I have waxed poetic on Skate before (or, if you want to be all marketing-droney, "skate."). Basically, Skate is to Tony Hawk what Tony Hawk is to Skate or Die. Skate is mind-bogglingly fluid and natural, a game of skill rather than of stat improvements.
Since it is a game of skill, that means that the play experiences of various players will be extremely different. For example, I am not as skilled as Darius but am better at halfpipes, and Jeff is the king of lame falls.
However, Skate's "plotline" assumes a linear skill growth throughout the game. This is fine until the player plateaus, at which point they boggle at these requirements. I thought I was doing pretty well, until my cameraman started asking for 90 foot grinds and 4000+ points in a single take. If I kept improving as fast as I had been, I could probably do these. But player growth is never linear, and mine is approaching the asymptote.
Portal is the opposite. If you can't beat Portal, your spatial reasoning is significantly below the gamer average. However, Portal offers a huge number of special levels and challenges for players with more skill. Skate, on the other hand, explicitly says, "hey, if you're not both talented and obsessive-compulsive, you don't get to see the X Games and other cool ending shit, screw you!"
This isn't a problem with the gameplay of Skate. It's a problem with the overall game's design. A game with a weak ending... all too common these days.
So, lesson of the day: if you're building a skill game, that's great! Just don't make it require extraordinary skill to complete the game. Make the high-skill stuff extras, or on-line competitions, or whatever.
Alternately, you can do something a bit more generative, so you can support a lot of different kinds of skills. For example, I would love to build a skate.park.