Two Week Megadungeon

I recently committed to running a fresh campaign, which required I construct a megadungeon from scratch in about two weeks. It was a big job. I was a little worried it was too big to get done in such a short time, but I managed it with days to spare. Some of my players have asked me to explain my methods, and I will endeavor to do so here.

Like any other creative project, I’ve found the best way to accomplish my goals is to lower my expectations. When I say I created a megadungeon in 2 weeks, I do not mean that I meticulously crafted some beautiful bespoke adventure module with hundreds of pages of room description. I can barely conceive of how I’d use such a thing at the table, much less how I’d create it. The dungeon I did create fits on 7 single-sided sheets of paper. One page of random tables and rules reference, and 6 pages which each have the map and key for a single level of the dungeon. Even within that drastically limited scope, I only completed my megadungeon by taking every shortcut I could find.

First step was to source maps. There are more freely available maps online than anyone could use in a lifetime. For this project I went with old reliable Donjon’s Random Dungeon Generator. With it I could quickly produce as many maps as I needed. Plus, generating all the maps with the same tool meant they had a consistent style, which is useful for avoiding all sorts of little annoyances that have cropped up in the past when I’ve tried to force maps from different sources to be part of the same structure. Somewhat arbitrarily I settled on starting with 6 levels to the dungeon, my only real reason being that it felt large enough to be a megadungeon, but small enough not to stress me out while keying it. The Magic Number Seven, Plus or Minus Two is probably a factor here. I played around with the generator’s settings quite a bit between each level. More densely packed rooms on this one, fewer rooms on that one, this map shaped like a cross, that other one is a circle. There’s plenty of knobs to tinker with, though I think only three of the settings I used were actually important:

  1. Straight Corridors + Remove all dead ends. These settings produce the most direct corridors. Even still they’re more meandering than I would prefer, but beggars can’t be choosers.
  2. No stairs. This is perhaps a minor point, but I wanted to have direct control over where connections between floors would exist.
  3. Map Style: Graph Paper. This is by far the most important setting. Not only does it waste the least amount of ink when printing, but it leaves a lot of white space available for making notes.

Once I’d gotten 6 maps I liked, I printed paper copies of each so I could easily make alterations and notes. On each level I added an elevator shaft adjacent to one of the rooms, usually one near the center. One of the most important elements of megadungeon design is that players have direct access from the entrance to many different parts of the dungeon. Otherwise, every single session begins with the party moving through the exact same set of rooms, which can be difficult to prevent from becoming tedious. The elevator shaft means they have 6 choices for what their first room will be each session. In a more traditionally themed megadungeon a grand staircase would work just as well. Within an hour of starting the project the bones of the dungeon were assembled.

Pixelated, low-color photo of an empty dungeon room, in the style of old DOS games.

Now I had ~225 blank rooms to key. That’s a lot of opportunities to get tripped up by Blank Page Syndrome, so I started by creating some touchstones that would help inform my keying later on. The first and least important of these is the dungeon’s origin. Who built it? Why? What calamity caused it to stop being used for its original purpose, and become a dungeon? These are all good things for a referee to know, but it’s also important not to get caught up in them. A dungeon in a ruined hospital may have a lot of beds and broken medical devices in it; and a dungeon in a ruined school is going to have books and desks; but that’s all filler for empty rooms. The treasures, traps, and creature encounters are going to be derived more from what happened to the structure after it stopped being used for its original purpose.

Next step was figuring out tyrants and factions. My six level dungeon has three wizards and a dragon as its tyrants, because as we all know a 2 is always a dragon, and a 12 is always a wizard. These primarily exist as big scary random encounters, but they also need a lair. I place these by writing directly on the map, aiming to be as evocative as I could. Instead of “Merlin the Great’s Lair,” I might write “Mirrored Palace of Merlin the Moody.” Later on the personality and style of each tyrant will affect what sorts of rooms are nearby.

I use this same style of keying through the process, writing the text directly on the rooms they describe. It limits how much detail the key can have, and in particular means small rooms must have more simply described contents than large ones, but these are limitations I can live with. In fact I’d say this limitation forces me to produce better work than I otherwise would. When I’m running games I quickly get frustrated whenever the forward momentum of the action is disrupted because I can’t find the information I want in a block of text. A few words, or maybe a sentence is enough for me to construct a shared imaginary space for my players. That’s especially true here where I’m writing notes for my own future self. Nobody else needs to understand this dungeon.

One major mistake I made here was writing the room keys in pen. I usually prefer to write in pen because it forces me to commit to what I write. I enjoy not being able to go back and change things, and instead be forced to adapt to whatever decisions I’ve already made. I feel the same way about keying the dungeon, BUT, many room descriptions will need to be altered during the restocking process described below. So for the love of the gods don’t make my mistake, key your megadungeon in pencil!

Chrono trigger screenshot. The primitive humans and the lizard people are good examples of opposed factions.

