Put a little differently, what Numbed has latched onto here are power sources (active class) and noncombat roles (interactive class), much like "striker" and "defender" apply to combat. Somewhat ironically given the approach discussed to "fuzzy" skills (which I love), the active class is more about method than about results (how do you attack? spells vs. swords vs. sneaking vs., um, praying), whereas the interactive class is results-driven in the same way that party roles are in 4e (what is the end result of you applying your talents?). Somebody's the face man; someone else is the locks and traps guy; someone else is perhaps the sage.
This, by the way, forms the crux of the difficulty in determining where healing should go: the active classes and interactive ones answer different questions: "how do you do damage (by what means and in what fashion)?" vs. "what kinds of effects do you achieve [outside of combat]?" The [bracketed bit] in the latter description seems to be the general intent, but of course one can pick a lock during combat, so I would argue that it is unnecessary. If you remove that, then it's pretty clear that healing should fall in the interactive classes, even if it is a function chiefly expressed during combat.
The "leader" role in 4e shows that as well: there are leaders for different power sources, each of whom has a healing function.
I don't mean to say that active classes are perfect analogs for power sources; clearly they are not. But they serve a similar function in the game: describing the flavor of where your attacks come from. They overlap role slightly also, in describing the kinds of attacks that each class makes.
But I maintain that interactive classes are the same thing as "noncombat roles": the different things that a D&D party might need done, distilled into certain archetypes. 4e was sorely lacking in this kind of thing, so an attention to it in 5e will be most welcome, and I hope the devs are reading this thread.
I can see this idea being quite useful actually. It could even give you a reason to have an intelligient fighter, or a priest with an int score. I really like this idea. Not sure about divorcing race from ability score bonuses though, I think divorcing them would be too much for a lot of people, it would be killing one sacred cow too many. Personally I think it might be a good idea.
Building on what Jeremy is saying above, plus a discussion in another thread, I was thinking about the interaction of roles. At first, it would seem possible to make combat roles a socket. So that I'm not being TOLD that I must be a striker. It works for the striker well, just add damage to singular attacks. It could work for the controller and leader. But being a defender is something you DO. It's something you prevent others from doing. Looked at that way, it seems as if the defender role is more of burden than a choice. It is to 4E what being the healer was to 2E.
Wow, I'm not surprised you caught the attention of Monte himself!! You managed to actually come up with a system where I just theorized and rambled... I feel the need to state some things (all actually) that I really enjoy about this model, and others that I would mix with my own ideas.
The points I really like and I hope D&D Next gets right away from your model:
Multiple, separate, and non-obtrusive "paths" that concur in building a character in which these elements could matter all the same: race, background, combat abilities, non-combat abilities, results-oriented skills (and super skills that "just succeed"), theme...
Feats that can give advantage to any one or two of particular areas of expertise of a character, helping to give more focus to something rather than something else. These could make two characters that took the very same "classes" look and feel very different or at least unique.
The "fuzzy abilities/skills" concept, the fact that two characters of different classes (or similar classes, or even equal classes) could have very different methods to achieve what they achieve.
Things I'd add, mix, shuffle, organize differently:
First of all I would ensure that even with the "fuzzy concept" in place, the Interactive methods of a Ranger would be mechanically different from those of a Fighter and so on. Players could then come up with their versions of this, but at least they would have the clear impression that their "various classes" interact with each other to create something unique. The example I made on my thread in which you posted: while investigating a crime scene, a Fighter that specializes in that niche in its Interactive "class" would probably have to re-enact the fight in its head to understand what happened, where maybe a Ranger could understand things from little details around the floor, scents, dirt and so on. True, this could be made with the "fuzzy" concept, but as you can see, one of them used his Active abilities (even if in its mind only), the other more his Interactive ones. This way they'd be radically different. I'm not sure the concept is viable, it's just something cool!
The reason for which I often wrote "class" or "classes" under quotes is that I would try to stay more on the traditional path, and call only the Active "part" of the character its Class. The interactive part could be called Expertise or something like that. After all Theme is not "thematic class". Plus, just as the theme would "feel and look" different from the Class, i would make the "Expertise" differ too, which brings me to...
I would probably integrate the Active and Interactive even a little bit more than what I said in the first point. I'd like to see (at least in the basic implementation of this) Classes limiting their choices of Expertise a bit. This would also reintroduce the concept of Multiclassing, which I feel necessary in D&D. That's to say, if a Wizard wants to become sneaky, even with its own methods and all, it might make sense to "take a level in Rogue" (this could also be made through feats), because that would open more areas of expertise to him. There are multiple ways of achieving this same effect, it's all a matter of finding the right words to use and the right mechanics, so that it all still "feels D&D", which is a central point of D&D Next.
A point that came up to me back in my thread, and that I see very tied to your idea: improvisation is good, role-playing is awesome, but the two don't always have to go together, and when roll-playing comes up as the mechanic to solve things, improvisation and role-playing become just ways to justify a roll. So in this new perspective I'd say that to solve problems, characters would have to "work their way", not roll it. Rolls are good for combat, to represent chaos and everything, but in the "areas of expertise", objectives should be achieved through a series of procedures. The DM could set a complexity level. Maybe some mystery requires such a deep analysis to be solved (both rom the player and the character) that it may require the characters levelling up (acquiring XP, something that in this context would be achieved by investigating multiple elements, maybe even multiple times), and when levelling up, choosing the needed skills, abilities, feats and so on required to make their characters good enough to solve the problem. These changes don't have to be permanent: retraining is your friend. But it makes the players perceive that they worked and built towards a solution. Less complex problems could just require less time and effort. A character spending a lot of time investigating, without necessarily levelling up in doping so, might receive temporary boons that building up together would empower him (them) enough to be able to solve the problem. This could be done with "quasi-items" such as 4e's Alternative Rewards. But there are countless possibilities. The point is: rolls are good for combat, or maybe use of skills in difficult contexts. But working on the solution fio a problem should require working, and building. Self-improvement and time.
