Striking the balance between preparation and improvisation

So, my dirty little secret is that I'm terrible at preparation beforehand for my sessions.

No, really, terrible. That's why I started this blog - in the hopes that it would force me to work on the Temple instead of procrastinating. (It's working, though the conversion is slow going what with my life up in the air. Still, I get a little done every night.) In fact, I'm so terrible at it that more often than not I find myself just completely winging sessions.

Fortunately, I'm good at improv. During the last WorldCon I ran for a table of players who ambushed me with a game. (They called me on my phone and asked me to come to the con suite. Once I was there, they wouldn't let me leave without running something. Crazy people.)

However, I don't feel comfortable relying on improvisation entirely when running a game I want to be consonant with the tenets of the OSR (so far as such a disparate hobbyist group has tenets, anyway). Why? Improvisation leads easily to things that look like illusionism, or, in long-running cases, to a world that is inconsistent. (You try remembering the intimate details of an NPC's temperment or the constitution of a town square from six months ago and see how well you do.) The crux of the matter is, giving power to the players relies on them standing on firm ground at least in some respects; they can't make meaningful decisions if any given previously-reliable piece of information turns out to be unreliable because you're not writing things down.

A little digression: I'm not talking about preparation like creating balanced encounters or plots. I don't play that way. I try my best to prep situations, not plots, and so I'm not getting bogged down in the details of how to predict player actions. Rather, I just have difficulty with my innate laziness.

Anyhow, the problem with improvisation is that it doesn't lead very well to a consistent world in which the PCs can move, and it's downright terrible at pro-active offering of options, which are the vital component of a game that purports to be player-driven. Unless the players can know something beforehand to guide their decisions, they aren't really making choices; they're just flipping coins.

That said, by no means do I intend to do away with improvisation. In fact, I cherish it; it breathes a life into the game that otherwise wouldn't exist. Reading the oracular dice (or my internal mercurial demon, which is frankly even more random than the bones) brings the world to life and provides entertainment for me and for the players. Furthermore, it is the other vital component to allowing player action to matter; if everything in your world is pre-scripted, then the players aren't agents, they're subjects.

How do I balance these competing forces, especially in the face of my natural bent to procrastinate and make it up around the gaming table? I'm still struggling with this question, but so far here are the rules of thumb I've come up with (and that I'm trying to implement):

• Build important NPCs and decide why they're important. This means, most accurately, who they are in the situation (e.g., village elder, blacksmith, ogre on 3rd level) and with whom they have what relationships. This does not generally mean "why they are important to the PCs" - I figure the PCs themselves will decide that. If they're combat NPCs, stat up their combat-relevant abilities and tactical inclininations, as well as treasure.
• Build important locations and sublocations, with at least a few hooks for the PCs. This changes a lot based on the nature of the location - if it's a dungeon, it must have rooms, and a story/rumours/inhabitants/surroundings for the PCs to know about. (There's a lot out there about how to make dungeons.) If it's a dungeon room, it has details on geography, current inhabitants and uses, traps, and treasure. (Bonus for history, but that can be made up on the spot if necessary.) If it's a village, it has relationships with surrounding areas, local geography, and starting attitudes of citizens, as well as at least a vague idea of the strength and focus of the economy.
• Build tools for randomness. This means random encounter tables, reaction tables, and the like. Only focus on those things it wouldn't be fair for me to make up on the spot, or for things I'd be tempted to bias, or for things I honestly can't decide. For example, random encounters with monsters is good tool fodder. So are rumours, though somewhat less so. NPC reaction tables are good, but I've no need of NPC knowledge tables. I can decide fairly well what a local tanner knows about The Crypt of St. Genevieve.
• Do other preparation if I want, but that's not necessary.
• Once the session is over, write things down. I struggle with this at times.

I'm curious what others do. I'm aware that after-action reports are popular; do folks have other methods they employ in addition or as substitutes for this in order to track important information? What sort of pre-session preparation do you engage in, and how do you do it? Do you find that it helps?

P.S.: I hope to have more Temple-centric posts soon. I'm just trying to avoid giving away too much information to players who might read this.