Factions are the most important megadungeon touchstone. The way different groups interact with the players and with each other is a huge part of what I enjoy about megadungeon play. The players will forge alliances with some, go to war with others, play diplomat to negotiate peace between enemies, or convince groups to wipe one another out. Like tyrants, factions control some rooms, influence others, and provide fodder for the encounter table. Coming up with factions is simple: some kind of creature + some kind of distinctive behavior = a faction. The Boastful Bovines are a braggadociously proud community of cow-people who live on the first level of my dungeon. I came up with between 2 and 3 factions for each floor, recorded the name of each faction using a specific color, and placed colored dots on the map to indicate the center of each faction’s power. This is the room where they are most safe to eat and sleep and conduct their private affairs. The colors aren’t necessary, but I often want to know what faction the players are most likely to encounter in a given room, so being able to quickly reference what faction is closest is handy.

By now every level of the dungeon has multiple points of interest I can riff from when keying other rooms. There’s no reason you couldn’t add even more if you like. All mine were social, but you could create all sorts of different touchstones. Perhaps one room could be where some tyrant of faction used to live, and they left traps, treasures, and monsters behind. A portal to hell would make a good touchstone, as would a magic crystal that absorbs all heat, a massive zone of silence, a miniature black hole, really anything that would have a ripple effect across multiple surrounding rooms will help make the keying process easier.

There’s still one more way we can break the problem down before we start keying in earnest. Broadly speaking every room in a dungeon will fit into a fairly small number of categories. Some will have creatures in them, some will have traps, some will have really weird stuff, some will have treasure which may or may not be guarded, and most of the rooms will be empty. That last bit may sound counter intuitive, but it’s vitally important. In this context “Empty” doesn’t mean a completely bare space, it just means a room without monsters, traps, treasure, or special weirdness. A kitchen would be an empty room, but it would still have pots, pans, cupboards, salt, vinegar, and knives in it. These rooms are important because they are where the action will spill into when the players retreat from battle. Empty rooms are where the players will rest, where they’ll devise wild plans for how to make use of something that was intended to be set dressing. Empty rooms are where the players make their own fun. The ancient texts actually advise having 60% empty rooms (1st edition AD&D DMG, Appendix A, Table V-F). I like 50% because I’m a dense dungeon kinda boy.

Photo of the water bottle described in the text. Light green, twist top, printed trees and an americorps logo.

With my desired ratios in mind, I made this little table to get myself started keying:

1-10: Empty
11-13: Creatures
14-16: Creatures with Treasure
17: Trap
18: Trap with Treasure
19: Something Weird
20: Unguarded Treasure

Each day I’d sit down with my maps, throw a fistfull of d20s, then key one room for each of the results I rolled. If I felt like doing more when I was done, I’d throw the dice again. As I got into the rhythm of things and the maps started to fill out, the dice gradually became less important. A few times each day I’d pull my papers out to spend maybe 20 or 30 minutes writing keys. Whenever I got stuck I’d pick something, anything, and do some free association. As I write this there is a water bottle beside me on my desk. Perhaps the dungeon could have a room filled with water? The bottle has trees on it, maybe there could be a room with a tiny forest, or a meticulously cared for garden? The bottle also has an “A” on it, sorta like the scarlet letter. I could have a room with some prudish creature in it which calls everyone a slut. Just like that I’ve got three dungeon rooms from a single object on my desk. You can do this with anything. I’ve personally gotten a lot of mileage out of using Magic: The Gathering cards for this. I also did a fair amount of stealing room ideas from other dungeons, video games, and blog posts. This dungeon is for private play, not publication, so plagiarism rules don’t apply.

As I keyed I jumped all over the six maps. If you haven’t already, you should free yourself from the expectation that your megadungeon’s layout will make logical sense. That’s impossible to accomplish, and boring to attempt. These are mythic spaces that run on dream logic. If I were to key adjacent spaces one-by-one, my brain would try to force those rooms to conform to some pattern. My goal is to create a wild network of weirdness first, then once that’s done I can let my brain off the leash to do its pattern matching thing. This produces wildly more interesting results for me. All of which is not to say that you should ignore inspiration that comes from referencing nearby rooms. That inspiration can be good, and it’s why we created the touchstones. Jumping around just helps to avoid a sort of tunnel-vision design, where you become bound to obey the logical implications of whatever room came before. Jumping around is also an easy way to create multi-room problems. You can simply place “THE BLOOD GATE” somewhere on level 3, then jump to level 1 and write “TOILET WITH THE BLOOD KEY IN IT” on some empty room. In 10 seconds you’ve created an adventure that could keep your players busy for a whole session.

Once the rooms are all keyed the dungeon is ready for play. Of course you’d also need an encounter table, but I have discussed my process for making those elsewhere. The only notable difference here is that I used 2d3 rather than 2d6. The fundamental method remains the same, but the ranges are somewhat compressed.

