Disclaimer: Some of the links below are affiliate links (meaning I get a small percentage of sales without extra cost to you).
I have been running megadungeon products a lot. I've run Barrowmaze (the first volume) for two regular players in person, I am still technically running Dwimmermount for a face-to-face group (although we only played like 3 sessions each year, and none so far this year), plus I am actively running Rappan Athuk online (we spent about 20 sessions down there, but the campaign is on the cusp of turning into something very different).
I have tried to summarise my thoughts about running megadungeons, focussing on some of the issues that I have. I must add that I immensely enjoy reading and running megadungeons, so these concerns aren't about invalidating the concept but rather things that if solved would make it an even better, smoother, and more rewarding experience. I should also add that these issues may come up in other types of adventures as well; they just seem exacerbated in megadungeons.
TL;DR: (1) consider the costs and benefits of grids and gridless dungeons; (2) add environmental cues around branching off points on the map; (3) use observable warning details around traps consistently; (4) secrets doors should always be interesting; (5) random encounters should have details that you can build on
Grid versus No-Grid
When I ran Barrowmaze and Dwimmermount, exploration was handled absolutely by the book. We rigorously kept track of turns, torches, and encumbrance. We counted the squares and advanced time based on the distance moved and the party's speed. Even though the Rappan Athuk maps also have a grid, I only tell my players rough distances and sizes (short, long, small, and large rooms/corridors, and their comparative forms). My players make wonderful maps, but the lack of exact distances makes them more like graphs than traditional maps. They represent connections in space but not in time. As we aren't using Roll20 or FantasyGrounds, this approach has quite a few differences from the traditional way of running dungeons:
- no exact distances mean less emphasis on accurate mapping; less pressure on the mapper makes it a more desirable player role
- describing rooms and corridors is easier for the referee, and information not related to dimensions can be conveyed more quickly
- map-based challenges (i.e. secret rooms placed in the "holes" left by other rooms) are extremely unlikely to be solved without other types of cues
- unless the referee counts the squares behind the screen, movement rate has a less consistent impact on the game; simply assuming exploring a room or passing through a longer corridor or a handful of rooms without interacting with them counts as an exploration turn seems like the way to go in the long run
Loops and Alternate Routes
We all know how important loops and branching off points are when designing dungeon layout. Dwimmermount is fairly decent in this regard, and even some of the Rappan Athuk levels showcase good loops (although in Rappan Athuk it's mostly the stairs and other level connectors that provide important decision points). Barrowmaze is definitely the weakest of the three; I had to add a handful of extra corridors, some of them secret, to counterbalance its oversimplified layout.
However, that just not enough. Even though making navigational decisions is nice, it's nearly not as good as making informed navigational decisions. In most published megadungeons, there is little or no indication as to what lies beyond a door or why the party would want to go left instead of right. Either the referee plants extra clues or they restrain themselves to environmental cues obtainable through a roll (usually a listen check in the published modules), which in itself is time-consuming due to layout (i.e. finding out information about the contents of adjacent rooms often warrants flipping multiple pages). Most intersections and doors are like this, in actuality.
The way I intend to solve this problem in my own megadungeon project is partially inspired by Dwimmermount, where the environmental cues are more or less consistent (e.g. red doors made of areonite signal that these parts were built by technomagical elves, so challenges and rewards would most likely follow suit). My notes are full of inscriptions and recurring pieces of decor, and even the most vicious traps follow a noticeable pattern, so that observant players would eventually learn to recognise them. My intent is to amp up the puzzle-solving part of dungeon exploration, without necessarily robbing it of the sheer wonder we all love about it.
Traps and Secret Doors
A lot has been said about good trap design, but I have yet to see a product that consciously follows through. The traps in Rappan Athuk are simply atrocious. They are often devious, but their deadliness actually promotes low-interaction play (except for the most careless of the brave), because the rewards of a level aren't in line with the deadliness of the traps. I feel like Barrowmaze and Dwimmermount are slightly better, but they too lack the necessary amount of warnings and signs surrounding the incidentally deadly traps.
