You have to think of movieclips as movies unto themselves that can occupy just a single frame if need be. And they play when you test the movie as opposed to playing/scrubbing the timeline. They are vastly more useful and manipulable than graphic symbols, but if you are stuck in a single timeline animation mindset, then it will be hard to get you over the hump until you start learning how to design with actionscript. And if your intended outputs are along the lines of strict animation, then maybe you don't need to to worry about movieclips and actionscript at all... I know there are plenty of animators who don't know actionscript and do just fine.
Hm, alright - so say I were interested in producing animation mostly. That is, something like most of the cartoons on homestarrunner.com, not very interactive once you get past the menus. Should I be using graphics instead of movie clips in this case? I guess the problem is in coordinating action. It seems that if you want to be able to coordinate the action, you have to use graphics instead of movie clips, but that cuts off all the nice features like filters you can apply to graphics. Is this just the way it is?
One other thing I'd like to clarify - to animate objects separately (that is, instances of symbols), they all have to be on different layers, right? So you can't have layers like "people" "clouds" "cars" if you want multiple people, clouds, and cars to move independently of each other, right?
If you are doing timeline animation, each animated object should have its own layer. If you're doing actionscript animation, separate layers are not necessary.