Umbran said:
Scientific method, via posts on a message board? Dude, you're sample is going to be self-selected up the wazoo.
Scientists do it all the time.

Besides, self-selecting evidence to support an hypothesis is more the academic method than the scientific.
Umbarn said:
I am not convinced there's going to be a single useful theory - this is a social/psychological issue dealing with human desires and "taste". They dont' have a real "theory of TV shows" that works worth a darn, and that's for a passive activity. You want to make one for an active passtime? Good luck!
I must disagree. What I'm after here is, essentially, a description of what an RPG is. A simple, straightforward description.
For all the gobbledegook people append to theories, at their heart they're really simple things. Take the Copernican Theory for instance. It comes down to, the Earth orbits the Sun. Everything else are only proofs demonstarting that fact. Copernican Theory (as amended and corrected by Tycho Brahe and Johanes Kepler) serves to describe a phenomenon. Namely, why certain planets appear to go backwards during part of their orbits. It also produced, in due time, two other theories, Newtonian Mechanics (the laws of motion) and General Relativity, both of which act as refinements to and expansions on the original.
In the case of RPGs a variant of Occam's Razor comes into play. Namely, The simplest description of a phenomenon - in this case, what an RPG is - is usually the best. We're not dealing with a psycho-social phenomenon, we're dealing with what a thing is. How people see it, how they deal with it is another subject altogether. How people handle RPGs is quite outside the scope of this topic.
People have a tendency to complicate matters. If they took the time to look at what tv shows (to use your example) have in common it wouldn't be that hard to come up with what a tv show is. The same thing with RPGs. People come up with complex formulations in an attempt to be all inclusive, when looking at what all RPGs have in common could produce a simple description.
What if a certain RPG doesn't fit the description? Then the best bet is that it is not an RPG as we've described it, it's a different sort of pastime.
I'm trying to keep it simple here. I hope to show that by providing a simple description of RPGs it becomes easier for players and authors to use and devise RPGs to appeal to a wider audience and allow for a wider range of experiences and play. But such an RPG theory needs to be basic enough to allow inclusion of different types of RPG. But not so broad it becomes useless as a description.
Nowhere near summing it up in a nutshell, but I hope I've made things clearer.