The Prison Megadungeon: Designing an Escape Megafunnel

Safe at Home 

Prisoners are thrown into The Pit, but the fall does not harm them. In an earlier age, these dungeons were part of The Palace, and visitors would float down this shaft as a thrilling diversion. The same magic persists in the structure's current purpose, but it is merely a small mercy for the condemned.

The area at and around the bottom of The Pit is the sanctuary. Most of the prisoners stay very close to this location. The jailers provide food and water from above, but only enough to keep the prisoners alive. The only other provisions provided are candles. So many candles. The Sisterhood of Benevolent Abstraction receive favorable tax treatment for the production of the candles, so they send scads of them. All supplies are simply thrown into the slow-fall field of the shaft and float down to the waiting people below. The prisoners never even see their captors.

The sanctuary is safe. Just as the shaft is enchanted with Feather Fall, some ancient Sanctuary or Hallow spell shields this place from harm, and none of the monsters of the surrounding dungeon halls will enter this place. No threats from the dungeon can come here. Adventure does not happen here, experience cannot be gained here, and nothing new will be found here unless the PCs bring it there. The sanctuary follows the same principles as West Marches settlements; it is a place between adventures, not a place of adventures.

The sanctuary is a source of new PCs, hirelings, and henchmen. There are always some number of prisoners here. How many there are and who they are is a quantum aspect of the game. They only come into focus when the players need them. 

It is also a place for downtime to happen. Any downtime activity that could be done in a settlement can be done here – if the PCs can come up with a plausible plan for acquiring the necessary resources. Expertise might be available within the pool of prisoners, or might be something a shrine could offer. The people in this space are really more of a resource the PCs can leverage, rather than distinct NPCs the DM will personify (although the DM may adjudicate what these NPCs need in order to do what the players want, and what they will refuse to do, in the same way the DM would step in to control the actions of a hireling who was otherwise directed about by a PC). 

Up Above

Returning up the shaft is not a realistic option. At least, not at first. The walls of the cylinder are 200’ of smooth stone, and there is not remotely enough material at the bottom to build even makeshift platforms. The hole is set in the middle of the ceiling, not close to the walls, and covered by a heavy metal trapdoor that is winched open and closed by jailers overseeing four burly, ribbiting froxen. Beyond that aperture, several layers of protection within the prison itself stand between a would-be escapee and freedom.

All that said, it is not impossible to escape in this manner. If the PCs come up with an ingenious plan, going up the shaft and turning the tables on their jailers should be celebrated. But we presume that doing so will require at least some engagement with the megadungeon itself, to gather either the power or the resources to execute such a daring reversal.

The First Shrine

Near the sanctuary is the first shrine, the home of Carnyhollow. Carny is a “demon” in a technical sense, but lacks the willpower and can-do attitude required of an extraplanar entity to manifest a discernible moral position. It mostly just wants to accumulate wealth in exchange for junk.

At request, Carny can transform an object or objects into something else of one-tenth or less the value. For example, if 50’ of rope is valued at 10 silver coins, Jerry would require at least 100 silver coins worth of treasure to produce the rope. 

He is indifferent to the type of treasure, but absolutely will not haggle or discount his price. 

He will accept magical items, but can only produce nonmagical things in return, and will be conservative in guessing a value for a magical item without a clearly implied value. 

He can only produce generic items and cannot customize them. He can create blank paper, but not specific books. He can create a padlock and matching key, but cannot create a key for a specific door in the dungeon. 

The equipment list for the system in use is a good proxy for what Carny can create, and the DM can use common sense and liberally reject any requests that stray far from it.

Carny is most immediately useful as a supplier of basic equipment and an immediate use for treasure, but the party can consider more elaborate schemes as they acquire more riches.

Any dissatisfaction the PCs may express with Carny's terms should also motivate exploration. It is hardly the only such creature offering such services in the dungeon, and any friction the PCs experience with any particular NPC or antagonist should serve as a motivation to explore further.

Three Tiers of Maps (for the DM)

We’ll go through (at least) three tiers of planning out the dimensions of our megadungeon. 

