Previously, I detailed my usual technique for creating self-contained adventures for episodic campaigns (like Slaying Solomon) or convention one-shots. That technique works very well, but it is far from the only way to prepare for such games.
Dating all the way back to the earliest days of Dungeons & Dragons, perhaps the oldest technique is the classic keyed map. It is possibly the best form of preparation for a certain type of game (dungeon or hex crawls). It is also a bit difficult (for me at least) to adapt to fixed-length sessions outside the D&D genre.
The problem is pacing. As a game master, my goal for a one-shot is to end the session with some kind of satisfying conclusion. I don't particularly care where the party ends up or even if they fare particularly well. But nothing annoys me more than an adventure that doesn't have a satisfying ending.
Going way back to about 2005, I can remember many excellent Dragonspire games (D20 and Risus) where the party explored adventure sites that I keyed out ahead of time. But that was an extended campaign in a non-episodic style, and for that reason I was perfectly fine with calling the session in the middle of a dungeon. But the end of an adventure and the end of a session are different things. I do recall trying (successfully?) to wrap things up nicely when we reached various campaign milestones.
In recent years, I have run only a single adventure off of a keyed map: my Karst Chantry One Page Dungeon (developed for Risus, run using Old School Hack). I had hoped that the adventure would take but a single session. But three hours into the first session, I realized that there was no way in hell that the party would get anywhere close to exploring the entire dungeon. And they certainly weren't going to pull out and retreat to civilization with any kind of consolation trophy.
No, it took a second session to get to satisfying end-point. That, in itself, wasn't so bad. But due to scheduling difficulties, two-part adventures are difficult for our group. We lost one player (who's character went two-dimensional in part two) and gained another (rescued from a dungeon). It work amazingly well, but it took quite a bit of juggling to get there.
Hmm, I guess it wasn't so bad after all. But it is still a form of preparation that I don't get a lot of practice at. I *like* drawing maps (especially ones keyed with DungeonWords). But I think I need to exercise more careful planning to make them suitable for one-shots. How long will it take to explore a given room? What if the players dilly dally? What if they take a short-cut? I think that a good, non-linear dungeon map could use some switches to allow a GM to adjust pacing. Perhaps various side passages become available if the party is making good progress? I'm not sure how I feel about that. It's funny that in non-D&D games, I have no problems stepping on the gas or the breaks to adjust pace. But a dungeon crawl seems to demand some degree of fidelity to what's on the map. Or not. Maybe it's just that more of my old-school sensibilities come out when I'm running off a map.
Of course, there is another idea: put the characters on the clock, or more likely, the dungeon itself. Link dungeon events to the real-time clock. Imagine a dungeon that is sinking into the muck, a self-destruct sequence, a gate home that is about to close, or a dying companion that needs the the dingus in the treasure room to recover. As part of the social contract, the players need to know that the real clock is roughly synched up with the game clock.