Monday, February 10, 2014

You Will Never Finish That Dungeon

Dwimmermount was in the news again.  Tenkar's Tavern, the tabloid news service of the OSR, spurred a discussion last week when it was announced completion of the Dwimmermount kickstarter is changing hands.  Alex M, the publisher and driving force behind the Adventurer Conquer King series of products, is taking over the writing of Dwimmermount.  He's targeting a March-April completion of the draft.

This post isn't really about Dwimmermount.

Actually it is about Dwimmermount, in so much as Dwimmermount represents many problems of the published megadungeon.  There are problems of  time, and effort, and level of detail - what I call the sandbox triangle*.  It's about completing the dungeon.

However, today I'm not talking about the creator, the referee, finishing the creation of the dungeon - putting the pen down and closing the book.   That's one avenue of discussion - but rather I want to talk about the referee's expectations for the players.  Are the players ever going to finish the dungeon?  I had an epiphany the other day.  Consider this whole exercise a variant of the My Precious Snowflake Encounter™  problem.  To wit:  when you sit down and write an encounter, do you expect the players to meet with that encounter?  Or do you create your setting with essentially fungible encounters, and it's fine if the players skip whole sections?

You see, there's this conceit that a megadungeon should be too big to map, too big to fully explore.  The players are only expected to experience bits and pieces of it.  Unfortunately, maps with openings and unexplored areas drive players absolutely farking nuts.  Players are completionists by nature.  They need to explore those unknown areas and finish every last room and corridor.  It's like a form of table top OCD.  Player psychology runs directly contrary to the megadungeon ideal of large areas unexplored.

Here's a novel thought:  maybe the megadungeon ideal runs contrary to the psychology of referees, too.

Let's say I've created my precious snowflake, the Coolest Trick Room Ever™ - the Room of 10,000 Spiders.  I've put a lot of time into that 10k Spider Room - I really want to make sure the players experience it.  What are some scenarios for the extended design?  Perhaps my dungeon is so large the players may never experience 10,000 Spiders before they move to the next level (: sad referee :).  I could put the 10k Spiders right in their path, so they're guaranteed to meet it - perhaps I don't mind a bit of the old Illusionism, shuffling around quantum ogres and whatnot.  Another expression might have the levels sized small enough to ensure the players find everything, giving in to their completionist OCD urges and the referee's need to see his or her work experienced and validated.  A different flavor of that approach is to put in hints, clues, and other pointers to help them find the 10k Spiders room, even if it was off their path - the blaring neon sign approach.

Do you write interesting, detailed encounters and place them in a design so large that there is only a slim chance the party will actually meet your Precious Encounter™?  What are you, some kind of masochist?

I submit that there are very few (published) dungeons that are so large in scope the players are expected to experience only a fraction of each level before descending downward.  Execution of the gigantic dungeon involves one of three basic approaches:

  1. The place is gigantic, but it's only detailed at a high level at first, and the referee will elaborate more details ahead of each game session.
  2. The place is gigantic and detailed at a granular level, and the referee spent a gargantuan amount of time writing it.  Oh the humanity.
  3. The place is gigantic, but only certain pieces are detailed at a granular level, and the referee is moving them around as necessary so the players run into the detailed stuff.

Which brings the conversation back to Dwimmermount.  Dwimmermount seemed to be one of those Type 1 settings - vast in scope, but lightly detailed; and now it's en route to becoming a Type 2 setting - vast in scope but detailed like a much smaller module.  Dwimmermount is somewhere in the neighborhood of 800 rooms; the original five levels of Stonehell is also in the neighborhood of 800 rooms.  Barrowmaze is somewhere under 200 rooms (the second installment brings it up near 400).  I have one of the early versions of Castle of the Mad Archmage, and it weighs in at almost 1,000 rooms - the clear heavy weight of the published megadungeons.

There are plenty of good reasons to run smaller lairs and dungeons.  They become self evident as you read the notes above - you never have to worry about the players missing a Precious Snowflake Encounter™ or your work going to waste, for starters.  For a published module, the experience you're presenting to the reading audience is homogenized - lots of people will have the same reminiscences.

Why would you put together a gigantic, sparsely detailed megadungeon where a fair amount of your work may go unnoticed?  This is what I've been mulling.  It implies that the encounters and areas, which might be interesting and unique, are still representative and fungible.  It’s okay if the players skip areas and head down to a more challenging level.  You've designed it such that the players don't have to complete everything.  (In fact, grinding XP on the easy levels seems anathema to the whole thing - like boiling ant hills or something).

I have some further thoughts on why the megadungeon is so alluring, despite the inherent risks to the referee's time, sanity, marriage, and social life, but I'll save them for the comments or a follow up post.  I didn't mean for this to be so discursive!

Before I forget, here's that funny epiphany I had the other day:

Pick One:  Do I want to spend my time working on the kind of dungeon that I, as referee, will never expect to finish, or work on the kind of dungeon the players will never expect to finish?