Here’s an example of what is produced by this dungeon crafting process:

Fully keyed dungeon map with a couple dozens rooms, and descriptions written on them in pencil. The three factions are Goblin Boomers, Ninja Slugs, and Jealous Banditos

If you’re looking at this and wondering how anyone could run a good consistent dungeon crawl from it, remember that the only person who needs to understand it is me. If you made a dungeon with this same process then the only person who would need to understand it would be you. When I write something like “Blue Chest, Difficult Lock, 2 Magic Spells & 1 Item,” I know that my future self will interpret that in a certain way. A Blue Chest implies the existence of a Blue Key somewhere. A difficult lock would mean the lockpicking check is penalized, perhaps by half the value of the sublevel the room is on. So a difficult lock on floor 2 would be at -1, a difficult lock on floor 4 would be -2, etc. There’d be some random junk in the chest in addition to the treasures, maybe some moth eaten linen? The 2 magic spells would be on scrolls, and I’d roll some dice to figure out what spells they were. The magic item would likewise be randomly determined off whatever table is most convenient. I might even ask my players if one of them has a favored magic item table they’d like to roll on.

Even so there’s a lot of missing details to that room description, but if they come up I’ll figure them out on the spot. At any given time during play I’m probably mentally filling in a room’s description. A good 40% of my game prep happens while I wait for players to argue about what they want to do next. There are days when I’m able to do this really well, and days when I’m not. I wish I was always operating at peak performance, but it has to be okay for the game to have a bad day. And the longer I run games the better I get at doing this. My worst sessions these days are still better than the best ones I ran 10 years ago. I think my players mostly come away from my games feeling good about the way they spent their time. That’s the important part.

A megadungeon is a living space, so even once it’s ready for play it will never really be complete. After each delve I review the rooms my players visited and consider what changes may result from what they did there. If the players kill a monster and steal its treasure that room might become empty for awhile, or a different creature with different treasure might move in. If they only injured the monster, then perhaps that monster was forced to make an alliance with a nearby faction while it healed. Now that faction will be ill-disposed towards the players in order to maintain their alliance. If a player character died, their meat might be sold in a cannibal market, or the necromancer 3 rooms over might raise them as a vampire to go hunt down their former compatriots. Restocking like this is how the players get to see their impact on the game world. I don’t get the opportunity to be a player very often right now, but when I do play, it’s this sort of thing that keeps me coming back to a campaign session after session.

In the same vein I also have a notepad where I record all the seeds my players plant. For example: if the party kills a group of goblins but lets one get away, I’ll record the existence of this goblin in my notes. Next time they roll a goblin encounter I might choose to double the number of attackers, and have that escapee leading them in an ambush. If the party funds a troupe of musicians, later on the party may go to a bar and discover those same musicians are performing on stage. Or maybe they won’t. Some seeds never bear fruit, and that’s okay. Simply restocking rooms and noting player seeds takes care of the vast majority of prep work I need to do for future sessions.

Eventually, if the game runs for a very long time, or if I get bored, I may expand the dungeon. I’ve already placed a second elevator shaft somewhere on the bottom-most level which could eventually lead to another 6 floors. Smaller expansions might be accomplished by having a wall collapse somewhere to reveal a whole new area beyond it, or some wizard could open a magic portal to a far-away locale. It depends very much on the direction the campaign takes over the coming months. For now I’m happy to ride the wave of the player’s throwing themselves at what I’ve created.

Monochrome green image of a dungeon door. A text prompt asks "ENTER THE DUNGEON?" Someone has typed "hell yeah" in response.

I must thank Ava, Anne, and Elias for prompting me to write this. I didn’t think I would have a lot to say when they first suggested it, but it got me digging deep and forced me to put words to some concepts I hadn’t bothered to articulate before. As it happens I’ve been working on another daunting project, a set of d100 tables that would serve as prompts for building better dungeons. It’s turning out to be the most substantial project I’ve ever worked on for Papers & Pencils, but in retrospect this post will serve as a good introduction. Those tables still need a lot of work, but they’ll likely be the next thing posted here.

Thanks for reading. I hope everyone is staying safe. Black Lives Matter.

8 thoughts on “Two Week Megadungeon

  1. Hot fucking damn, this is totally rad!

    Some of the things I’ve bounced off of when trying to create megadungeons in the past are neatly handwaved away with your approach. I love it.

  2. I’ve been running a lot of megadungeon content recently and generating it on the fly, and the one thing that has already struck me is the difficulty in handling room encounters that are *close* but not *in* this room. It’s the classic issue of goblins waiting in the next room while there is explosions and screaming going on next door.

    I’m currently planning/writing a boss-based megadungeon, but rather than having *any* fixed encounters, the wandering encounter list is much, much larger, and specific to each faction-zone. You might have a few encounters that are fixed to specific locations, but only if that encounter is immobile, or if it’s specifically a lair. Most encounters are roving scavengers in the megadungeon, just like the players.

    1. That’s probably the best approach if you’re randomly generating things on the fly, for sure.

      In my case, the dungeons are pretty active and densely packed with noisy creatures, so the PC’s noise usually doesn’t rise to the level of drawing attention.

Leave a Reply

Your email address will not be published. Required fields are marked *