Apparently, one of the great lessons of script-writing is as stated above: whenever you're writing a scene, you should 'arrive' later than you think and 'leave' earlier than you intended - that is, you should get stuck straight into the action, deal with the action, and then move on as soon as practically possible.
As a GM, I find that's very much against my instincts: I want to lay out the scene in a "ground state" and then let the PCs start to change things with their actions. And, at the tail end, I don't really feel right about moving on; that's largely something for the players to decide once they've done everything they want in the scene.
And yet, it's right enough - it's very easy to let things drag out in an RPG if the players don't know quite how to get started, or if the GM isn't willing to just push things when the players haven't themselves called a halt. And it is definitely the case in films that a scene generally works best if there's no extraneous filler that just bloats up the running time.
Building a Scene
When constructing a scene, therefore, I think there are two aspects to consider:
- What are the "moving parts"?
- What needs to be resolved in the scene?
The "moving parts" are anything that can move the scene forward. The biggest of these are, of course, the PCs, but as these aren't under GM control they usually don't count. Other elements can include the NPCs, a ticking clock, or really anything else.
The question of what needs to be resolved should be fairly obvious, as it's the whole point of having the scene. There may be one such thing, or many. One important thing to bear in mind, though: these are questions to be answered; they should not pre-suppose any particular answers - it's just as valid for the PCs to be captured and thrown in prison as it is for them to escape!
Framing the Scene
When setting out the scene, the GM should quickly lay out the location, but should be particularly sure to call out each of the moving parts of the scene - where are the PCs, where are the key NPCs, and so forth. If there's a ticking clock, this should be mentioned!
As the moving parts are called out, they should also be put into motion. This should probably include the PCs - perhaps they walk into the bar, perhaps they respond to the threat represented by the ticking clock, or something. But they should probably start doing something!
That's the key to "arriving late" - the scene starts with things in motion. It's not the PCs job to push things into motion from the steady state, especially since the players may well not know what they're trying to do, and they may not actually know what moving parts they can push!
Playing Out the Scene
The scene should largely play out as normal. However, the GM should be mindful both of the moving parts and also the issues to be resolved.
Each moving part should generally stay in motion. Depending on interactions with the PCs and with other moving parts they may change direction, but they generally shouldn't come to rest. Or, rather, if they do then they should drop out of the scene (and be seen to drop out of the scene).
For example, in the "Speeder Bike chase" scene in Return of the Jedi, as soon as Luke and Leia race off on their bike, Han ceases to be in motion - he can no longer affect the scene. As such, he drops out and we don't see him again. The action doesn't keep cutting back to Han and Chewie standing around doing nothing!
Likewise, as the various issues are resolved, they should likewise be highlighted as such.
It's worth noting that, of course, the players might well introduce new moving parts or new questions to be resolved into the scene. That's absolutely fine - the players are, after all, co-writers in the game so they can and should help drive the story. In which case, add those elements to the board and carry on - again, these new moving parts should stay in motion or drop out of the scene, and the issues should be resolved as the scene progresses.
Ending a Scene
The scene should end as soon as either of two things occur:
- There are no more moving parts still in motion.
- All the questions have been resolved.
Strictly speaking, the first of these should be unlikely, as the PCs will mostly remain active and in motion. However, the GM should keep watch on player involvement in the scene - as soon as the players start to lose interest, or as soon as they start to go in circles or to seem lost, it's time to move on.
But the second point is usually the more likely end point for the scene. And here, again, it should be a question of calling it as soon as all the big questions have been meaningfully resolved, without necessarily dealing with every micro detail. For example, again from Return of the Jedi, the battle on Endor ends as soon as Chewie reveals himself as driving the AT-ST - there's no need to chase down and eliminate every single enemy Stormtrooper.
As soon as either of these conditions is fulfilled, the GM should move on from this scene to the next one. If there are any loose ends to tie up (such as those last few Stormtroopers), either just ignore them or narrate the results - there's no need for those last few rounds of combat to be run through. Beyond that, unless a player brings up something pressing, move on and start the next scene.
Holding Things Back
One other thing that can be considered is the option of holding back some elements from a scene until such time as new life is needed. This is particularly useful if the players start to get lost but there's no clear way to resolve one or more of the questions. In which case, perhaps a friendly NPC suddenly drops some new advice or information into the pot. Or perhaps some bad guys kick down the door, giving the players a 'kick' to get them out of their cycle.
This is probably best done sparingly, as often a scene is complex enough as it is. However, it can be useful in the situation as described, and especially for the likes of planning scenes where the players may cycle between several perfectly good plans without settling on any single one.
The Scene Management Board
Having put all this together, there's probably something that can be useful for the GM to help with this, that being a scene management board. As the name implies, this should be a board (or notepad, or whatever) on which the GM tracks the elements of the scene. It should have five sections:
- Scene Framing: The first column on the board will be about the scene location - terrain and location features, appropriate adjectives, and so on. Specifically, this will exclude things like traps, as these are generally moving parts, but should include general terrain features such as lava flows.
- Moving Parts, active: As the name implies, the second section gives the moving parts in the scene - the names of the relevant PCs and NPCs, the ticking clock, traps, and the like. These are those moving parts that are currently active in the scene; when a new moving part is introduced it should be added to the board; when one ceases to move it should be removed from the board.
- Moving Parts, reserve: A second, smaller part of this column may be added to give some moving parts that the GM has held in reserve. This is an optional section that may not exist in all scenes. Note that this section should not include parts that were once moving but have now stopped - these should be removed from the board entirely.
- Questions, active: Simply put, these are the questions that the scene is hoping to answer. As each question is answered, it should be moved to:
- Questions, answered: As each question is answered, it should be moved to this section, with the answer noted. As far as is possible, the GM should resist moving a question back to the 'active' section - it's seldom interesting to answer the same thing twice.
Of course, the board may or may not be helpful. Also, it may or may not be appropriate to let the players see the board - on the one hand it makes things nice and obvious (here's the scene, here are the moving parts, here are the questions); on the other hand it's a pretty artificial framework, and just as the power cards in 4e tended to have a chilling effect on player spontenaity, so too is showing the board likely to limit the players' ability to cause chaos in a scene.
On balance, I think I'm leaning towards not showing the board.