• A diagram showing only hard and soft barriers or exploration complications; verticality and danger levels; and treasure targets.
• A very loose and sloppy map that converts this diagram into approximate rooms and overlays factions and major points of interest onto it. 
• A proper keyed map that fills in room-by-room contents. 

Three Tiers of Maps (for the PCs)

To create concordance between planning the game and executing it, we will aspire for symmetry. We’ll provide (at least) three kinds of maps that PCs can find.

We want maps to be a highly prized resource. Even the humblest map might at least indicate the position of a trap or threat. In a high lethality game, that could save a PC’s life. 

Recent and improvised. The PCs are not the first to explore these halls. Other prisoners have ventured out into the darkness, hoping to find treasure to trade at shrines, or even dreaming (foolishly) of escape. A map may be a crude series of lines on trash wedged under a door, or hurried chalk markings on a wall, or a self-applied tattoo on a corpse that didn’t make it out.

These maps are helpful for understanding recent and proximate threats and opportunities, but they won’t convey any deeper understanding of the dungeon.

Modern and carceral. Maps were required to turn this place from The Palace into The Pit. Perhaps the jailers were sloppy or indifferent to how this might aid prisoners. Perhaps they were in a hurry to leave this dangerous place and didn’t remove all signs of their work. Perhaps some of them were left behind here, buried with their plans. Perhaps that was intentional. These maps are helpful for understanding barriers and obstacles, and where the divergence between the structure's original purpose and its repurposing can create opportunities for the PCs.

Ancient and original. Beautiful blueprints made thousands of years ago. These maps are helpful for finding secrets and understanding magic, and locating particularly valuable treasures. None of these maps are perfect, because each is of its respective time and ignorant of what came before and after. But they are all useful, because they are foundational, and in many respects contain information not known anywhere else.

The Prison Megadungeon: Into the Pit

The Pit* was something else in the past, something important. But now it is a hole in the ground where The City sends those it does not want. They do not execute you because the dungeon is hungry and it must be fed. The Pit is a prison. It is a dungeon. You are here.

Funnel-First Design

Characters start as level 0 ordinary humans with no special abilities and no equipment. Lair of the Lamb does this well, and it is useful in a number of interrelated ways.

• Character creation and the rules are part of gameplay
• Problems must be solved diegetically (“the answer is not on your character sheet” is pretty obviously true if it is a blank piece of paper) 
• Adding “stuff” to your character increases complexity gradually

Diegetic Classes and Ancestries 

Most funnels work like this: The players each control several level 0 commoners. Many of those commoners die during the adventure, “funneling” the players toward a dwindling group of survivors. At some point the funnel adventure ends, and the survivors graduate into conventional level 1 PCs. 

Most of the funnels I have seen do not require that the events of the funnel relate in any way to the resulting choice. If your surviving commoner’s highest stat is intelligence, you can make them a level 1 wizard, whether or not they had any interaction with the arcane during the funnel. That’s fine, but it is an example of intrinsic growth. Our PCs are nobodies who start with nothing, so we want to feature extrinsic growth more prominently.

So we’ll make the level 0 to level 1 transition more literal. A PC who wants to become a wizard needs to find something in the world that makes them a wizard. And this is true of every step of character creation and advancement. It must be something within the dungeon, not within the character. 

Shrines

What makes our wizard a wizard? If we choose a classless system like Knave, this problem solves itself. Equipment functionally determines “class,” so we already have an extrinsic solution to this problem. 

What if we are running a system with dedicated classes? The aspiring wizard would simply need to do two things:

• Acquire at least one distinctly wizardly item
• Expend treasure at a shrine

Shrines can have different aesthetics and names, but mechanically, each shrine is a place where treasure can be exchanged for XP. And that includes the XP required to advance from level 0 to level 1.

Many systems do not specify how much XP is required to go from level 0 to level 1, but we can reverse engineer it from later level progression. For example, D&D 5E.2014 requires 300 XP to advance to level 2, and 900 XP to hit level 3. We could choose 100 XP as the threshold to hit level 1. 

For a game feature that we want to be readily available to players – like reaching level 1 in a basic class – we want the price in treasure to be low, and the shrine to be close to where the PCs begin. More niche or difficult goals can require either more treasure, or visiting a more distant shrine (or both). Or a very specific type of treasure. 