*The sandbox triangle is a project management truism adapted to sandbox gaming.  The sandbox game balances freedom (scope), effort (time), and detail (quality).  If you don’t have a lot of time, but you want the players to go anywhere, the sandbox won't have much detail.

If the players tell the referee their plans in advance (ie, we're going to Goblintown next week), the referee can use the allotted time to make Goblintown more detailed.  The referee's prep time in between sessions hasn't necessarily increased, but by self-limiting their scope, the players have allowed the DM to focus on more details in a specific area.


  1. _Do you write interesting, detailed encounters and place them in a design so large that there is only a slim chance the party will actually meet your Precious Encounter™? What are you, some kind of masochist?_

    Well, they're a GM, masochist is somewhat implied. :P

  2. I've got tons of encounters the players have never touched. They got semi-completionist on the first level of ASE, and then gave up on that because they weren't getting enough gold. For the next couple of levels, it's been more directed at accomplishing in-game goals - get some rumored bit of loot, slaughter some faction of humanoids, etc.

  3. A lot of food for thought. One idea that occurs to me as I read this is that a megadungeon may be well (better?) suited for a GM who has multiple gaming groups and is going to run lots of different groups of players through the same place. So even if group A in the online game misses the "My Precious Snowflake Encounter," there is every possibility that group B in the weekly FLGS game may still run across it. Or group C, playing a year later may find it.

    That said, I've shied away from the megadungeon for the reasons you mentioned, and I tend to gravitate toward creating dungeons that can be completely cleared out in a session or two.

  4. I enjoy the creation. I love it when the players find my 10k Spider Room, but I'm a tinker by nature. If they don't find it maybe I'll reuse it someday. Maybe I'll quantum ogre it. But they probably stumbled into my 20k Firefly Room and that's fun too.

    The thing I worry about with a megadungeon isn't individual parts, it's the thought of players never reaching the bottom. Whatever is down there. Any party can explore the upper reaches of a megadungeon, but I want to see a party reach the bottom.

    It's a related concern, but I think different in some key ways. It's tied to wanting to see a party stick with a game long enough to reach that level of development. And it digs into my imagination of wanting to see what it is that the dwarfs woke up when they dug too deeply and too greedily.

  5. The irony about Alex M from Autarch creating a Type 2 setting is that ACKS itself recommends using a fungible design in making the campaign setting. Specifically, that you create lairs and only place them on the map as players encounter them.

  6. I think that when you publish a megadungeon, you are really doing it as a creative process and a work for others. like to think that sometime, someone will use my megadungeon. Some adventurer out there will get to that out-of-the-way room of Castle Triskelion. I think that this is certainly true of the Castle of the Mad Archmage.

  7. Pat's approach is also where I would try to go with a solution if I was running such a vast dungeon. Since the individual areas are less relevant in the aggregate, the game needs to shift from "did the players find the special room", to "are the players making interesting choices about where to go?" How do I maximize the player-driven nature of that type of game? Embrace the strengths of the sandbox. The trade off is that some work will go unused - let's hope you didn't invest too much time; you're hoping to gain a unique table top experience by presenting such an extensive locale and giving the players extreme liberty.

  8. I think the old AD&D 2E boxed set Dragon Mountain would qualify for discussion here. That set had extra maps that weren't even keyed for some of the areas, and the authors encouraged the DM to add, rearrange, and generally cause havoc. Their advice even included changing passages, that the mountain itself might change in layout in the inhabited areas by the next time it was explored. My players actually went off on a tangent some of the unmapped kobold warrens, and I had to do everything on the fly, which I love to do anyhow. But that's just one example of what TSR was doing with the megadungeon in that era. They're meant to be open-ended. And, since the DM controls what the players experience for the most part, there's nothing that says you can't take your 10,000 Spiders and dump it right on their fat heads whenever you want. I like to have a few pre-designed but unplaced encounters that I can drop in when it's appropriate in just that way. Things lagging along, players growing restless from slogging through yet another rats' nest with 2000 cp? Drop the spiders on them in one of the next rooms. Problem solved.

  9. I design my dungeon in a similar fashion to designing a "surface" campaign...I make a general map of what is in each area of the dungeon level (using a grid designed in Publisher). Then, as I do my detailed maps (as needed or as I have time), I drop them (in jpeg form) into the grid. Because we are playing online, I just upload the jpegs to Roll20. I find if I create too much detail at once, I exhaust my small reserve of creativity.

  10. Dungeons are dynamic environments. Each level of a campaign dungeon need not be too large, but they should be deep, and each time a level is cleared out, new critters will take advantage of the open real estate. Planned encounters need not be elaborate. A few random rolls on the appropriate table, a few notes is all that's needed. You can let your imagination feed off the vide you get from the table to flesh out the details for the encounters your players actually er, encounter. No need whatsoever for overly detailed notes.