Clearly, Mirror's Edge isn't my favorite game this year. But that's not to say it was all bad. As Brinstar has detailed, the main character, Faith, embodies a number of positive traits. Sadly, I think she also highlights the real problem of designing "franchise" characters: one isn't enough.
Here are the good points: Faith is an Asian-American woman, but her race and gender aren't mentioned in or relevant to the story at all. She dresses sensibly, given that she's in a pretty athletic career, and her body shape also reflects that without being overtly sexualized. She's stubborn, but not snarky (there's nothing worse than a "clever" character written by not-so-clever writers). Perhaps, you might be tempted to say, she's still a little rough around the edges.
The problem is that there might be one decent character there, but there's really no-one else for her to interact with, so those rough edges are never really sharpened. Everyone else in the game is dishwater-dull, from the other messengers to the cops to Faith's sister. The same goes for the dystopian setting--it's vague to the point of nonexistence. That's bad for the players, but it's worse for the protagonist: without anything to push back against, Faith doesn't have much ground to define herself. She has no points on which to take a stand, except for the most basic (I think most of us are anti-framing and anti-betrayal).
There are two trends in game writing at the moment: strongly-defined protagonists, a la Jade from BG&E, or mute stand-ins like Samus Aran. In the latter case, the surrounding world has to be made as interesting as possible. It's not a coincidence that the Metroid Prime titles tag everything in sight with scannable text, or that Half-Life 2 devotes so much work to giving the Combine little bits of "business," like their constant radio chatter and introductory set pieces. On the other hand, if you're going to make the protagonist an actual character, you can probably get away with a less defined world (BG&E's setting does what it's supposed to do, and not much else), but you'd better have someone for that main character to talk to, and their actions had better be strongly tied to concrete, interesting motivations. Jade, for example, is constantly interacting with her companions, and she clearly has strong opinions about each of them. Faith has neither--both her world and her friends are generic--and as a result, she herself is uninteresting.
It's a shame, because as I played Mirror's Edge I was reminded strongly of William Gibson's Virtual Light. Like the game, Gibson's book concerns a city messenger and outsider who gets tangled up in a class struggle. But Chevette Washington (Gibson's protagonist) is surrounded by interesting people: Rydell the reluctant rentacop, Sammy Sal and Bunny the bike messengers, and the eccentrics living on the ruined Golden Gate bridge. Chevette is not only defined for the reader by her interactions with these characters, she kicks off the plot herself when she impulsively steals the titular virtual light glasses from a sleazeball partygoer--at every step of the way, we're learning something about her. Faith never even displays that much initiative: her story really begins when her sister is framed, and she spends the rest of the game reacting to events.
Storytelling in games is like the weather: everyone complains about it, but nobody does much of anything about it. Progress has been slow, but (as opposed to other media) it's often shaped by the technology and culture surrounding it. One of the advantages I see from widespread console multiplayer is that it may build support for ensemble casts, as opposed to mascot characters. Gears of War, for example, is nobody's idea of a well-rounded drama, but its characters are inarguably much more lively than Faith is. Mirror's Edge gets caught on the wrong side of this trend for a variety of reasons: the first-person perspective, emphasis on time trials, and a primary mechanic of player-vs.-environment. I'm not sure that better characters would have saved the game entirely--it's got plenty of its own issues, as I've noted--but they probably would have made its failure a lot less aggravating.
Imagine that someone invents a machine that makes omelettes: brilliantly-colored, spicy omelettes made with breathtaking speed. Taken by its combination of verve and simplicity, you order the machine. But when it arrives, to your dismay, you discover that the omelette-making process is actually fraught with danger--80% of the time, due to a misstep in the instructions, it sets your kitchen on fire. Also, for some reason, the manufacturer has added a mode for making breakfast sausage instead. The machine is a very poor sausage-maker, but it keeps getting stuck in sausage-making mode, and until the sausage is successfully cooked you can't get back to the omelettes (and the kitchen fires, which are starting to lower your enthusiasm somewhat for the whole idea of breakfast).
Mirror's Edge is this omelette-maker. It's filled with absolutely gorgeous visual design, presenting parkour from the first-person perspective. Except that it doesn't work, about half the time. The controls are overly touchy, especially strafing, and the context-sensitive options aren't nearly sensitive enough. Worse, the part of the game that's really fun--the running, in between falling--is interrupted regularly with fight scenes. Often, you can't run from the fights, because the soldiers are very good shots and the escape routes are (intentionally) via slow and exposed pipe-climbing. It's like someone on the design team said "We've really got something here, with the running part of the game. Let's make sure to take it away from the player on a regular basis."
There's always a lot of comparison to Prince of Persia whenever a game tries free running and acrobatics, and with good reason, since it did it best. But people often take the wrong lessons from this, citing the "rewind" function that largely canceled out dying-as-punishment. That wasn't the genius of the gameplay, however: what made it really good was in fact the inaccuracy of the controls, the way that the Prince would do the right thing as long as you hit a button with something close to the right timing. PoP realized that the fun wasn't in being a precise platformer, but in the simple thrill of directing a complicated flow of leaps, grabs, and wall-runs around the game's carefully-crafted spaces.
It's strange, actually, that a title with such aggressive visual editing as Mirror's Edge--it's practically monochromatic--would have such weak editing on the interaction side. Eliminate combat from the mix, and you've cut the game down to basically two buttons, up and down. Get rid of strafing while you're at it, since all it does is let me swerve off catwalks by accident, and make it work more like the first Metroid Prime games (which also took a third-person gameplay conceit and moved it into first-person). Those changes would work the level designers a bit harder, but the end result is leaner, more focused gameplay.
What it all comes down to, really, is that when asked to make a choice between realism and fun, Mirror's Edge chooses the former. No doubt, in real life, Faith would be riddled with bullets almost instantly, and so in the game, she is. As a result, the player is discouraged from approaching situations with speed and daring, because it's a process of trial-and-error fatalities made worse by Faith's clumsiness. By contrast, it would be highly unrealistic for players to be able to sprint through a gauntlet of enemy fire, bullets whizzing by but rarely breaking the flow of action--unrealistic, but much more rewarding. As it is, the game just feels unfair: it gives you the tools to do one thing fairly well, and then punishes you for trying to use them.