What follows is three broad ways of preparing for play. They are basically refined and slightly more narrow versions of a post I made before this blog at Theory decides. The versions written here have slightly different naming schema and extensively use Montola’s theory.
Disclaimer: I find scripted play generally distasteful, pointless or alien. That might influence something.
When game master and the group builds a setting and the players characters (with varying amounts of input from players and GM on the different aspects) and then the characters are placed in the setting, do stuff, and the setting responds, game is sandbox play. A setting generated by improvisation in play based on “what would really be there” can also be sandbox play.
In terms of the model discussed, a sandbox has accidental attractors in that a given group of characters might or might not care about them. Maybe the want to slay the slumbering dragon, maybe awaken it, maybe take its stuff, maybe they ignore it completely. There are things happening in the setting, but they go on independently of the player characters, who are free to go and do as they will.
Sandbox play is hit-and-miss: If the characters don’t have agendas of their own or happen to bump into something that engages the players, the gaming will be dull. There usually is a slow start where players get used to their characters and the setting and have little time to start doing something interesting. On the other hand, given characters with strong principles and goals, sandbox play can create wonderful organic stories and experiences. If characters are of the sort who always get offended by something or always are scheming to the over the world, at least something will happen.
There is a strong starting cost, or need to be good at impro, to run a sandbox game well.
There are strong attractors the player characters are expected to follow. The expection may be tacit (that is what roleplaying is) or explicit. It may be part of the game rules (Rune). There usually is a setting where things that don’t directly touch the player characters happens, but they are on the background. Often there is a particular story that is being told. It may have been designed by all participants (the crazy Swedes are up to no good with that kind of stuff, I’m certain) or by the GM in solitude. There certainly are other methods.
Strong attractors are the key here. If all the player characters are united in purpose (save the world), share similar values (alignment and interpretation of it) and have well-defined solution to most problems (fighting), the game should go along just fine. GM knows what kind of hooks and rewards to use, players know what they are supposed to do. GM can plan excellent events while the players have fun dealing with those.
The great strength of scripted play is that preparation is both useful and efficient. What is prepared is often also used. If not, it can be recycled to some later situation. The great flaw is the tendency to stick with what one has prepared. Some games make this near mandatory. The myth of impro being somehow difficult (more difficult than using prepared material, at least) is a result of relying on preparation. Railroading happens when GM creates more and more attractors that actually lead to the same place when players diverge.
Scripted game (as in a series of sessions) is built so that attractors draw the PCs together. Avoiding bifurcation points is important. An alternative is to place them so that one has time to prepare, whichever attractor is followed after the brief chaos.
Players create characters. Game master builds or tweaks everything else so that characters are engaged, but the direction they move to is unknown. Essentially, volatile play means that GM constructs bifurcation points the players will bump into. The Forge people call a specific sort of bifurcation point a bang. More generally, there are two sorts (not dichotomous) of volatile situations that can be prepared: Those which rely on player making a decision and those which rely on dice making a decision. Generally speaking, the first are more enjoyable, at least in my opinion. Combinations, such as the player deciding which dice get to make the decision, are possibly. See for example many combat systems.
There is room for using attractors, too. They should be used to keep the player characters interacting and the game as a whole coherent. Otherwise all the characters might end up doing their own thing separate from the other PCs, which is generally not as fun as players interacting. It is also more work for the game master.
Good rules are things that don’t require much preparation, or at least much preparation for specific occasions. Improvisation is practically necessary technique, so rules that make it possible or easier are always nice. Rules which make resolution unexpected but not overtly random are another good tool: Stuff like action points that give a significant bonus to rolls, for example, allow success at unexpected situations that the player finds important.
Problems include the aforementioned bubble play, where PCs don’t significantly interact, and inconsistencies. When much detail is generated on fly to drive PCs towards a given bifurcation point, there is significant risk of an inconsistency or three appearing. Usually they don’t matter because they are not noticed. Sometimes things do go messy.
And the lesson is…
Play the way you do, but know that there are alternatives which can look totally alien. Experimentation is a good thing; some techniques transfer well between gaming styles.