Oh and of course, from the DM's perspective: ways of differentiating the various non-combat situations and "adventure phases" so that they feel mechanically or "pseudo-mechanically" different... In Exploration, positioning counts a lot, perhaps even more than in true Combat, but in a more abstract way, and some abilities could grant benefits that differ based on the context they're used in. I wouldn't necessarily formalize these elements, a narrative way of differentiating them in the ability description could be the best way to make the game feel "less clunky" than 4e. All in all, it would be more on the DM's side. But an example of different mechanics I had made in my post is the "stealth encounter", where a well-hidden character could use his implement (Rogue: blackjack / Fighter: sword pommel / Wizard: stunning spell) to disable a guard/enemy in one single blow, silently enough, overcoming the mechanic of HPs, which is something that doesn't matter in stealth encounters the same way it matters in combat encounters.
Thanks a lot for your contribution and thought-provoking ideas!
But being a defender is something you DO. It's something you prevent others from doing. Looked at that way, it seems as if the defender role is more of burden than a choice. It is to 4E what being the healer was to 2E.
hmmmm, well its certainly a popular choice at my house, fighters that lockem down, shielding swordmages kiting around the battle field in particular.
Ofcouse I love LazyLords too... so maybe I am an oddball.
I don't have much more to add except some terminology.
If you wanted to fish around in the old books, using the term "proficiency" for interaction class might add the right flavor.
And you could always marry the concept of "alignment" to "theme."
...Waitaminute. I've got that backwards.
Logically, if your interaction class determines the methods and solutions you have at your disposal outside of combat, your fuzzy skills, then the classic concept of alignment (which I am loathe to bring up but which many people are willing to fight for) would work better there.
It would give people a reason to chose different, more unusual alignments because they would offer different ways of solving problems - as well as a valuable roleplaying guide. The way an Evil character approaches an interrogation of a prisoner as opposed to a neutral (unaligned, whatever) one as opposed to a lawful good one was used as an example way back in the Red Book. Why not codify that with some useful benefits?
Addressing a couple things mentioned above. First, I realize the names I gave the classes suck but I hadn't thought of anything better. Race + Class + Expertise + Theme seems to fit the bill. Also, how many classes (the combat class) should be in the base depends on the type of mechanic used for combat. If it's solidly 4E with powers, then probably more classes to cover role + power source. If it's 3E, fewer classes with the differentiation coming from particular choices in how you do things. I can only really see it as mostly 4E in combat mechanics, even though I don't play 4E. Perhaps if we apply a little "fuzziness" in the description of powers, we can get away with a bit fewer classes. As for limiting choice, that should be mentioned in the book as one of the options a campaign can take. Limit the classes dwarves can take, then limit the expertises fighter can take. It doesn't break the system but I see it solidly as just an option.
One of the problems I see in this myself is Vancian casting. Originally I thought splitting combat and non-combat could solve the problems it has, but now I'm not so sure. I LIKE the idea that Knock is just an Open Locks skill that always succeeds vs. a DC 20 lock at first level. So, if spells aren't used in combat and they shouldn't duplicate skills, what do spells do?
I'm thinking about doing away with a guy who does healing as a job. Healing is a great mechanic. You could design the game is such a way that healing isn't required, but it's not what I want. I'm half of the opinion that if the fighter needs to be healed by the cleric, then it's the fighter's action that it comes from. Even though the classes that heal in both 3E and 4E were much more playable and fun than the before 3E, a group still absolutely requires a healer. From a mechanics perspective, why?
I have a few things to say about non-combat stuff, but it applies generally to any edition so I'll start a thread for that later.
Sounds rather mechanics-heavy, but I may agree with the sentiment. As I understand those paragraphs, we have selectable abilities, no combinations of which are ideal (avoiding how 4e class-build or race choices made the player want to pick certain class powers over others all the time, and no 3e "red tape" Feats/abilities, period).
For division of those abilities, I'd go for concise, Basic-edition simplicity. Selectable ability lists: Race, Class, and Neutral (anyone can take these), or just Class and Universal (the latter includes Race). You'd pick, say, one selection per level, 2-3 at first level instead. Abilities would be the best of each incarnation of the game, meaning these aren't all just attack powers, as adventuring also means exploring, resource-management, traveling, and other NPC interaction.
Multiclassing would be just taking class abilities from another class's list, a limit being: the most class-list abilities they have must be from their class.
I'm skeptical of skills beyond the Thief/Rogue; they seem to make players fall to rolls instead of making judgments themselves. To answer your Knock spell question, here Knock and Lock-picking could be the same ability; Wizards select the Thief ability, and possibly roll to succeed.
I didn't divide combat and non-combat because (1) campaigns & play-styles differ, (2) some abilities are useful in and out of combat, and (3) players should consider the character during character creation, not the tactical feasibility of a stat block.