Notes on Creating a Zylarthen Dungeon: Part I

1. Embrace the Random: I propose that creating a dungeon is a bit like writing a novel. For many people, starting one is intimidating because they assume that you have to have all of this stuff in your head beforehand. Most people are incapable of having all of that stuff in their head beforehand, so they get scared off. What they don’t understand is that much of the ‘stuff’ will emerge organically—later. You don’t create by setting down what you’re already thinking, but to discover what’s, so to speak, already there. Random generation is the engine that allows you to discover things. In Vol. 4: The Campaign, I cite the comments of C.S. Lewis on this. The other stumbling block is feeling responsible for every decision. Should I put this monster or that monster in that particular room? What is the right thing? What if I do the wrong thing? That feeling of responsibility can be daunting. But random generation takes it off of your shoulders. Or so it seems. It’s all sort of irrational. But it’s how many of us think. Or at least it’s how I think.
2. What Initially Reminded me to Embrace the Random: In his introduction to the seminal contemporary megadungeon Barrowmaze, the author, Greg Gillespie claimed that much of the dungeon was generated randomly. I thought, wait, that’s legal? You can create a dungeon randomly and then sell it for money? Count me in, man!
3. Level Variance: On P. 9 of Vol. 4 I laid out tables of monsters for each dungeon level. Each table spanned six or more monster levels, such that you could, say, get 4-24 of the weakest vs. 1 of the most powerful. I then wrote: ‘For greater variation, consider a wider possible spread of monster levels: die 1 = one (table) level lower, die 2-5 = specified (table) level, die 6 = one (table) level higher.' For this dungeon I increased the extremes such that die 1 = two levels lower, die 2 = one level lower, die 3-4 = the same, die 5 = one table level higher, die 6 = two table levels higher. Thus, on say the 3rd level one might encounter anything from 3 Giant Rats (at the absolute lowest possible end) to 1 Basilisk (at the absolute highest end). Is this variation good or bad? I have no idea. Keep in mind, though that ideally the players will often not be able (at least at first) to easily identify the level of difficulty of the monsters. How powerful are those frothing mammals swimming through the water at us? Who knows?
4. Non-Random Patterns: For new monsters, I decided that there would be a 1 in 3 chance that they would be the same as previously encountered ones. Die 1-2 = a random roll of all monsters previously encountered in the entire dungeon in numbers appropriate to the level, die 3-4 = a random determination of previously encountered monsters on that level, die 5-6 = a random determination of previously encountered monsters in that section. This creates meaningful patterns that make sense. Hopefully the repeating ones will be interesting and not boring. If they seem boring, perhaps there’s a way of making them more interesting.
5. A Mistake: I screwed up slightly on the algorithm for treasure not guarded by monsters. It’s supposed to be 1 in 6 of all unoccupied rooms. But my Excel randomizer algorithm only did 1 in 8. I will reroll unguarded treasure for the first two levels.
6. Just So You Know: That said, obviously much of the flavor, tone and mechanical interest will come from traps and features added later. But the monster distribution helps to trigger ideas for this.
7. Why are there so many Martians? Is this a good thing? Out of the 99 monster assignments in the first two levels, 6 are Martians. This is very close to precisely what the tables would predict, as Martians make up 1/18th or 1/19th of each table. The Zylarthen dungeon monster encounter tables are not naturalistic. That is, I didn’t set out to have them track some previously decided scheme of proportions. Rather (with only a few exceptions—dragons, most obviously) each monster is represented precisely once in the tables (and I decided that the different ‘colors’ of Martians would each be a separate monster). Each table for each level generally includes one monster from each of 15 or so categories—getting 'weirder' the higher the die roll. Thus, there is generally 1 mundane mammal, 1 giant insect, 1 prehistoric, 1 undead, 1 Martian and so on in each table. Since the humanoid Martian species are clustered in the lower tables (as opposed to the Orluks, White Apes and so forth in the higher ones), there’s roughly a 1 in 20 chance on these tables that a new monster will be a Martian. I guess I think this is fun. But other dungeon creators are of course free to reroll if they find this bothersome.

More notes in a subsequent post…