All correct, except that Williams' book has some flaws as well. The merits of ...Ne4 are well explained by Williams. Personally I am not so sure, if the exchange always benefits Black. The question is, which is the stronger piece, Black's king's knight or White's queen's knight? This question is hard to answer. Two objections against the knight's exchange, is that it takes precious time. Black is in a very concrete danger to get behind in development. The other one is, that the knight is missing for an eventually attack against the white king. I also have noticed, that Williams himself is not doing very well with ...Ne4 lately.

Now let me try to help poor Inn2  . That game Belov-Williams is not that critical at all. Black should have played 16...Bxg2 17.Qxg2 Ng4 with a reasonable game.
Now I am going to be pretentious and try to explain better than Sadler.  Originally the move ...a5 was directed against Botvinnik's plan. In the late 40's he met the Classical Dutch by playing Ba3 iso Bb2. This seriously hampers Black's plan to play ...e5. The antidote, developed by Simagin, is to play ...a5 meeting Ba3 immediately with Nb4 (either via Na6 or Nc6). On the longer term the idea is to get ready for ...e5. If White succeeds to prevent it (eg by playing Nf3-e1-d3), Black prepares counterplay with ...b5 (hence the moves c6, Na6, Bd7, Rb8).
Here I simply don't agree with Williams, that White can play for an advantage without any risk. It is true, that White's play is easier. Black's play is reminiscent of the hedgehog, his position remains cramped for quite a while. The good news is, that if Black manages to take over the initiative, that White's position often falls apart very quickly. At the moment I am not sure, if White can prove even an edge (with b3 and Bb2, that is).
The next logical step is: why not ...a5 before Qe8 (so on the 7th move) ? In fact the idea is simple: Black sometimes wants to play 8...Ne4 and then after 9.Nxe4 fxe4 10.Ne1/Nd2 the answer d5 is stronger - pawn d4 is under attack. Of course White often does not take the knight. Williams also answers the question, how Black must proceed then. Black should pay some attention to Broomfield Phillips, 4NCL 2002/03. He played all the normal moves and still faced some problems in the endgame.
Finally Black might even consider playing 6...a5, while the pawn is still on d7. The reason is, that variations with 7.b4 or 8.b4 are quite dangerous. I can tell from experience.

The sad news for Inn2 is that 7.b3 offers even a wider choice: except 7...a5, 7...Ne4 and 7...Qe8 Black also may try 7...Nc6. After 7...Qe8 8.Bb2 (8.Nc3 transposes of course) Black again has a wide choice. So instead he has to be ready for even more nonsense ....

But I really cannot help Darthmambo. I would say to investigate all three. The reasons are clear. It is good to have several options open. Then there are some transpositions. Finally there is a lot of crossover of ideas.

I could never quite figure out the idea behind 7... a5. I mean, 7... Ne4 and 7... Qe8 sure look like they do different things, but looking at the position without knowledge of theory, 7... a5 just seems like a random shot. Sadler made a reasonable explanation of 7... a5 a few years ago in one of his NIC book reviews, but i still think its over my head.

So far i've evaded this nonsense with 7. b3, of which i think its quite critical with a prominent expert going down recently: see Belov- Williams Kusadesi 2006!

I am thinking of playing the classical dutch. I just don't know what line to play. Qe8 looks tatical, going over to the king side, etc... a5 is a good all purpose move, then there is ne4 (but I think there was a chess pub post about one line in it not being so great for white). It seems like when you play the classical, you have to be ready to transpose to one of the 3. Well any thoughts would be great.