White gets an open position. Why are you basing your opinion on GM game percentages?
Why do people play 1 e4 e6 2 d4 d5 2 exd5?!

Because they have tried 3.e5 a dozen times without knowing any theory and get tired of losing the d4 pawn in every game.

21.5 for white 26.5 for black and 52.0 for draw
You already answered your own question. The problem seems to be that you wrongly assumed that every player believes drawing is a bad thing.

1.e4 e6 2.d4 d5 3.exd5 is not even drawish.
The high percentage in draws is because people use it for preagreed draws.
For example.The following game played on the last round of the semifinal of the USSR championship.Both players had already qualified for the finals but quick draws were not allowed in Soviet championships.
This recipe is well known among tournament players many years now.If you want a quick and easy draw and noone to suspect that it was prearranged just play French exchange and exchange everything.
Played the last round too.The player with white needed the draw to secure second place(he couldn't be first even with win), the player with Black needed the draw to secure an IM norm.
Overall, a large percentage of the draws in French exchange were draws because both players wanted a draw.
There is no reason why a position must be drawish after only one pawn exchange.
All the more reason why Database Statistics are TOTALLY USELESS!
Another case in point: I am in an email correspondence tournament. In one of my games, I had played the following line (as White):
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.O-O Bg4 6.h3 h5 and now it's White to move. I played the move 7.c3 as opposed to the "main line", 7.d3.
If you look at 365chess.com, 7.c3 was played 209 times with an 88 percent draw ratio.
To conclude that a draw is inevitable is complete hogwash. Go further, and 26 games feature 7...Qf6, which is what my opponent played, and that has 53% of the games as wins for White!
Now let's say he had played 7...Qd3, which has an alarming 96% draw ratio amongst 176 games.
Problem is, go a level deeper, and 162 of those 176 games featured 8.hxg4, which has a 100% draw ratio because it forces a perpetual after 8...hxg4 9.Nxe5 Bd6 10.Nxd3 Bh2+ with repetition.
However, had my opponent played 7...Qd3, I would have played 8.Re1, not the drawing line. In the 13 games with 8.Re1, White scored 6 wins, Black 1 win, and 6 draws. That's only a 46% draw ratio, not an 88% draw ratio.
So don't just base move validity on database statistics without delving deeper and realizing that one draw line, like the case with the Exchange French, can severely skew the numbers.
It's horrible
That is probably what most 1.e4 players think of the French (opening)!

All the more reason why Database Statistics are TOTALLY USELESS!
Another case in point: I am in an email correspondence tournament. In one of my games, I had played the following line (as White):
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.O-O Bg4 6.h3 h5 and now it's White to move. I played the move 7.c3 as opposed to the "main line", 7.d3.
If you look at 365chess.com, 7.c3 was played 209 times with an 88 percent draw ratio.
To conclude that a draw is inevitable is complete hogwash. Go further, and 26 games feature 7...Qf6, which is what my opponent played, and that has 53% of the games as wins for White!
Now let's say he had played 7...Qd3, which has an alarming 96% draw ratio amongst 176 games.
Problem is, go a level deeper, and 162 of those 176 games featured 8.hxg4, which has a 100% draw ratio because it forces a perpetual after 8...hxg4 9.Nxe5 Bd6 10.Nxd3 Bh2+ with repetition.
However, had my opponent played 7...Qd3, I would have played 8.Re1, not the drawing line. In the 13 games with 8.Re1, White scored 6 wins, Black 1 win, and 6 draws. That's only a 46% draw ratio, not an 88% draw ratio.
So don't just base move validity on database statistics without delving deeper and realizing that one draw line, like the case with the Exchange French, can severely skew the numbers.
7...Qf6 is absolutely fine, actually it's precisely what Black should play if he wants to avoid the early draw.

1. You get an easy to play position and don't have to worry about French pawn chains...
2. If desired both sides can play for a win, for exampling by castleing in opposite directions.

All the more reason why Database Statistics are TOTALLY USELESS!
Another case in point: I am in an email correspondence tournament. In one of my games, I had played the following line (as White):
1.e4 e5 2.Nf3 Nc6 3.Bb5 a6 4.Bxc6 dxc6 5.O-O Bg4 6.h3 h5 and now it's White to move. I played the move 7.c3 as opposed to the "main line", 7.d3.
If you look at 365chess.com, 7.c3 was played 209 times with an 88 percent draw ratio.
To conclude that a draw is inevitable is complete hogwash. Go further, and 26 games feature 7...Qf6, which is what my opponent played, and that has 53% of the games as wins for White!
Now let's say he had played 7...Qd3, which has an alarming 96% draw ratio amongst 176 games.
Problem is, go a level deeper, and 162 of those 176 games featured 8.hxg4, which has a 100% draw ratio because it forces a perpetual after 8...hxg4 9.Nxe5 Bd6 10.Nxd3 Bh2+ with repetition.
However, had my opponent played 7...Qd3, I would have played 8.Re1, not the drawing line. In the 13 games with 8.Re1, White scored 6 wins, Black 1 win, and 6 draws. That's only a 46% draw ratio, not an 88% draw ratio.
So don't just base move validity on database statistics without delving deeper and realizing that one draw line, like the case with the Exchange French, can severely skew the numbers.
7...Qf6 is absolutely fine, actually it's precisely what Black should play if he wants to avoid the early draw.
pfren,
The point wasn't a knock on 7...Qf6 vs 7...Qd3, it was simply pointing out that White scores well after 7...Qf6 or after 7...Qd3 8.Re1, and that if you are playing the White side, while it may read "88% draw ratio" for 7.c3, going deeper reveals that those numbers are mis-leading because one other line, 7...Qd3 8.hxg4, distorts the numbers, and that 7.c3 is a perfectly viable line to play if you are playing for a win. In no way am I saying that 7.c3 wins, but also saying that playing 7.c3 does not equate to an "auto-draw" if White doesn't want a draw.
The same can be said about the Exchange French. As Anand once said, if White wants a draw, White has a draw. However, some will still play lines that are at an attempt to win despite the mis-leading high draw ratio.
Statistics aren't useless, unless you don't know how to read them. You make a valid point, prearranged draws contribute to the high drawing rate. Still it makes sense that the Exchange French produces more draws than let's say the Modern Benoni or the Najdorf Poisoned Pawn.
But the main misunderstanding by the thread opener is that black's higher win rate means he is better in this opening. Just a hint: Of all the games where the French Exchange has been employed, compare the average white players' elo to the average black players' elo.

