During brief discussion with Phil I verbalised the idea of good game design not being the same things as good rpg design. This is obvious when discussing, say, Chess. I argue that it is also true when discussing roleplaying games, given the way I define good game design.
The definitions have my bias clearly articulated; they are there for all to see. If you have different base assumptions or definitions, your conclusions may also be different.
Game design is building a (semi-formal) system where players can make mechanical choices that have mechanical consequences. Good game design makes this process of decision-making interesting: There are few null choices that have no effect and the best choice is often enough very hard or impossible to see, if it even exists and is unique.
Rpg design is building a fiction and a system that describes how the choices the players make affect the fiction. Good rpg design makes the process of play interesting: There are actual choices to be made, they are about something the player cares about, and there are several roughly as lucrative alternative ways of making many choices (in this paragraph several can be arbitrarily large, but not too small).
Good rpg/game design does not imply that the game itself is good, because there are numerous other factories related to that. As such, if one is only interested in how much enjoyment can be derived from a (roleplaying) game, fixating too much on the quality of the (rp)g is not advised. There is correlation: On average, well-designed stuff is more enjoyable.
Do note that the other kinds of design are immensely important (and not part of the above definitions): Designing the game so that it has a suitable social footprint (the time, effort and commitment gaming takes), building the game so that it encourages the creation of certain kinds of fiction, building functional character sheets, elegance and other usability issues, and doubtless other factors. I may someday extend this post to explicitly include some or all of those things. This is not that day.
My thesis is that good game design and good rpg design, as defined above, are not very tightly linked. One can have an rpg that is well-designed game but not very interesting fiction-wise; likewise, a well-designed rpg need not have interesting mechanical elements.
What I am not saying is that the two design issues are orthogonal; they certainly affect each other. I am also not saying that they are independent; the quality of one factor tends to influence the other for the positive, because it is common to link certain fictional and system-level effects together.
Examples in the abstract
Assume a game with very complicated (and intense and fun) combat system. Assume the output of the system is the amount of hit points the participants have at the end of the combat. All other variables that change only affect the single combat encounter and any used resources are recovered with a moment of rest or such. This combat system is (one can assume) good an instance of game design, because it has many (mechanical) choices that are interesting. It is not good rpg design, because none of those juicy choices are persistent; all that remains is the number of hit points one is left with. To be honest, there are other potential choices one can make: Which opponent to kill, how much of one’s abilities to reveal, for example, but they are pretty minor and would work with almost all combat systems.
A game where each (player) character has a number of memories (some of which are utilitarian, some have emotional value, some both) and the character can sacrifice them to demons in order to get wishes or other benefits could be well-designed, rpg-design-wise; if the character sacrifices too much, that character can no longer enjoy from the achieved victories; if too little, something bad will happen. OTOH, sacrificing the utilitarian memories (where was the artifact hidden again?) can have much the same effect as sacrificing nothing: Failure at preventing the bad things. Game-design would only make this interesting if the memories with emotional value gave some sort of benefit; otherwise they are like spell points.
On D&D 4th
From what I have read, 4e is focused on encounters and the designer are doing game design. What about rpg design? No idea. Experience for achieving certain story points could do that, but I am more than slightly doubtful. This does not mean that “there will be no roleplay in D&D 4th”. The system just will probably not do all that much to promote the kind of roleplay I am looking for.
Bonus: Proof by antithesis
Assume that all good rpg design is always good game design. See the two example above. They are non-trivial counter-examples to the antithesis and hence the antithesis is wrong, from which it follows that the thesis is true. QED.