Used sporadically, I can actually accept the simplest traps. There's nothing wrong with reinforcing the paranoia. People should expect that any door, chest, or floor tile might be trapped, and as long as it's not just a save or die effect but rather some resource tax, it's actually okay. Heck, no one should be surprised if you get cursed, struck by lightning, or summon a divine guardian when you steal off an altar.
Simply noting a secret door on the map I will never accept. All three of these megadungeons are guilty of this, and it makes me sad. Secret doors should be an opportunity to invite the players to interact with the contents of a room, yet most of the time it's assumed pushing a random spot on the floor/wall makes part of the wall slide aside. It's utterly unimaginative, and it promotes the kind of rolls (the dreaded perception checks, uhm, sorry, "find secret door" rolls) that most designers preach against vehemently. Without any guidance from the text, running these secret doors devolves into pixel bitching, which is just as undesirable.
I love the concept of random encounters. My problem is that most dungeons have very boring encounter tables. I don't need "1d6 ghouls", I need "1d6 ghouls dressed in old-fashioned noble garments" as my imagination immediately starts racing. I need these traits so that I can handle non-violent interactions more easily. I could (and do) come up with such things on the fly (it turns out gnolls in Rappan Athuk smoke weed, for instance, and the undead guardian of a great treasure on the first level of Dwimmermount is a soldier who still thinks the empire he once served is strong), but on my own I tend to gravitate towards the "surreal dreamscape gonzo" or the "gonzo with lots of media references", which doesn't always fit the campaign.
I understand how someone might think that omitting such details would lead to easier adaptability (and to a certain extent, it does), but that raises the question: how exactly are these better than the standard encounter tables in the rulebook? The very least, a simple table of a few possible activities would be nice. The other alternative is giving the creatures such clear motivations that it becomes fairly easy to improvise context. This attempt can be seen in Dwimmermount (as most wandering creatures belong to a faction), and from the three megadungeons I have run, Rappan Athuk is the weakest in this regard.
I have also read a few other megadungeons, and I actually want to run some of them at some point (particularly those that might offer further insight into running and designing megadungeons):
- ASE is very captivating thematically, but neither the gameplay nor the format it's written in is that much different from Dwimmermount or Barrowmaze (which, in this regard, offer almost the same sort of experience, except that Dwimmermount has a deep history and has vertical levels). The special rooms are amazing, though, so I might just end up running it at some point anyways.
- Castle Gargantua is a procedurally generated megadungeon, and as such it differs greatly in presentation from most published megadungeons. I also appreciate all the unique monsters and items packed in there. It seems like generating stuff on the fly is time-consuming, so I would either do that in advance, or just use a software to do the heavy lifting for me.
- Dungeon Full of Monsters is a pretty cool modular megadungeon. I haven't read it in its entirety, so I only have vague impressions. I love the general layout. The texts seem to be brief and useful. I think I should probably run it at some point just to see how different it feels compared to traditional presentation.
- Mad Monks of Kwantoom is another procedurally generated dungeon. It's actually a solo game, but I think it could be played reasonably well with a small party of players. My issue, again, is that such detailed generation on the fly is probably too slow.
- Stonehell and The Castle of the Mad Archmage would be interesting to run to experience first hand how their minimalist presentation helps (or hinders) the referee's job. I am not that much inspired by their content otherwise, even though I have no doubts they would produce a lot of cool anecdotes.
- The Forbidden Caverns of Archaia is something I'd really like to love, but the presentation just doesn't click with me. I find it hard to read the regional map, and that would definitely hurt my ability to convey the necessary visual details for my players.
- The Maze of the Blue Medusa is way too imaginative to pass on, and I suspect it would feel very different in play compared to any of the above mentioned megadungeons. I'm especially curious how much utility I can get out of the PDF in my online games.
What issues do you run into when running megadungeons? How do you solve these issues? What other published megadungeons should I add to my run_me list (based on what they could teach me about design and refereeing)?