This whole "pillars" paradigm is worrisome to me. It reminds me too much of "roles" in 4e, where a class is defined by a metagame concept and not how it exists within the game.
IRL, a species can be catetorized as a predator or scavenger or herbivore, for instance. That doesn't change what they are or what they can do. But if you were trying to create an abstract computer model of animal species, you might find it very helpful to have an individual species 'inherit' traits of a category like 'predator.'
You can attempt to break a game down into singular data points--Roleplaying, Exploration, and Combat.
The only worrisome thing I see in the 'Pillars Paradigm' is the tendency to label one of them 'Roleplaying' instead of 'Interaction.' RP is an activity that can take place to the extent the PCs desire it across the Pillars. Interaction is a broad grouping of challenges that are solved using socials skills and other abilities that can influence them. Roleplaying is something the players do, combat, exploration, and interaction are things their characters do.
The reason it fails is because it is an artificial form of game design. It excises the creative process in favor of a logical, analytical construction of ordered game mechanics producing a sterile result. ... I fear, this is what 5e is going to do with the paradigm of pillars: "This is an Exploration option. It should not increase your ability to explore by more than 1.75, which we have calculated by averaging the possible number of scenarios that it might apply and then dividing it by pi."
Game design is an abstract and artificial exercise, and experts can and should use some pretty deep math and theory if they want to do it really well. Games, especially RPGs, are very abstract by their nature. (LARPS, I suppose a bit less so.)
The above questions are rhetorical.
They are also /all/ completely bogus. In any situation in-game, you'll be somewhere, which you take as meaning it's exploration. There will be others you can talk to, if only your party members, which you're trying to imply will make it RP. There will be, if not other creatures you could resort to violence against, at least objects you could break. There will also be things that actually matter to overcoming the challenge, and they'll generally fit neatly into one of the 'Pillars.'
Attempting to rigidly define them and shoehorn classes into one of the three pillars--and trying to balance them around this, no less!--cannot succeed.
D&D already has drawn a pretty sharp line around combat, for instance, and doing so has only served to improve the game. Now, shoehorning classes into one of the three pillars /would/ be a terrible mistake. If the smoke you were blowing above were solid and all three pillars really were always relevant all the time, it would actually be OK to have classes that specialized exclusively in a pillar. Because that's flatly false, and D&D challenges really do tend to be either combat, exploration, or interaction - even if they can move quickly from one to the other, particularly when a non-combat challenge is failed, provoking a fight - and because campaigns can focus heavily on one pillar over the others, or on each pillar to varying and different degrees -
no class can afford to be bad at any of the pillars.