Here's something I said on Vincent's blog, which I think is kinda smart. It's about classifying different kinds of rules. I'm usually pretty skeptical about classification systems, but I think maybe games are better when they use a variety of different "kinds" of rule, which this may help with:
Rules can usually be broken down to an "if/then" statement. That's almost never the best way to communicate them, or to remember them, but for the purposes of this analysis, it's useful.
"If [thing] then [action]"
The "thing" referred to above can:
a) Happen in the real world, or in the fiction
b) Happen frequently, or rarely
The "action" referred to above can:
c) Tell you something happens in the fiction, or tell you to do something in the real world
d) Be very predictable, or very unpredictable
I think that covers what Ben's talking about in this post, as well as Vincent's "Mediating Cues" stuff.
Very often, if the "thing" is very common, and the "action" is very predictable, we tend to internalise the rule, and it's never overtly invoked. That's what Ben means by "continuous" rules. They become part of the landscape of play. For example:
"If [someone is talking about something 'their' character does] then [give that credibility - it happens in the fiction]"
That's a very common rule, which is almost never invoked during functional play, because it happens very frequently, and the action is very predictable. The rule is rarely stated outright, but the rule influences our play anyway.
"If [a character is going aggro], then [roll +hard, ...]"
That's a rule from Apocalypse World. It's almost always invoked when it comes up, because it happens infrequently, and the resulting action is unpredictable (in this case, because you have to roll the dice).