Friday, August 26, 2011

Top-down vs Bottom-up

Here is something I was thinking about for some time now but never had the catalyst to blog about. In a sandbox campaign and in a way in classic campaign settings to you often start with the big, important factions. You place your Leach King in one place and the Orc Alliance in another, add a kingdom or two each for every race. The idea is to have those big powers generate conflict between each other and for the PCs to get involved in those conflicts. You often add successively smaller powers like local clans, noble families etc.

I'll call this the Top-Down Approach because it reminds me of top-down parsing in programing language compilers. Whether or not the setting is built before or during play is not important here. What matters is that you as GM always have the big picture about who is doing what in your game world. Maybe the players do not know why the Black Eye clan is besieging Farnburough Castle - but you as GM do know in one way or another. Everything that happens has a known (but maybe unrevealed) reason withing the game world. Everything is - at least to the GM - causally plausible.

Now imagine a campaign in which the PCs have encountered the following:
  1. In the first session, the PCs dealt with a zombie that haunted the village of Oppendunst every night. They found his fresh grave in the cemetery and killed the poor undead bastard a second time.
  2. In the second and third session the PCs dealt with an onslaught of starving kobolds that raided th local countryside in waves. They found out that the kobolds were suffering from a hunger-inducing illness and managed to kill most of the tribe and force the rest of them to somewhere else.
  3. The PCs also met several NPCs in the local communities.
The PCs suspect Loranna, an NPC they have met, to be involved in this because they think her behavior was weird (even though you never intended that). During postparation, you decide that she is indeed the missing link. She found an object in her garden and, after noticing that the Zombie was obviously looking for it wandering around her house, hid it in the forrest. And that's it, you stop right there.

What is that object? How did it get into Loranna's garden? Why was the zombie looking for it? How did it affect the kobolds? You do not answer any of those questions, in fact you watch out not to fixate on potential answers.

Because here is the thing about the Bottom-Up Approach: You will find out or decide on the answers to such questions during play. If the PCs find a place called Old Kobold's Den in the forrest and you spontaneously decide that this is the place where the Thing is, then it really is! But if the PCs go after the kobold tribe (you make it The Quest For The Lost Tribe) and find it? Well it turns out a shaman had found it and that is how it affected the tribe.

If one of the characters has a backstory involving an old family line stemming back to the age of The Empire of Satin then the Thing could be an Odd Cube inscribed in the Lost Language of the Court. But if the players have more interest in fabulous creatures it would be The Curious Giant Feather That Shines Silver And Blue.

In Top-Down you decide on as much as makes sense to you but reveal to the players only what their PCs would know. In Bottom-Up you decide as late as possible but reveal almost all you know immediately.

In Top-Down you build setting elements starting from the causal root and grow them from there until they touch the here-and-now of the PCs. In Bottom-Up you start at the here-and-now of the PCs and hold off developing the setting until new elements spontaneously and forcefully demand their introduction because the PCs bump into them or because you find a natural fit for what was previously undecided.

I think this is about the same thing as what Zak is talking about. His heavy use of random tables to generate interesting encounters is way to seed a Bottom-Up Approach. The tricky thing to hold off on deciding on what all the loose parts mean as long as possible. It goes against most GM's inclination to plan, map and plot.