I'm speaking mostly from my own experience of running games. I've run a number of games from first level (or whatever the starting stats are) up to the threshold of high level - 9th level in 2e, 13th level in 3e, 13th level in 4e (not a good example... 21st is the true "high-level" for 4e), and 5 dots in a primary Arcanum in Mage: the Awakening. My common pattern is that low-level sessions are single-session or two-session adventures (for the obvious reason that the PCs don't have the reserves of power to do more). Low-mid adventures start stretching out a bit more in scope, with mini-arcs of 3-6 sessions. This continues into "mid-level" adventures - yeah, I know the terminology is imprecise. All throughout these adventures, I've been seeding other things, usually far more than I could reasonably run; this means the PCs have plenty of choices of things to pursue, or not. Finally I run a big storyline that takes a whole heap of sessions; in D&D that single goal might take three or more levels to complete.
When it's done - and I've been lucky that games have lasted long enough to finish their "big" stories - I have a bit of a problem. The players have solved the major problem that they invested time and energy into resolving. The central questions of the campaign are answered, and I have built all that I readily can on the foundation that I've laid. That is, my problem is how to cross the bridge into high-level play.
What I need to do, but never seem to consider when I'm on the spot in a campaign, is flash the timeline forward six months or a year, so that the characters rest on their laurels a bit and enjoy the fruits of their victory, rather than trying to have them move to the next thing in-character. This makes sense to the English majors out there who think of Freytag's pyramid. The adventure that ended mid-level play was a climax. I need to embrace the denouement and let the tension ease off for awhile. This doesn't have to last long - even one whole session might be stretching it. Once the tension has tapered appropriately, I should start in with new exposition, either picking up on earlier threads or introducing new ones. Instead, I try to jump from the climax to a new rising action.
A campaign's structure plays into this dynamic as well. For these purposes, I think there are four structure choices (if you can think of more, tell me):
- Serial Drama: This structure has fallen a bit out of favor in the online communities I frequent, where it's seen as an invitation to railroading. It can be that, if the GM isn't careful, but there's still room for intricately-plotted games. In some of the best games, players gradually solve the GM's intricate puzzles and hurl themselves against challenges that push both players and characters to their limits. (For more about the virtues of this style in contrast to pure sandbox play, read here.)
- Character Drama: This kind of game is driven by the drama of the primary characters rather than external conflict. Hillfolk and Smallville are notable cases of this kind. The GM may be little more than an impartial adjudicator of conflicts. There usually are still NPCs (in the tabletop implementation of this structure); their main point is to apply new pressures to the relationship dynamics between the main characters.
- Procedural Drama: Depending on how you look at it, a lot of dungeon crawls could just about slide into procedural structure, but GUMSHOE-family games like Ashen Stars and Mutant City Blues are great examples of this. (If you're noticing how much Pelgrane shows up in these categories of structure, then you've got a pretty good understanding of what makes Pelgrane tick: experimentation with structure is their bag.)
- There are a lot of good and useful advice about procedurals to be gained from just watching TV shows with an eye toward structure. You're in luck, though: Leverage exists. Also, the show is fucking phenomenal and John Rogers can teach you everything you need to know about building a reversal into every procedural story.
- Also, Shadowrun. SR, in its many incarnations, is one of the best-known and most popular examples of a game built from the ground up to embrace this structure.
- Sandbox: The OSR has waxed eloquent about sandbox play, and other bloggers have excessively conflated sandbox with the West Marches style or huge wilderness hexcrawls. The only obligatory element of Sandbox structure - at least as I'm using the term here - is that the GM seeds plots rather than planning outcomes, and lets the currents of emergent play do whatsoever it will (with a judicious eye toward making sure things stay engaging and fun for all involved - Sandbox is not code for absentee GMing).