Here's a summary of everything I've been writing about so far, collected in a single condensed post. It's still damn long, and you probably should still read the other posts if you haven't, but here's the "short" of it.
1. Roleplaying is not a specific game-thing, it's a medium
"Let's roleplay" doesn't really mean anything without proper context, it is about as well-defined as "let's play cards" or "let's play sports" and that's ok. It covers a range of activities that can be quite different from each other. We gamers are well prone to identify "roleplaying" with one specific behaviour, ofen equating it with "how we play" (or "I play"). Thus distinctions between "true roleplaying" and "doing it wrong" are drawn, but they are all false, because none of them have immediate primacy. So roleplaying is not something specific we do, but rather a medium for doing various things. Often described as simply "this thing we do".
2. We use roleplaying with certain aims, to fulfill goals and expectations
These impulses are rarely identified and recognized and even less often spoken out. A gaming group usually falls into its own groove, and develops its own system of play, but we often work from many unspoken assumptions and adopt unquestioned norms. Communication is always of prime importance, but these things are often hard to communicate, for various reasons. When an idividual's aims are not satisfied he usually rebels (gaming always, always presuposes freedom) or disrupts the rest of the group by pursuing his aims, or becomes bored with the game. When a group's aims are not satisfied play becomes boring, stale, frustrating, unsatisfying as a whole. Games and techiques are judged "wrong", "broken" or simply "bad". A huge amount of this can be resolved at a purely social level, with a dialogue above and beyond the gaming table: establishing what we are doing.
3. Why do we say "Just roleplay it"
"Rule zero", "just roleplay it", "my gm is great he can make any game fun", "it can work with the right players" and simmilar claims and maxims are basically all saying the same thing: Just make a judgement call. The human ability to decide, draw conclusions and make judgement calls regarding fictional content easily trumps any sort of rules in the sense of fulfilling our aims. It is however these aims and the framework of setting, genre, theme etc. that informs our decisions. This is why even the most freeform games are not "system-less": our decisions on what happens are systematic. Whether we're using a set of published rules, some homebrew concotion or doing freeform, we are developing our own system of how we decide what happens in the game. What happens should serve what we are doing.
4. You can't do anything
Following rules is one way of determining what happens, explicitly or implicitly. Thus, the kind of outcomes (behaviours, consequences, narratives, etc.) that a game favours (suggests, imposes, rewards etc.) either support or break what we're doing. This is how games are about something. Even games that aren't about anything and promise they allow you to "do anything" are fundamentally more or less friendly to "what are we doing". I think it was Ryan Macklin that called GURPS (a supposedly "Generic Universal" RPG) a "Gritty Urban" RPG: by virtue of its rules, GURPS is more or less friendly to what you're trying to do. We drift, tweak and houserule to make it fit better but beyond a certain point it breaks: you end up with something entirely different or need to start from zero or pick up a new game. Rules, as a method of determining what happens and how we behave in the fictional context, are part of our system of acheiving our aims. Just as a musician can't make an album that will please everyone, you can't write rules that will satisfy all aims.
This is also why Jared Sorensen has been jammering on about how "fun" should not be a priority of game design. The question is fun how? for whom? why? Design a game that does something, that is about something, and do it well. Then people who like that will find it fun. This systemic support for genres, themes, settings - but above all: aims - is where design happens. We can't design people or gaming groups, we can design rules for them. All design, much like art, is for someone. Mostly for ourselves, but for someone. For the people who will find it fun, for the people whose aims sync with the game, that's the purpose of a game. When you create a new rule the question is what does it do, how, and what is that saying. How are people going to use it? Even more important than individual rules are the relationships between them, the interactions between parts are where the real meat is. If I gain enoug experience, I level up, if I level up, I get a higher attack bonus, if I have a higher attack bonus I can hit bigger monsters, if I hit bigger monsters I get more experience. That's design, making those choices, so the game says something. Whether you're houseruling, hacking, drifting or designing, you're trying to make the game say what you want it to. When you "just roleplay" you're saying it yourself, with all the pitfalls that might entail.
This is my framework for discussing RPGs, more or less. Now that I've established it perhaps next time I can finally start talking about some actual play stuff.