(Note: this is mostly about design, maybe about critique).

So, you read over game book X, and you put it into play.  Within a few sessions, this perfectly ordinary thing has happened - rules from the book have turned into habits in play that you don't need to think about.

Okay?  That's 'the collapse of rules' I'm talking about in the thread title.  I bring it up here because I think that games (and designers) approach this stuff differently, and I think the difference matters.  I will use some examples; they probably aren't perfect examples, but hopefully they're good enough that you can bear with me.  I'll also use some "on the one extreme, and on the other" stuff - this is just to illustrate, not because I think the extremes are important.

Pervasive Structure And Optional Structure

A game that has pervasive structure has bits on-hand that informs so much stuff in play that play can become the habits borne from the rules.  Dogs in the Vineyard has pervasive rules; the structure is everywhere, informs everything.  Spirit of the Century also has pervasive rules; the fate economy gets into anything and everything.  Amber's methods for judgement are pervasive...   but they aren't all rules.  

On the other end, we have optional structure - rules that you can "go to" when you think it would be good to use rules for this thing or that one.  Many, even most, traditional games are built this way; really old games that have totally different systems for resolving different things are especially like this - the core play is not in the rules-habits; the rules are there to be "gone to".

--

Degree of Collapse
The other measurey thing is the degree of collapse - how invisible can the rules get, how much can they vanish into habit?  Looking at the three "pervasive examples"...  Dogs doesn't collapse much at all; the structure always stays visible.  Spirit of the Century collapses quite a bit, but not all the way; you'll still be skipping tokens across the table, even if it feels totally natural.   Amber comes pre-collapsed; many of the 'end result habits' are shown off more clearly in the examples than in heavy structure.

--

Okay, so what?
When you're doing game design, some kinds of systems (like fate points, say), are always pervasive.  If you link those things very strongly to a purely optional piece of rules (like a totally regular combat system), stuff happens.  Let's say you made combat the most commonly mentioned way to spend those points; if you do, you run the risk of making people think that the game is "all about" combat - or of thinking that the pervasive bit (fate points) are actually a part of the option (and ignoring it except when thinking about combat).

Also, be aware of how well your structures "collapse"; this likely will require some playtesting.  Basically social rules with a single mechanical bit (compare numbers, roll a die, whatever), collapse easily.  Rules that involve actual physical handling (miniatures, lots of tokens flying around, dice getting bid back and forth, big hands of cards at all times) collapse less easily, or not at all.  And some kinds of fun (seriously in-character stuff, say) need the rules to be "collapsed" and out of the way - while other kinds of fun (tactical challenge) thrive on "uncollapsed rules".  Don't aim for a kind of fun that your structures sabotage.

I wouldn't be so dichotomous.  I believe I'm struggling with a game system that very much uses 'like fate points' but still remains as optional as the group assumes.  The use of points in this game (well, actually they're d6s) can be on any die roll; only die rolling isn't the point of play.  It's a way to raise tension or resolve conflict between players.

In Scattershot, the way you gain these dice (let's call them experience dice, cuz that's what they are) is by invoking one of the tropes of that specific game (games have tropes, players have tropes, non-players have tropes; in Scattershot, everything has tropes).  The way you use them is by bending die rolls (or game events) in favor of how you want them.  Neither of these is particularly pervasive.  In fact, I'd go so far as saying that many regular games, like D&D, don't spend all their time roll-playing.  I tend to call non-mechanical play Informal Play.

Informal Play is 'informed' by rules and dice and whatnot, but not ruled by them.  People just talk and that's the play.  Sometimes dice come out when people can't think of what they'd want to have happen; sometimes they come out when the want to gamble; sometimes they even come out when someone wants to disagree.  And just as quickly, they can go away again.  Or just stay out; I'd call that Simple Play.

Scattershot can scale up to much more complicated than that, but at every point these experience dice stand ready to man-handle play.  Thus a game that (probably badly) tries to straddle both sides of the Pervasive and Optional fence (if there is one).  I've also tried to create a collapsing system for the rules too.

Like you say, it all takes a certain amount of playtesting.  I'm hope to get a rough draft up for some playtesting next month, if anyone is interested.  (Drop me a line.)

What am I trying to say?  Perhaps Pervasive and Optional are not extremes on a continuum, but overlapping possibilities in playstyle.

Fang Langford

(Who, you're still working on Scattershot Fang? Good to know..)

Anyway, as an example that I'm thinking of right now--I think Top Secret SI would make a better zombie fighting game thant AFMBE--AFMBE is a consistent modern rules set, but it lacks the elegance of the combat that TSSI has for hitting and hit location--which is determined by the same die roll. The lower you roll the better you've succeeded, but also the more likely you are going to be to hit something vital.  AFMBE simply uses "roll randomly for hit location", and that is entirely optional. For a zombie game (if you use the classic "Shoot them in the head") it seems to lose something in conveying how important location is to combat.

The rules choices help enhance play for certain kinds of games. Some rules choices are just flat--simple, not specific, and so they can be used readily for anything. Rarely, however, are they optimal for anything.