I often wrestle with a very simple problem when thinking about running a game, one I boil down to "Why not just use Risus?" For those unfamiliar, Risus is one of the simplest RPGs out there, with rules that easily fit on a page, but with enough potential and extrapolated depth that hundreds of pages have been written about it. It's a good, solid little game, and if you want to go from zero to playing as fast as possible, it's a great tool to include in your toolbox.
For me, it's something I view as an 80% solution. For most games, Risus provides me rules enough to cover 80% of the situations that will come up in play, and the question is how important, interesting and fun the other 20% is. Surprisingly, that 20% is a big deal, since most of the differentiation between games can be found within it. It tends to cover things like magic and quirks of cosmology which are often the most noticeable differences, especially within a genre.
I often judge other RPGs by this yardstick, asking what they are offering me over just playing Risus and faking that last 20%, but that's slightly unfair. Faking that last 205 is easy to say, but the ability to do so is a skill, one that is a function of a lot of time and practice. It would be easy for me to _say_ that it should be easy to fake it, but that would be a load of crap.
Still, over time I've come to realize that's only part of the problem - risus is very easy to set up, so easy that it can sometimes be hard to play. That is, the structure of Risus mechanics don't provide any direction of play (whereas, in contrast, D&D's point towards a fight or pursuit of cool powers/loot) so they depend upon other structures. The problem is, however, that those other structures are often MORE work to build than the rules ones, which undercuts the whole idea of fast setup.
What do I mean by structure here? Some combination of system, situation and setting. Structure provides context, and context is a necessity for meaningful action. All of which is to say structure is a necessary part of a good RPG session, but it's an open question WHERE that structure comes from. It's rare (but possible) that it come from only one source, but even in a mix, one is usually more pronounced that the other.
And this, I think, is the heart of what I wrestle with when I think about the gap between games as written and games as played. All games need some sort of structure, but when it comes from different sources, it makes drastically different demands. Any two Burning Wheel games will have in common that they are Burning Wheel, but any two Amber games will have in common that they are Amber. Obvious when you say it, but the real truth is one revealed in play.
Consider the language people use to discuss games. When someone has a good Burning X game, I can tell because of how they talk about it, not by the setting elements they reference (which will probably be few and far between, outside those immediate to the characters). In contrast, two Amber games played with radically different systems will still be talked about similarly because the priority is the setting. When the streams cross it can be interesting, but that's somewhat tangential to my concern.
So now I'm wondering, how do you talk about that? I know how to present system in such a way as to provide structure, but setting and situation? I know some tricks (families, faces and dungeons) but they're crude stone tools compared to the shiny steel and silver of system design. But maybe I can use them to build some better ones.