I just read some posts from another blog on Why 4E Died and an addendum to the previous post. More interesting than why the edition had so short a life (which I could care less about; I already have my books and can play 4E in perpetuity) were some of its fundamental flaws. With many of the recent announcements about 5E being less than hopeful, I've been thinking about tweaking 4E on my own. Namely, providing universal house rules that fix math issues, paring down the rules into an optional simplified version of the game, reducing option glut and re-balancing certain elements, and completely revamping the Magic Items system. Some of Kris Hansen's (the author of the above posts) thoughts are relevant.
Powers: I agree that there is a lot of untapped potential here. Lately I've wondered if fewer powers with more flexibility might be a good idea (this has already been done with the Hunter (Essentials subclass) Ranger). Also, options that "build" off of previous powers as you level (like a Great Cleave option that you can use in place of an encounter power), and more at-wills in general. Also, a system to cover "improvised" actions in a way that's more comparable to powers. Perhaps reducing power redundancy and/or streamlining the power list for each class. A huge undertaking, to be sure.
Races: Cultural options for different races would have indeed been awesome. Racial utility powers also came into the game much too late.
Classes: Some interesting thoughts, but I doubt I'd touch much here. One thought I've had lately was making the Druid similar to the Berserker, with Wild Shape functioning as an on/off switch to go from a controller to a striker or defender. I may just homebrew a new subclass for that.
Skills: Flat out disagree with the author here. I like the existence of skills, but I'd re-do the mathematical details while keeping the basic skill list (which I find waaayyy more functional than 3.xE's). In short, I hate the +1/2 level scaling of 4E skills and I'd keep DCs flatter across a character's career with less numerical growth (ideally I'd alter the whole game in such a way, but then I'd have to re-write all of the monsters, and no thanks).
Feats: Obviously I'd bake in Expertise and Improved Defenses for free, but I'm strongly considering scrapping the feat system altogether. For a simplified version certainly, but in general I'd probably have to just reduce the importance of feats. After all, without them multiclassing and hybrids won't work properly. Still, there's a lot that can be done and I don't think it would take much effort.
Combat: Disagree with the author a lot here, as I think 4E's style of combat is an intentional feature. Still, a "simple combat" option would be nice (especially for less important encounters), and I think power/feat/magic item tweaks that I would be making anyways would go a long way toward reducing option paralysis. In short, I can probably get away with reducing average combat length but I wouldn't go so far as to try to remove the grid. If you don't want to play with a grid, you probably don't want to be playing 4E. I see the value in both styles (indeed, it's been refreshing to go gridless in The One Ring).
Items: Only barely touched upon here, but I'd make the most drastic changes here. First, inherent bonuses would be the default, and I'd make my inherent damage bonus (which don't stack with item bonuses) standard. Magic items would, by default, be reminiscent of 1E when they were uncommon and getting one would be special and mysterious. I might consider implementing the Magic Item Reset from Square Fireballs. Too soon to tell. Also, I might flesh out basic implements a little bit; much like a weapon-user can choose between a +2 or +3 proficiency weapon, I might make Superior Implements selectable by default, with an Accurate Implement being the equivalent of a +3 weapon.
So there you have it. All of this game-design talk spurred by 5E has really forced me to think about the state of D&D, and I think that a (somewhat) heavily house-ruled 4E that fits the preferences of my own gaming group (and possibly others out there) would be valuable. 4E provided an excellent framework, and it doesn't look like 5E will build on that. Someone should, and I'd like to at least attempt it.