John Harper made a point on twitter the other day that speaks directly to something that bothers me in a lot of game designs, especially ones I've had a hand in. The problems is that a lot of games pick up inertia with quick rules and a strong premise, then grind to a halt when it comes time to pick stunts, powers or whatever other specific fiddly bits provide the exceptions the the baseline rules. Things grind to a halt as players flip through the book, reading and reviewing their options before making decisions that they're really worried will be the wrong ones.
This is a problematic way to start a game, and a number of strategies have been established to address it, such as setting up quick-picks and packages, or simply preparing characters in advance of the game. These can work, but they're ultimately duct tape and a band aid sort of solutions.
Video games handle this much better, especially more modern video games, since they are designed with an assumption that the manual will barely be glanced at. Players learn how to play by doing it. There are numerous strategies that support this, including familiar control schemes, but the most basic is to start the player with a smaller set of capabilities and options than they will eventually have in play. Historically, this was the domain of "tutorial levels", segments of play that were outside of the regular scope of play, where you'd be walked through the various details of rules and interface.
More modern games have made that tutorial a part of play design. The initial situation of play is usually constrained in some way: you might have only one weapon, one spell, or control only one type of unit. You will play a level under that constraint, and then the next level (or after soem other benchmark) you will expand your capabilities. You'll pick up another weapon, learn a new spell, add more units and so on. If situations require special rules, you'll get the chance to discover that as it comes up (rather than going back to the book). The net result is that you learn to play the game by playing the game, which is pretty slick.
Obviously, different games handle this to differing degrees. For many first person shooters, the ramp up is very small, while some real time strategy games use the entirety of their single player campaign as a ramp up. From an RPG perspective, the most interesting is probably MMOs.
World of Warcraft, for example, starts a character off in a fairly limited environment (a "newbie zone") with clear direction (The guy standing in from to you has your first quest) and well-concealed safety bumpers (there are lots of enemies, but they're the kind who won't attack you unless you attack them first, so they look more dangerous than they are). Beyond that characters start with the ability to make a basic attack (swing a sword, knife, staff or whatever) and perform one special ability (cast a fire bolt, make a power attack, something like that).
For the first 10-20 levels, new abilities come rapidly, but not so rapidly that you don't have time to try them out and get the hang of them in play. The speed of advancement levels off at higher levels (especially in terms of new abilities gained), but those early levels give you a chance to get a grasp on the class. But the thing is, while you're getting that grasp, you're still doing the same sorts of things that you'll be doing later on - getting quests and killing stuff. The fact that you are learning does not sacrifice the play experience.
The fact that video games do this well is not, I think, an indication that this is something that ONLY video games can do well. It would be very easy to conceive of a game like, say, 4e being tweaked into a model like this, so players enter play with perhaps a single at will ability, but gain other abilities quickly, perhaps between sessions when the time required to make choices does not detract from play (MMOs address this by offering very few choices - you get X power at Y level, and that's that. There will be some elements of choice, like WoW's talent trees, but even those will be fairly constrained).
The main thing this requires is a bit of rethinking of how we handle advancement, particularly that we might want to think about shifting some of the things we think about as normally part of character creation to advancement. Coming back to those stunt/power choices that really bog things down, there might be some real benefit in giving fewer of them at the outset, but make the first ones easier to gain.
There are adjustments that would have to be made. One thing you'd want are strong defaults that reinforce character concepts. For example, if you wanted to do this for Leverage, the GM might just pick on talent for each role and just give it to the player at the end of chargen, then let them add another talent at the end of the next two or three sessions. This constrains things slightly (so that, for example, all Hitters are ass kickers and all Hackers have "DO you have that thing I gave you?") but the trade off of quickly entering play really seems to more than make up for it to my mind.
Anyway, something on my mind.