I have no problem with the white player going for the exchange. Shouldn't it be a part of a French player's preparation - to study how games go in that variation?
It's like a Sicilian player complaining about the Alapin.

some people don't like the French or don't want to play into theory.
Exactly - also, people at all levels can appreciate not wanting to play into your opponent's preparation/knowledge.

As Black I like to play 1...e5 pretty much whenever it doesn't lose a pawn. But what about against 1.e3!? If White plays 2.d4 does Black have anything better than going into a French exchange with 2...exd4 3.exd4 d5
Also, the French Exchange, or at least the "Monte Carlo" variation can arise if White declines the Budapest with 3.e3: 1.d4 Nf6 2.c4 e5?! 3.e3?! exd4 4.exd4 Nf6. I used to play the Budapest, and once I realized this, I actually started playing the Monte Carlo against the French, on the grounds that I'd have to be careful in my assessments since I could wind up on either side.

As Black I like to play 1...e5 pretty much whenever it doesn't lose a pawn. But what about against 1.e3!? If White plays 2.d4 does Black have anything better than going into a French exchange with 2...exd4 3.exd4 d5
Also, the French Exchange, or at least the "Monte Carlo" variation can arise if White declines the Budapest with 3.e3: 1.d4 Nf6 2.c4 e5?! 3.e3?! exd4 4.exd4 Nf6. I used to play the Budapest, and once I realized this, I actually started playing the Monte Carlo against the French, on the grounds that I'd have to be careful in my assessments since I could wind up on either side.
The scary thing is that there are a number of openings that can lead to "Exchange Variations" of other openings.
In addition to your 1.e3 scenario, I had 2 Exchange Frenches in the 2014 US Open, one against a player almost 1400, the other against a 2150. I won both games. The caveat? I didn't even play the French either time. I started out with a Petroff all 5 games where I had Black, and while rounds 3, 7, and 9 all transposed to the Four Knights Game, rounds 1 and 5 were a French Exchange.
The French Exchange can be reached via the Petroff after 1.e4 e5 2.Nf3 Nf6 3.Nxe5 d6 4.Nf3 Nxe4 5.d3 (instead of the main lines, 5.d4 or 5.Nc3) Nf6 6.d4 d5 and you have the position after 1.e4 e6 2.d4 d5 3.exd5 exd5 4.Nf3 Nf6 with two extra moves having been played by each player!
So even if you don't play the French Exchange, it's probably good to know it, unless you really do play something that has no chance at transposing, like the Najdorf Sicilian.
By the way, when I mentioned "Exchange Variations" of other openings, a couple of other examples come from the Old Benoni lines (1...c5 instead of 1...Nf6 and 2...c5). After 1.d4 c5 2.c3 cxd4 3.cxd4 d5, you have an Exchange Slav and after 1.d4 c5 2.e3 cxd4 3.exd4 d5, you have an Exchange Caro-Kann.

Honestly, I don't understand why people play the exchange variation. As Hikaru and many others have said many times 'White is the one that has to push for the win'. I don't even think it is a viable way to secure a draw, in a lot of lines black can castle queenside and start a attackon whites king. Can someone tell me why people play the exchange? I just don't get it

Honestly, I don't understand why people play the exchange variation. As Hikaru and many others have said many times 'White is the one that has to push for the win'. I don't even think it is a viable way to secure a draw, in a lot of lines black can castle queenside and start a attackon whites king. Can someone tell me why people play the exchange? I just don't get it
I've seen White castle Queenside as well before. It's not a 1-way street.
In no way do I condone the Exchange Variation, and I'd never play it as White as I feel White's only two chances at an advantage on move 3 are Nc3 and e5, but it's not all sh*ts and giggles for Black as though an advantage is automatic by any stretch of the imagination.

Honestly, I don't understand why people play the exchange variation. As Hikaru and many others have said many times 'White is the one that has to push for the win'. I don't even think it is a viable way to secure a draw, in a lot of lines black can castle queenside and start a attackon whites king. Can someone tell me why people play the exchange? I just don't get it
I knew a girl who had a very limited opening repertoire. As a matter of fact, I had trained her for a few months in 1995. She was always playing the same "meek" openings: Italian game, French Exchange, Alapin Sicilian, Caro Kann Panov as white, and Sicilian Sheveningen (rarely the Petroff), King's Indian Defence as Black. She never, ever played anything else as white against the French: she was pretty familiar with IQP positions, the only variety was 3.exd5 exd5 4.c4, or sometimes 4.Nf3 Nf6 5.c4. She never bothered to study openings seriously, and she was frequently lost, objectively, right out of the opening.
Well, she never achieved something more than becoming World Junior Champion U-20, and WGM. Here is a rapid game she has played:
And yet, another one, against a low rated kid:
I have lost the scoresheet of the only game I have played against her. I was white, and she played a very dubious line of the KID. The position got very complicated, I miscalculated, got the worse of it, and lost.
To cut a long story short: Who cares about the opening? Just get a playable position, and play on!
It's horrible