I think 4e does a great job in teaching the art of DMing (seriously, the two DMGs and the DM's Kit are the best DM books ever). And it gives many tools to facilitate improvisation. These just have to be emphasized, so as to not fall by the wayside.
If by DM books you mean, literally, "DM books" then you may be right.
If by DM books you mean "GM books" then I don't agree at all. I know at least three better GM guides - Maelstrom Storytelling, the Burning Wheel Adventure Burner, and HeroQuest revised edition (some parts of which are cribbed by Robin Laws for his work on the 4e DMG2). And these are not just better GM books in some abstract sense - they have better advice for a 4e GM than the 4e GM guides do!
The main thing that is missing from the 4e GM guides is a discussion of the story (as opposed to tactical) elements of encounter building and resolution. Worlds and Monsters is actually a lot better on this, and it's a pity that such a good GM's book was relegated to "preview" status. In the 4e DMGs and Kit, the only story element that is discussed from the point of view of game play (as opposed to the point of view of the fiction) is languages (in particular, the explanation of why the game has only 10 languages).
Experienced DMs, on the other hand, don't really need to be told that they can change the rules. If they don't like something, they will change it anyway - and probably would even if WotC didn't mention the possibility at all. The one thing that WotC should definitely do for those DMs, just to cut down on arguments, is put a nice, clear statement in the PHB that "the DM can change the rules."
I don't really agree with this either. If we're talking about particular game elements - power mechanics, race mechanics, weapon mechanics etc - then I'm not sure the GM should be given the unilateral right to change them. This seems to me a matter of group consensus (although in a traditional group the voice of the GM will presumably be the loudest).
If we're talking about the basic rules of character building and action resolution, then I'm certainly sceptical of the idea that the GM has the unilateral power to change the rules. This is particularly so in a ruleset like 4e, where the quality of the rules as printed is measured not in terms of their fidelity to a simulationist ideal, but their tendency to produce a particular type of game experience by distributing narrative authority to various game participants in particular ways. As a comparison, I'm thinking about those parts of the Burning Wheel advancement rules which say "You can't substitute one sort of test for another - this is hard and fast" - the rules aren't a simulation that the GM is encouraged to tinker with, but a way of producing a particular solution to the problem of "always use your best skill to try and maximise success", and if you try to solve that problem a different way then, in effect, you've departing from what Burning Wheel was written to be as a game.
That's not to say that it's a crime to tinker with the rules of a game. But the designers should have the courage of their convictions - present the rules that they think will deliver the intended "D&D experience", and explain how and why they will do this. (The Adventure Burner is an excellent example of this, in my view.)
If we're talking about page 42 stuff (which [MENTION=607]Klaus[/MENTION] suggests with his reference to improvisation) then I see that as something different again. This isn't about empowering the GM to change the rules. This is about encouraging the GM and players to use the action resolution rules - including the page 42 elements of them - to their maximum extent.