Designing your Interface.
For the past few posts, I've been dancing around a core idea that doesn't get much play: the resolution of your interface.
I don't mean 800x600 or whatever. I'm talking about the detail level of the control that your interface gives the user.
On one side is the quick time event: the user doesn't have much control. He either pushes the button and moves on, or fails to push the button and gets moved back. It's a lot like Chutes and Ladders: the sense of control is largely illusion, and any skill test that might be there is simple pass/fail.
On the other side is Quake and buddies. Allowing you to control your position and direction with exquisitely fine control, the game's complexity arises from the natural, physical complexity of the rules combined with the nature, physical detail of the controller.
A game with that high level of physical complexity but a less complex controller is not the same at all. There are many games which have the same basic physical nature as a first person shooter - for example, Crackdown. However, their control system is not as fine: Crackdown's aiming system is automatic instead of the skill test you'll find in Quake. There are other examples.
I spoke of "grooved space", such as what you find in Sly Cooper. Sly Cooper's physical complexity is about as high as Quake, but the gameplay is radically different because the way you interact with that complexity is very different. Sly's acrobatics are largely helped along by Sly himself: you jump for a spire, he lands on the spire, no need to worry about pixel-perfect exactness.
As pointed out by Mory, this can lead to a boring game: it reduces the amount of skill required by a tremendous amount. However, in Sly's case, this reduction does not harm the game. It helps the game, because the game's main skill focus is on timing. All of the jumping around and combat and so forth is made much simpler so that the game can focus on timing.
The game eagerly switches between kinds of timing, running the gamut from Solid-Snake-like sneaking to Sonic-the-Hedgehog-like sprinting and jumping. It even has its Guitar Hero moments. But this breadth of experience wouldn't be possible if the low-level skill challenges were really hard. It would require too much effort for the player to just DO things, let alone do them with proper timing.
There's no "right answer": a game with a very low-grain interaction can be fun: play Rock Band. A game with a very high-grain interaction can be fun: play Quake. And everything in the middle can be fun, too. But they are very different approaches, and their side effects need to be understood by the developers. That's why Sly 2 & 3 weren't so hot.
As a side note, I've been giving examples of games where the space is complex and the controls are simple, but the opposite is also possible and, in fact, almost indistinguishable. Thinking about space and how the player moves through it as two separate elements is not a good idea: it's really one element. Game space-time, you could say.
For example, in Crackdown, movement is very high-grain. You have as much control over your movement as you do in Quake, and you have much better movement capabilities. However, it's not usually a high-skill situation: the world is carefully designed to fit your capabilities, so being a bit rough on the controls works out fine. It's the same result as Sly's "yes, I can land on that spire", but approached from the other side.
Do you see what I mean? What games are you playing, and where do they fall on this scale?