So to get back to talking about process: once my blank maps are ready, it’s time to start filling them in. That means rolling dice.
I begin by rolling on a set of charts to create a piece to add to the physical shape of the dungeon — a hallway, a room, a flight of stairs, and so on. Over time I’ve added and subtracted and changed probabilities to generate pieces I like at a rate that I like. And I’ve toyed with different charts to specify different details about each of these pieces; how long a hall is, the material a door is made out of, and so on. There’s a balance, I think, between having variety and having an over-complicated process. I’m fairly satisfied with what I’ve got so far.
The trick I’m most concerned with is getting the frequency of improbable events right. For example: most of the dungeon rolls should turn up, as I said, rooms and hallways and stairs. But there should also be the occasional massive cavern — like the Innsdene cave, or the hob-goblin city. Those require a whole set of tables of their own, of course, but what I want to observe here is that the difficulty is making sure that they’re rare, but not too rare. One every so often adds variety. One every few feet is too much.
Finding the balance has required a certain amount of working with the maps to get a sense of how the probabilities work out. At the moment, I seem to generate roughly one major cave per 22 by 28 inch map tile. That feels about right. Not every cave has something as major as a village in it; some of them may be almost empty, like the one with the grim beetles that menaced Innsdene, or like the cave-with-an-island-in-it that Kezia’s found. These caves still tend to provide a focus to the tile, somehow, as though anchoring the encounters that develop around them. I have to work out how the presence of the cave (and its inhabitants) affect the creatures I generate in the nearby rooms. I think it’s worked so far, and I still have any number of possible combinations of size, shape, and inhabitants that can emerge — as it stands, the charts can potentially generate caves that stretch up to something like a mile across.
Another note about the physical process of creating the map: I don’t normally focus on one strand of the map at a time. Say I start with a room, and find three halls leading away from the room. That gives me three possible places to stick a given die result. So when I make a roll on the master chart of ‘what comes next,’ I’ll work out what it is I’m looking at — what sort of room I’ve generated, how the corridor bends, what sort of junction I’ve found, and so on — and at which of those three would be the best place for the new puzzle-piece. Maybe a bend makes sense at the end of a corridor that seems to be heading toward an already-existing wall. Or there’s a big blank space that a room would fill nicely. Or even a space between two nearby halls that the piece I’ve just rolled up with would connect.
I’ve usually got more than three strands going at once. That gives me a range of possible locations for anything I roll up, however odd. But at the same time, I find that it’s better not to have too large a map surface in play at once — for one thing, if I do, no one part of the map ends up as developed as I’d like (this is a surprisingly time-consuming process). But if I work with one map tile at a time, then it all moves forward and the results seem to harmonize fairly nicely. The trick here, as elsewhere, is to find the right set of constraints to encourage lateral thinking and problem-solving without invoking constraints that thwart the whole process.
Next time, a few words on how I generate the creatures, treasures, traps, magic and other stuff that goes in the rooms that I roll up.