Hungry Gods

Pelor will not help you. Even if he was canon, his sunlight doesn’t reach down here. If you want divine aid, you’re going to have to ask a dungeon god.

If we’re using Knave, the Knave 2 rules for relics work well as-is for this kind of thing. But what about other systems?

Gods work a lot like the shrines described above. Valuables go in, rewards come out. But gods are picky about what you tithe to them. Or they may want you to do something for them instead. They have more agency than a shrine, which is typically going to be a reactive location, versus the personality we get with a god.

The gods in Lair of the Lamb are good dungeon gods. They’re all freaks, and the worshippers of one of those gods are directly responsible for the PCs’ predicament. But they nonetheless offer power, and the PCs are not in a position to refuse that offer of power without serious consideration.

For further examples, read this post and this one. This post really centers “gods” (really, any powerful entity willing to trade power for something worldly action) in a way we want to emulate. A lot of D&D treats deities as highly intrinsic to the characters who follow them. In many modern-style games, they are really an extension of character creation, moreso than a part of worldbuilding.

Magic

Extrinsic advancement is all well and good for our fighters and thieves. But what about wizards? 

First and foremost, no spells will be learned by virtue of leveling, so any magic a PC is wielding is coming from what they’ve found in the dungeon. That means even a humble first-level scroll is a  major treasure for a low-level PC.

Intelligent monsters may possess such items. What about non-intelligent monsters? Well, a great feature of The Monster Overhaul is the frequent sidebars explaining what monsters taste like, and what magical effect (if any) can be gained by eating them. We can build on that to give wizards a Dungeon Meshi experience, deriving magical power from monstrous meals. We can also give otherwise-ordinary monsters magical abilities, with the understanding that defeating or capturing (or eating) them could be the key to learning a new spell.

Meet the Neighbors

You may have noticed that I said at the top that players begin as level 0 humans. Not elves or dwarves or thrikeen or dire molefolk. Humans. If you want to play something else, you’re going to have to find it in the dungeon. And the most ready place to find non-human PCs is amongst the dungeon factions.

Factions can offer PCs various rewards, but we have an additional one to offer in this megadungeon. Establish friendly relations with a faction, and members of that faction are available to your PCs in the same way humans were. They are a quantum resource for character creation and NPC recruitment, just as the “starting pool” of humans was at the beginning.

That also means that “recruiting” a faction creature as a potential PC can earn the enmity of any factions that opposed the friendly faction. If the rock baboons hate the gecko-kobolds, do you still want to ally with them, gaining them as a resource, but losing the kobolds? It should be difficult to recruit creatures of opposed factions, unless the factions in question no longer produce interesting conflict for the PCs.

Non-faction creatures (i.e., unique creatures) can potentially be recruited too, but there is not a quantum supply of such creatures available as PCs. If a unique creature is recruited and then dies (because, say, a PC used them as an adventurer), it is gone for good.

*We're going to throw down a lot of proper names throughout this exercise. All of these, from The Pit itself down to the lowliest NPC, are placeholders that we can always change later. 

The Prison Megadungeon: Under the Influences

Writing the (ongoing) World’s Largest Rewrite series has been fun, but I ultimately think it is better as a source of monster ideas than something that should actually be finished. But where to put those monsters? A prison megadungeon without the cumbersome "world's largest" trappings. Let's make a megadungeon.

Caverns of Thracia is the megadungeon I have run the most. Arden Vul is the megadungeon I have spent the most time thinking about recently, thanks to the 3d6 Down the Line podcast. Megadungeon guidance in the greater world of blogs is too vast and diffuse to cite fully, but I will credit a few inspirations.

Arnold K’s Lair of the Lamb. It is not a megadungeon, but it is quite strong in treating the prison not first and foremost as a place that adventurers will delve into, but as a place ordinary people will try to escape from. The very idea of doing a prison megadungeon like the WLD and not making it a prison escape scenario is a little strange, after all. I don’t think I even mentioned that in my first World’s Largest Rewrite post. We'll get into this more in future posts, but the concept of this megadungeon is that the PCs begin as prisoners.

