I recently found myself rolling for a couple hundred rooms, so I thought it might be worthwhile to note my dungeon stocking strategy. The long and the short of it is, I use the suggestions from the D&D Rules Cyclopedia: 1/3 empty, 1/3 monsters, 1/6 traps, and 1/6 vaguely-defined “special”. I’m no expert on the history of the hobby, but I gather those are the Moldvay rules. Here’s how I apply those rules.
First, I’ll get the map together. The first bit of the dungeon, including the part that the party has explored thus far, I mapped by hand, on five-to-the-inch blue-lined graph paper, in a fit of middle-school nostalgia. When the mapping techniques of the 1980’s encountered the digital technology of the 2010’s, when I scanned those hand-drawn originals, I concluded that the old ways were overrated. Since then, I’ve drawn my maps directly on the computer. I use The Gimp for the maps, making heavy use of layers. As I gain experience mapping this way, I’m accumulating brushes and macros and such to improve the process.
For the most part, there’s not much difference between my Gimp-made maps and my hand drawn, once I’ve straightened out the crooked lines. One thing I haven’t done enough of, as yet, is including extra information directly on the map, like icons for monster lairs or treasure caches.
Once I’ve got the general outline drawn, I import it into SketchUp and do a rough, dirty 3D view, just to make sure the place still fits together. A couple of times, now, this extra step has saved me from having hallways come out unexpectedly halfway up walls on other levels. There’s a lot that I could take advantage of, during this step, that I don’t, simply due to time constraints. (Yes, yes, and my built-in laziness. Entirely admitted. Now, go ‘way, kid, ya bother me.) I’ve tried more than once to set up areas with complex vertical aspects, like balconies overlooking balconies overlooking galleries, usually inspired by one of the subterranean settings in the Lord of the Rings movies, using SketchUp to produce the original and then extracting the flat map from that, with limited success. The problem I’ve run into is communicating the structure to the players well enough that the complex map becomes a feature, rather than a hindrance.
I’ve fiddled with this sort of three-dimensional mapping before. Long ago, I mapped out a partially-constructed skyscraper for a three-way battle using the HERO System rules, using a big vinyl mat and floor-by-floor maps. That fight is still talked about today. Half the combatants could fly, the rules encouraged comics-style knockback, and the groups approached from several directions, so the fight sprawled over the entire map by the time it was done. I think the key factor to this success was the relatively-simple environment: being only partially-constructed, the building was a framework with floors. More a series of stacked boxes than anything else.
Later, for the Space Cowboy game, I used SketchUp to build a model for the PC’s spaceship. Since the campaign setting had no gravity-manipulating technology, the ship had to spin for artificial gravity. I couldn’t find any pictures that I could use for a cargo-hauling slower-than-light interplanetary spacecraft, so I set out to make my own. By the time it was done, I had mapped the entire ship. I used the model to render a books’ worth of views, so the players could see what the characters were looking at. This whole process was consciously aping the way the TV show “Firefly” was filmed, in an effort to make the Cabra as much a part of the party as Serenity was a part of that show’s cast.
It worked really well, largely because the characters spent a lot of time on the Cabra. First, when I put the thing together, I had to get the relationships of the rooms straight in my own head, which helped me immensely when it came time to adjudicate movement around the ship. It also gave me a weird case of vertigo, for a while, where I would look at a room and envision it rolling ninety degrees to one side. That sense of location came in handy in many unexpected ways. Once, someone set up an ambush by taking advantage of the weird blind spots and angles made possible by the ship’s layout. Lots of “day in the life” role-play grew out of the description of what happens inside the chicken coop when the ship goes from “under thrust” to “zero gravity” to “spin gravity”, where “down” moves from the floor to a wall. (Osolo liked to watch the new chicks, the first time they experienced it.)
I’m sure that level of detail would be wasted, in the dungeon. A balcony, sure. Ledges in a cave, great. A giant doughnut spinning so that you stand with your head pointing towards the doughnut hole and your feet aiming away from the center, such that if you start running in on direction, it feels uphill the whole way, yet you still return to where you started? Nah, that sounds like overkill.
