Different classes use their powers in different ways which often are not applicable to other classes, so a generic pool of martial and magic powers isn't going to work without reducing variety in the game, which becomes rather self-defeating.
I can understand that a number of people have difficulty filtering through various options - they're the audience that the 10 Essentialss products were aimed at. However, if they have an issue with additional options, they basically have to stop purchasing anything other than fluff books, as every additional class book will add complexity, and WotC has little choice but to keep producing books.
I have to say that I'm definitively on KM's side of this question. There is a very definite cost to each extra game element which is added to the game. This cost is payed in a large number of different ways, some of which spring to mind are:
1) Increased development complexity, the new element needs to be assessed against the other existing game elements. As the number of elements it can interact with increases the complexity of this task increases as the square of the number of such elements.
2) Decreased focus, the development team has only so much manpower. There is some finite number of game elements they can adequately support.
3) Increased decision burden, players now have to filter through this long list of (at present) 1000's of elements which may be applicable to a specific choice point. It is all well and good to say "make a good way to filter them" but effectively you have to know they're out there to even look for.
4) Increased planning burden, in 4e you not only need to wade through the choices you are faced with AT THIS MOMENT, but you need to also understand how those choices impact all the other possible choices you're going to want to make with this character in the future. As the number of choices increases this becomes more and more difficult.
5) Added and often redundant page space, books only have so many pages. When class N requires basically recapitulating practically the same powers and feats as classes A, B, and C already had that's just more marginally useful or even wasted book space.
Honestly when I cracked open the PHB1 in 2008 and first went through it this was THE first flag that popped up in my brain was the failure to provide for any feasible way to share material between these extremely large heavy-weight class descriptions. I can't really comment on why the 4e devs felt this design was superior (or even adequate) in the long run, but experience seems to be bearing out that its costs are increasing steadily over time and threatening to collapse the system at this point. It is already close to reaching a point where additional support may be practically impossible and this is likely a large contributor to both the push to use subclasses in Essentials and the general slowdown in and decrease in quality of new support across the whole system.
What I see is that they have arrived at the same conclusions I've outlined above and their solution is to simply fall back to a small number of core classes which can share game elements going forward. They could have done this in other ways, but given the need for backwards compatibility the 'subclass solution' was a fairly reasonable compromise. I think they should stick to their guns. 4e at least may remain a viably supportable system for a few more years and then 5e can do a deeper rethink and restructuring to get all the way there.
I'm not entirely sure what the most ideal structure would be, but I AM sure it involves a vastly reduced list of powers overall. I'm taken back to the solution which worked well for casters in AD&D, particularly 2e, where there were only 2 spell lists. Any new class could draw on those 2 lists and/or extend them in various ways. I think perhaps it would make sense to have power source based lists which are then allocated in subsets to different archetypes to create classes. A fighter and a rogue can draw from the martial list. They may be allowed different subsets or get different riders/class feature interactions, etc to distinguish them, but fundamentally they'll be reusing a large part of the same game elements, as would warlords, rangers, barbarians, etc. There are many possible minor variations and tweaks that could be made to such a scheme, but in the long run it seems almost inevitable in light of our experience with 4e, IMHO.
