice seems to contradict what the engines are saying. Don't get me wrong, 3...dxc4 is a good move and one I played in the brief period I tried out the Slav. However, my point is that it is after all a sideline, so you can't base your main repertoire recommendation on what to play against something that isn't the absolute best move. It's like saying "I play the 2...Nc6 Closed Sicilian as black" - OK, but what do you do if white plays the critical Open Sicilian lines? You need to give something against the Chebaenko setup I showed.

Should that really be the object for a player trying to improve or a lower-rated player against a higher rated player? There's absolutely nothing wrong with the Exchange Slav, but the nuances are too fine for lower rated players. I would say that playing Exchange Slav (or French) as a lower-rated against a higher-rated only risks exacerbating the difference in strength between the two.

my137thataccount: Agree wholeheartedly. The engine evaluation matters very little. All coaches tell you that you shouldn't really use engines in that way, it's not good for your chess. The way an engine plays chess is vastly different from the way we humans play.

I play 1.d4 only from time to time, so learning a full-fledged line against the Slav was out of the question. So I settled for 1.d4 d5 2.c4 c6 3.Nf3 Nf6/e6 4.Qc2, often continuing with g3 and trying to play a kind of Catalan where Black did not hoose the sharpest option. The problem moves are early ...g6 (intending Bf5) and early...c5, but in any case you get a playable middlegame. There is also a sytem combining Qc2 with Bg5, slightly more energetic and requiring a bit of preparation.

This is an interesting option. There's also the option of the actual Catalan Slav, which I don't necessarily recommend as it's a bit trickier than the regular Catalan, but nonetheless it's an option.

Catalan Slav looks cool.  Great name too.

I used to play this line too. As you say, the crucial line is the one where Black plays g6. My personal feeling is that 4.Qc2 is like the Breyer Varation 4.Nbd2 (similar idea, Catalan development with kingside fianchetto, Qc2, Rd1, and e4 and recapture on c4/e4 with the knight should Black play dxc4/dxe4). They're both fine for occasional outings, but if you are going to play 1.d4 regularly, you will need at least one option which follows more well-trodden paths (such as 4.e3 which I suggested earlier or 4.g3 as my137thataccount proposes). Both 4.Qc2 and 4.Nbd2 suffer from the same problem. They work perfectly if Black plays the Semi-Slav via the Slav move order. However, if Black develops the light-squared bishop outside the pawn-chain, he has easy equality. I used to play both of these variations, and the basic problem is this that if you're likely to face the same opponent a second or a third time, or if your games can be found in a database, your opponent is highly likely to have swotted up on the right set-up (4.Nc2 g6 and Bf5/Bg4, 4.Nbd2 Bf5 or even Bg4-h5-g6).

https://youtu.be/-AdSN6CpuWQ

And the weird 3.Bf4 is definitely not the London System

Basically, 4.e3 and 5.Nbd2 are aimed at the same purpose: Prevent white from taking on c4 and introducing the structural complications peculiar to the Slav. 

So we're looking Here:

Here:

And here:

I love that Deirdre included games with her suggestion! I've analyzed two of them....some suspicious play by black, but my opponents (and myself) also play suspect moves. Older games definitely feel more down-to-earth than a lot of these otherworldly 2700 games.

This is the line I mentioned, when I originally recommended the Slow Slav. When Black counters this in Semi-Slav fashion, you develop the queen's knight to b2 and the dark-squared bishop on b3 followed by Ne5 and f4, followed by Qf3 or Ndf3. It's a great line for White. But also against Bf5, Bg4, and g6 the Slow Slav works fine.

You didn't show any games where black plays Bf5 before e6

I was going to the make the same point as you, DeirdreSkye. We have to remember that most players up to somewhere close to E2200-2250 strength are better off taking a slightly simpler approach to the openings in order to focus on strengthening other aspects of the game. We're no by no means claiming it to be a silver bullet against the Slav, but it takes a lot of pain out of preparing against it.

I might try myself the slow slav, but I think that I will stay with the xchange. Maybe the Bf4 e3 lines with delayed kings knight development. Will take a look at it. 

As I said before, beware of this line:

I study these by going through whole games and committing them to memory (as much as a 1400 can memorize...), so the process is slow going, but I found this line by a very strong player named Ivanisevic (who I'm ashamed to say I hadn't heard of) playing a 2700 called Ragger (who I ALSO hadn't heard of!_

It's a line of the Grunfeld Defense.

Seems like a roundabout way to get a Gruenfeld....in traditional move order, it looks weird