So, one way or another, I’ve got a map. While drawing it up, I’ll be thinking about how to use it. I might think, “Hey, it would be cool to use a cave fisher”, and so I would include some caves and some places for a cave fisher to lurk. Or I might look at the map as I’m drawing it, and think “This looks like a good spot for an ambush” or “If I lived over here, I would want a lookout over there”. I don’t usually get to much detail, yet. Moldvay recommends filling in those set pieces now, but I just make a couple of notes, often with lots of question marks.
Then, I roll. I use a dice-roller app, so all I have to do is poke a button and interpret the result. I keep a shorthand tally, just a couple of characters, on my map key. As others have noted, once you’ve rolled a few dozen rooms, you memorize the table and things really speed up. This gives me a long list of “empty”, “trap”, “monster”, and “special”, with or without “treasure”.
The next challenge is to figure out the specifics. I can work on fleshing out a specific area within the dungeon, based on the rolls. If the random results conflict with my ideas from the mapping stage, I’ll often move the results around, rather than simply overrule them. For instance, if the dice say “empty, with treasure” where I thought to put a guard-post, I’ll find some nearby area where the dice call for a monster, and I’ll swap the two random results. My thinking is that this will keep the distribution of results close to the random results.
On very rare occasion, I’ll skip the random roll entirely. I can think of one good-sized area where I knew what I wanted. In that case, the number of monsters is closer to 1/1 than 1/3…
My original plan was to use the random monster tables from the Cyclopedia, but replace the originals with their Dungeon Fantasy equivalents. A goblin is a goblin, a dragon is a dragon, and a blink dog is an enigma.* In the end, though, that started to look like a job of work, without satisfying results. It would be too easy for the dice to have the lion lie down with the lamb. Any random-roll system with enough complexity and nuance to give the orc overlords a bunch of goblin slaves to push around would also be a pain to use, especially over several hundred rooms.
So, I fake it and just write down whatever seems sensible. By this point, I’ve got a pretty good idea of whose turf I’m working on, what sort of creatures I want to use, and how they’re likely to see the area. I’ll use the random rolls to lay down the final details. For example, let’s say the dice indicate that there are monsters with treasure in this room, and this room’s only access to the outside world is through a room where the dice placed monsters without treasure. Perhaps this displays a master/slave relationship between the two groups? The area I’m working on is dominated by kobolds; perhaps these two rooms are a group of kobolds and their giant rat watchdogs. Or maybe it’s the biggest, baddest kobold with the tribe’s treasury with his kobolds-at-arms sleeping out front.
There’s enough detail given by the dice to act as a springboard, while leaving enough wriggle-room to keep things more-or-less logical. A trap can be a booby trap left by other adventurers, a tomb protection measure left over by the original builders, or a pair of magical boots with a scorpion hiding inside, ready to sting the unwary. A monster might be the Lich King of the whole level, a nest of giant spiders, a tribe of hostile humanoids, a rust monster, or a lost party of adventurers. If I want to give them impression of limitless, abandoned, dust-covered rooms, I emphasize the vermin, clouds of bad air, and opportunities for cave-ins. If I want to depict a bustling city of shadow elves, I have a lot more strongboxes with poisoned needles, guards on duty, and evil wizards conferring with demons. The underlying dice rolls remain the same.
The “empty” and “special” results can be tricky. If pressed, I would say that my interpretation is that a “special” is anything that isn’t a monster, trap, or treasure, while still being persistent — it can’t be carried off or cleaned up. It might even be a source of monsters, like a portal that lets in demons, or treasure, like a vein of ore. (Or traps, I suppose, though I can’t think of a ready example.) The big stone heads that the group keeps finding are specials. On the other end, just because the dice say the room is empty, that doesn’t mean it doesn’t have any contents; it’s just empty of monsters, traps, or treasure. I’ll throw around dungeon dressing as appropriate in otherwise-empty rooms. The room with the broken-down dresser, where the party spent some time scratching their heads and casting See Secrets, was an “empty”.
But it could have just as easily been an “empty with treasure”, with a fearsome Search penalty. Or maybe it was a “monster, no treasure”, and the party has ever since been trailed by an invisible stalker. Gotta keep ’em guessing.
* I’ve wondered for decades what those blink dogs are doing on the 2nd level of the dungeon. I just cannot wrap my head around a group of a Lawful, teleporting dogs that hang out underground. It’s a failing, I know.