an article praising improvisation in GameMastering and more or less discrediting the usefulness of preparation. Or, we should say, OVER-preparation. As GMs, we can sometime do too much planning, which can lead to frustration for the players (railroading) and the GM (elements left unused). In the comments section, one reader challenged to write the opposite side of the coin - just what IS worthwhile planning, and what's a waste of your time? I'll take that challenge, thanks.
There's one hard and fast rule, essentially: If it's not going to be used, don't spend time on it. The rest of my advice really just trickles down from this one rule.
Maps, for example. GMs love their maps. But maps don't matter unless there's an element of exploration to the scenario. If your heroes live in a city, and need to get to City Hall, then guess what, they just do. A fight in a hotel lobby only needs the environment to be described, and unless the game's mechanics require measured distances and miniatures, really doesn't need to be a full diagram. I mean, I get it. We likely all cut our teeth on dungeon crawls. That's where we picked up the habit, and it's a kind of plot cheat, in a way. It's a map in lieu of a scenario. The scenario is simply: grind through this space. Explore, kill, loot, next room. And while there's some joy to be gotten from that, it's not the be-all and end-all of role-playing plotting.
Insuring a rhythm is important. You may want to plan out your "encounters" in such a way as to provide both action and role-playing opportunities for different types of players, but also to give a session variety. Planning a quick rundown of your scenario is quick and easy and provides a solid structure on which to then add details. Start with a hook, then try to alternate action (fights, chases, physical obstacles) and non-action (deduction, social interaction, puzzles) beats, and make a note of the exact type of action/non-action moment it provides (and the time any of these might take). At a glance, you'll see if you've got too many fights, for example, and either streamline the build-up to the "Level Boss" or switch something to a chase instead. Or is there too little action, which will get boring for the player who doesn't really get involved in puzzle-solving? Once you've got the balance right, you can hang things on that "tree".
The set-up is important, but the resolution isn't. What hooks the players in, what events occur without their agency (the villain's plot, what has gone before, the fact that an earthquake hits no matter what), that's all worthy of your attention. Once the players are thrown into the action, however, there's no way to plan for every possible contingency, so having an idea of what COULD happen and having a cursory idea of how NPCs would handle it (or at least how to quickly show a false lead to be a red herring) is all you really need. One of my favorite moments ever is one where my Doctor Who RPG players beat Siren-like aliens by countering their song's frequency with Queen's Bohemian Rhapsody played loudly from a boosted ipod. I set up the circumstances in the first part of the adventure, but let the players take the lead in the second half, at that point simply adjudicating difficulty numbers for their own solutions. Memorable COOPERATIVE moments ensued. Concentrating on set-up rather than climax has the dual advantage of not railroading the players AND reducing the GM's frustration when all his best ideas become the "road not taken".
Key NPCs are, in fact, more important to pre-prep than plot elements. If you know who your villain and allies are, you can drop them into an circumstance and make choices for them. They are loaded with a set of attitudes, but aren't programmed robots that follow a specific course and no other. If they are, then the players should be able to exploit that, but you don't want to be railroading the players so they conform to what your villain is actually prepared to do. And again, you don't need to write entire biographies of these characters, just the details needed to play them. Having quick stats on hand is useful for when players decide to interact with a random character, but stats aren't always needed. They want to talk to a store owner? He only needs a quick personality and maybe a "social"-type attribute if he's going to be convinced of something. In fact, you only need stats if there's going to be a conflict of some kind.
What if you don't prepare for something and it comes up? Improvise. But also TAKE NOTES in case players want to return to that improvised NPC or location. One session is really the preparation for the next. We call that a continuing story, or in our line of work, a campaign.
And while a campaign is not a novel, and you don't need all that background stuff - you're not Tolkien! - if it comes to you, there's no reason to not set it down for future use. The point is not to do extra work. Having ideas, out of the blue, that's not extra work. Don't stifle your imagination on account of the above rules. It's about efficiency, that's all, and the more efficient you are as a GM, the more time you'll have to PLAY rather than PLAN.