In short, I suppose I am saying that I found two problems with 4E. 1) It was grindy; 2) I very rarely took the opposition seriously. The new math helped with
[URL=http://www.enworld.org/forum/usertag.php?do=list&action=hash&hash=1]#1 [/URL] , but -in some cases-made
[URL=http://www.enworld.org/forum/usertag.php?do=list&action=hash&hash=2]#2 [/URL] even worse than it already was.
Well, if that was your experience. Still, the 'new monster math' was mostly making monsters more bad-ass, so I don't see how it couldn't have helped at all with 2, let alone easily made it worse.
As with dissociation and realism, at some point you do have to make a willful leap past the game's abstraction and other concessions to being a game to get to the fiction it's helping you resolve. :shrug:
Looking to the DMG 2's Skill Challenge numbers, the new math in that area of the game made
[URL=http://www.enworld.org/forum/usertag.php?do=list&action=hash&hash=2]#2 [/URL] far worse while also increasing the grind I felt during many skill challenges.
Grind was not something I ever noticed in SCs. Skill rolls don't take long to resolve, and there are only so many required before you inevitably hit n success or 3 failures. Can't really see how that could get grindy... unless it just devolved into arguments about whether given skills were useable or not...?
I disagree that developing the mechanics and fluff hand-in-hand are only good for one story. You have mechanics designed for one story.
And that's that. If you want a different story, you re-design some of those mechanics, add some, maybe delete others. You end up with something that's better than the original for the 2nd story, but probably not as good as something purpose-designed for that story like the original was for the first one.
OTOH, with more 'generic' core mechanics, you can add-on flavor and custom-to-a-setting (or genre) mechanics, without having to substantially mod the core. You get better consistency and probably retain more balance for less re-design work that way. By the same token, when the 'proper' (IMHO) arbiter of story, the DM, goes to design his own campaign, he's not re-designing core mechanics, just tweaking peripheral ones.
However, when designing a "modular" game, I believe it is better to start with mechanics that make sense for the core game and then provide ways to build upon that later.
Sense for the core /game/, yes. But not for just one story or setting or campaign. The more generic you make that 'core game' you're designing for, the easier it'll be to simply add to it to evoke something more specific. The more you build flavor into it, or make the core about simulating /one/ set of flavor, the more difficult you make it to adapt or expand or make 'modular.' Modular games, far from being absolved from having solid core mechanics, need to be very versatile and robust from the beginning. This isn't 1974, and we're not groping in the dark anymore.
I believe that works better than trying to create a modular game where the mechanics don't really correspond to the core story you are trying to tell, and then you try to patch that later with a variety of different methods which may or may not work and may or may not work to different degrees. I've been lead to believe that because I'm familiar with this:
GURPS Dungeon Fantasy
and find that it works exceptionally well. I know plenty have a dislike of the system, but I found that -for me- it is an excellent example of how to have a D&D style game and tropes while retaining modularity.
The 'G' in GURPS stands for 'Generic,' y'know.

GURPS was designed as system first, story to be added later.
You had mentioned fitting mechanics to a story. Maybe you're right, but that should be what modularity means, right?
If done well, /yes/, a robust core system, able to handle a wide range of concepts, easily tweaked and added to when more specificity is needed.
That last sentence brings what I'm trying to say somewhat back together and saves it from my rambling. Mechanics and fluff should -in my opinion- be built with each other in mind.
I think what you're hoping for from 5e - a very modular game that, like GURPS, can handle a lot of very different stories and tones and styles - would not be best achieved by doing that. A game system designed to work with a specific proprietary world in a specific genre with a specific tone and meta-plot, absolutely, should be designed that way, with the mechanics being designed along side to support and evoke all the fluff and flavor - the result may not be good mechanics, but good mechanics aren't the point of such a game, the point is the setting and feel. In a game that tries to cater to many settings and feels (styles), the mechanics need to be good, because the game /isn't/ the setting and feel, but a toolkit to create them.