So I have been reading up on a bunch of different systems, especially rules-lite systems. (both the classics and the new wave versions.) and seeing the various ways gameplay and simulations of the game world are modeled through game rules have been very illuminating.
It's pretty clear that all RPGs fall on a continuum of how many rules the designer thinks are necessary. And over time there is a clear tendency for more rules to be created and added onto a game. No surprises there.
So we can look at the inital rule set for a game (before it has begun to be 'patched' with additional rules) as what the designer originally thought was needed for the game to function. I suspect that a lot of self published games start off as the GM's own set of house rules, that he decides work better then what is already out there and so chooses to publish them as a new game. So I think that those initial rules are a pretty good picture of how they RAN their own game- the rules they decided were necessary based on their own gaming group's dynamic. (there are alot of problems with this though, since I am sure many game designers look at what other games include as 'required' rules when creating their own game. But the gist of my thought is that the areas they chose to expand the rules were the most contentious areas in their own game.)
If you look at RPGs without rules have no constraints, and can devolve into squabbaling along the lines of every game of Cowboys and Indians ever played for more then 10 minutes. But I am sure there are Master GMs who can make it work- they have the player's complete trust and are able to act as master storytellers who keep the game flowing without dice or rules.
But they are a rare and mythological breed. For us mortal GMs, at the very least we need a combat/task resolution system. Something that impartially answers the question of 'Did I succeed?'. Again there are many talented and experienced GMs who can create multi-year campaigns where the players have epic adventures, all with that one simple mechanic agreed upon.
But again, players are sneaky folk, and they might realize that they can redefine their character to succeed more often, annoying the other players. So the character sheet is created as a social contract between the players that they will keep their character consistent and not make unfair changes. Now we have a level playing field and some rules- this is looking less like a story and more like an actual game.
Of course a game can also be won, if you figure out the constraints and find the spots to exploit.
From here the options explode into a multitude of directions, with various rules being added to patch the weak spots, which them themselves later need more patching. Eventually you end up with one of those games with a encyclopedia of rules that aims to account for every eventuality.
My point to all of this, is that rules are a crutch to overcome a GM's fallibility. Now, no GM is infallible (certainly not me,) so this is not a statement of me saying we need more freeform gaming. Its more the idea that whenever contemplating adding a new rule to a game system, one should think of it as adding a crutch to fix something the GM is having a hard time handling by fiat.
Is it easier to create another rule that needs to be remembered, or use your GM skills to simply handle it in game?