e4 is a good move because it opens up a diagonal for white's lsb. What about Nf3 though?

I wanted to preserve the option of playing f2-f4. Nf3 doesn't score too well on the opening databases.

In the Benoni, black often has issues finding a good square for his light-square bishop (because white can establish a grip on the light squares with pawns on d5 & e4), and playing an early Nf3 allows black to solve that problem with ...Bg4 and ...Bxf3.

Too early to say "this is completely wrong". White, by not playing c4, is allowing black to use the lsb to support a possible pawn storm on the queenside, made stronger by the dsb once it is fiachettoed. Also, by not playing c4, white is leaving black the possibility to play c4, that can be easily supported by Qc7 or Qa4, a6 - b5... and bla bla bla...  

I would probably have played 3.Nf3, 2.e4 or 2.c4 rather than this knight move. It just looks counter-intuitive, to me.

e4 is still playable here because e4 is supported by the knight on c3.

I think Nf3 is inaccurate (see my previous comment). In similar Modern Benoni positions, I like to play an early h3 to keep the black LSB out of g4, then build up the center with f4 and Nf3 etc.

Nf3 or perhaps Bb5+ although as the d1 to a4 diagonal is open to the BQ one must watch out for queen forks on the bishop and the opposing king, particularly if the c3 knight is diverted elsewhere!

More and more I will play with the principle: "do not block the own bishop with central 6 pawns". I know it is possible, but my, I don't like it!  

What was wrong with BXB+?

As white has a space advantage he wants to keep the pieces on the baord. BxB only leads to equality, white wants to preserve his slight plus. Look at the opening databases if you don't believe me.

Also, black's LSB is a liability, so why do black the favor of exchange it?

Joanne, I think the point is to force the pawn on f7 to take back, so to open the king's position. By BxB+ QxB, and now if d5xe6, Qxe6. Ok, you can make a lot of considerations here. If BxB+ and QxB white DOESN't take in e6, and black has to lose time making room for his b8N to develop, so white wins a tempo. But black can go out on Na6, since the white's lsb is not there anymore to take it... etc. etc.  

Over the next couple of moves, note how the black LSB is simply getting in the way, and not doing anything active.

I was looking at 7...b5 here. It seems to rid Black of that light square bishop at any rate: 7...b5 8.Bxb5 Bxb5 9.Nxb5 and Black can retake on e4 or just continue developing by ...Nc6. The move played allows time for White to play his thematic 8.a4 advance.

I have to agree

and challenges a possible b5.

and clears the way for kingside castling.

Just to be a little annoying...   we were told that Nf3 is not the best since you want to keep the possibility to play f4... and now, all of a sudden...  )) Of course this is a good move, in my opinion. But f4 would have been a plan: just planning f5, forcing black to open the diagonal.

The situation has matured a little since that observation in re f2-f4. The knight move played is not merely a developing move: it prevents Black's threatened 9...Ne5.

Yes, and also keeps it out of d4

0-0.

The bishop on e7 is potentially overloaded.

Ba2 which, as in your previous note keeps the bishop on that a2 - h7 diagonal.

11.Be2 seems a reasonable alternative.