The Two Week Megadungeon. I will not be following the steps here precisely, but like LotL, the TWM process emphasizes lean efficiency over exhaustive completeness. That will be important for, you know, actually making discernable progress. 

Designing the Dungeon with Numbers. This post is specifically about the numbers one through five and how they imply different things when used in design. We’ll want to keep this in mind as a counterbalance against the sprawl of the megadungeon. That sprawl is, of course, part of the appeal; no adventuring party is going to explore all of Arden Vul, and no two parties are going to have the same experience there. But that sheer amount of content is overwhelming if we don’t carefully curate how choices are presented.

For example, we don’t want too many areas accessible to the players at first. If the PCs have more than five choices, it is probably more than they can really conceptualize meaningfully. More than five options becomes a list that has to be studied. And a list can easily become a chore.

But a choice between three to five options – with perhaps three to five decision points within each of those initial options – is much more digestible. The PCs will eventually have many more choice points, but these will be presented fractally, as the PCs engage with the dungeon; not all at once, and not too much, too soon.

Landmark, Hidden, Secret. This is one of those great blog posts that can inform all kinds of adjudication and design. We need to make sure that LHS design is baked into the megadungeon. You know how modern-style players are always talking about how the "exploration" pillar feels underrepresented, relative to the "combat" and "social/roleplay" pillars? LHS design, in conversation with strict time records, random encounters/events, and other procedures, helps deliver that exploration aspect of fantasy dungeon games.

The Lost Art of Getting Lost. It is fine to say the dungeon will be lethal, but players are smart and good at playing conservatively. They will probe carefully and back away from the unknown. Doors that slam shut and force the PCs to find a new way back to where they started really test those skills. The same is true of dungeon features like pits, portcullises, and sliding walls. The humble iron spike becomes a lifesaver when it can guarantee a path backward. But, of course, the noise from hammering in a spike triggers a wandering monster check. And when time passes, that faction you offended before starts removing your spikes… oh crap… 

Because our PCs are prisoners, we can also address some of the obstacles noted in the post. We can control the “economy” in a way that is not possible for “open-world” adventuring, so we don’t have to worry as much about torch supply trivializing light. Our PCs will begin with plenty of candles, and no other source of light. Imagine a simple torch as an exciting “treasure."

Metroidvanias and Megadungeons and Information Architecture in the Castle Automatic. Two posts from the excellent Rise Up Comus blog. The first post creates a taxonomic framework for understanding megadungeons, and why some fantasy adventure game features and character abilities do or do not work within that context. The second is on information design. We don’t strictly need to worry about information design, as we are designing this as an experiment, and maybe to run someday, not to publish (haha… unless?) But good information design is its own reward. Every DM can get by with their own chickenscratch notes and kludged together prep process, but it’s worth interrogating how we could do better and make things clearer, even if that just transmits a clearer picture to players.

The Monster Overhaul. I have cited this so many times during the WLD series. A good monster manual is much more than stat blocks; it is a worldbuilding and session-planning tool. And the Overhaul is not a good manual, it is a great manual.

Make RPG history and look insane while doing it. Blades of Gixa is one of the successful projects  that emerged from the Dungeon23 movement. Indexing location and history in the way this video demonstrates is an awfully interesting idea, and is something we can mess around with for bridging the ancient-lore-to-current-events gap, and avoiding the dreaded “this room used to be…” problem that plagues many dungeon products.

Various Cryptic Keyway posts and projects. Of course I can also take from things I’ve already done, whether iterating on a good idea or salvaging a half-finished thing for parts. Even before the WLD series, previous posts have suggested a similar concept to this one. This experiment will keep some of those ideas. And if it fails, it will be the fuel for something else in the future.

I’m not interested in turning the WLD posts into a dungeon itself. I ultimately think the “world’s largest” conceit is a mistake that undermines good dungeon design. But working through those posts has been useful for generating ideas, and I like (with credit and thanks to OSE and the Overhaul, of course).

I can also scrape content off of my aborted Dungeon23 project, as well as several other dungeons started and left unfinished over the years. Those could even still turn into their own things someday, but a currently active project (like this megadungeon) always gets to steal ideas from projects that are dormant.

We’ll add to this list of influences as we go.