This one is a look at Ico by L.B. Jeffries using his "Zarathustran Analytics," a set of analytical tools he came up with to critique, or maybe a better term is break down (more akin to analysis), games.
Basically, the piece is an application of a concept to the game to see how well two things happen: 1) how well the game lives up to the ideal laid out by the concept and 2) how well the concept helps us understand why the game is successful.
The concept he's working with is Jonathan Blow's notion of disingenuity, the idea that a game's story wants to suggest one thing while the gameplay does another. You can hear Blow talk about this idea here, and you can see a similar idea (with a different name, the mouthful "ludonarrative dissonance") used in Clint Hocking's piece on Bioshock.
L.B.'s main point is that the gameplay not only mirrors but actually intensifies the game's story. Everything from the controls to the particular aspects of the environment you have to worry about (and how little you have to worry about) help increase the sense of being a child, an abandoned child who doesn't quite understand his world, who is trying to take care of the one thing he loves in the only ways a child can understand. The emotional attachments you develop to your companion are deep, but also protective and limited, like a child's. It also creates a sense of "guilt" in the combat when you realize that you aren't really in danger from the "monsters," and that they, too, are abandoned children like you. Nonetheless, you're pit against them.
There's more, and the piece is worth reading whether you've played Ico or not. (It also makes you want L.B. to hurry up and write about Shadow of the Colossus.)
But the main thing that works here is using Blow's concept to guide the discussion. In essence, what we get is the judicious use of a methodology that already has plenty to go for it. Blow pointed out a real problem in games that, once you hear his lecture, can really kill the experience of a lot of games, or at least create some serious emotional distance. But L.B.'s analysis uses that same notion to show how Ico actually increases your emotional investment, even in ways you may not have been explicitly aware of when playing but which, on reflection, were obviously working on you.
Having a concept/methodology/*idea* to guide a critique like this is particularly useful because it clarifies what the writer is trying to say about the experience. It allows him, for example, not to fall into the "review" trap of having to talk about the graphics from a general "they look good" perspective, and, instead, can focus on what he's decided is the game's purpose: achieving an emotional effect. The game, instead, looks like an intellectual, emotional, and even artistic product because the concept with which he begins doesn't assume that the game is a piece of software first and foremost. Instead, "disingenuity" highlights the relation between emotion and gameplay, and everything gets judged from that perspective.
The real thing to learn from this (apart from a better appreciation of Ico) is that we (writers and players) should be on the lookout for these kind of perspective-orienting ideas, and use them to think about games. That seems like a common place idea, and it's essentially how professional literary criticism works...grad school is about giving you plenty of models of reading and then setting you free to practice them, test them, change them, etc. But since most game writing simply works out of reaction rather than perspectives (or even "theories" with reasonably well-defined perspectives for analyzing games), even the best game reviews seem like we simply got lucky that someone had a good insight.
L.B.'s piece is an exceptional example of how many great insights can be gained from the straightforward application of a concept to a game. If only there was more like it.