|The creeping dread you experience while looking at|
this picture is analogous to the dread you should feel
at reading this essay/extended musing as it has no
gameable content. Proceed with caution.
I think that system does matter, to a point. One of the initial premises of Edwards's essay is phrased like this: "I'm suggesting a system is better insofar as, among other things, it doesn't waste Herbie's [our hypothetical GM] time." That's hard to disagree with. I love it when the writer of a game, game supplement, or adventure doesn't waste my time. (And don't fool yourself: time-wasters are as abundant in the small-press and DIY publications as much as they are in the big dogs of the industry.) And yet, while not wasting anyone's time is incredibly valuable and generous, it doesn't explain why so many games that do waste your time by being badly designed can be both popular and startlingly fun. In Edwards's example, Herbie is just a great GM who could make anything fun, but that doesn't really seem to fully answer the question of how one talented GM can make a game system that is working against him fun for an entire group of people.
The idea that "system matters" becomes even muddier with the introduction of the thesis that there are essentially three ways to approach an rpg, and there is little to no overlap between these differing styles: "Three player aims or outlooks have been suggested [gamist, narrativist, simulationist], in that a given player approaches a role-playing situation pretty much from one of them, with some, but not much, crossover possible." This is treated as a given, but without good reason. I have two objections here. First, it's silly to posit that there are only three primary ways through which to approach gaming. The dead giveaway is the notion that there are three of them; three is the go-to number for shaggy assertions, wherever you may find them, because it looks like a number of possibilities have been suggested without burdening the reader with a more comprehensive consideration of the actual number of possibilities. Three is too neat, too tidy, and altogether suspect; never trust three.
Second, the notion that there isn't much crossover between those categories does not map to my experience. I could not tell you which of those categories I favor more because I honestly think they may change from game to game or session to session, if not from scene to scene or moment to moment. And I don't think that I'm unique in that at all; in my experience, other players also shift their approach during play or approach different games in different ways. Far from there there being little crossover, I'm left wondering if those definitions have any utility when leveraged as they are in Edwards's essay. Even if we do endorse the idea that a system should be engineered to fit one of those styles of play, how a well-designed game is meant to only attract the players whose approach fits its authorial intention is left unanswered. Okay, so you've designed the best narrativist game ever made--how does it manage to warn off gamists and simulationists and how do we explain what has happened if they do have fun playing a game that isn't supposed to "matter" given their approach?
The answer isn't in the GM or the game system; the answer is in the social contract at the table. While I would certainly prefer that all games be more thoughtfully designed to create the kind of experience they're meant to evoke, I'll always prefer playing with a group of people who have embraced the social contract that it is a good thing to sit down with a group of people and create a fun experience for each other while playing a game.
There are games that I'm willing to accept as being poorly-designed. RIFTS and Vampire: The Masquerade are two games where the actual systems involved work at odds with the themes, imagery, atmosphere, and aesthetics belonging to those games. Yet, both of those games are much-beloved and it is impossible to deny that they've given a significant number of gamers a lot of pleasure over the years. While the game systems might not help create that fun, they also clearly not capable of stifling that fun either. Sure, both of those games could do better jobs at enabling fun at the table, but the prevalence and prominence of positive experiences people have with those games indicates to me that system always already takes a backseat to the social contract wherein a group of people sit down together and agree "These are the ideas we're interested in, let's have some fun."
In literary studies we've had Roland Barthes's thoughts about the disconnect between authorial intent and textual interpretation as an idea of merit for a long time; perhaps it is time to think about the disconnect between game design and games played under interpretation by gamers whose social contract dictates what they want from the game, how it should be played, which rules matter, etc. If the author of a text is "dead" as soon as it is in the hands of a reader, is not the design of a game "dead" when it becomes the imaginative property of the people playing it?
It's odd to me that the social contract is such a powerful director of our experiences in the hobby, but it's seldom talked about. It helps explain a lot. The social contract covers everything from player expectations to what lines people would rather not cross in a game, as well as how we might best contribute to a shared, mutual, and positive experience instead of diminishing it. Not only does it explain why we can get a great deal of enjoyment out of playing badly-designed games, it also bears thinking about in the context of whose job is it to create an entertaining game (see here and here for a spirited exchange on the subject) and also when thinking about why some campaigns fail to get off the ground or falter along the way. "System" can't address much of that, but the largely un-examined social contract does.