Differing Approaches to Dungeon Design: Hand-Crafted vs Procedural-Generation - Part 2
As mentioned in the previous post, there are some limitations to the “hand-crafted” approach to dungeon design. One of the primary limitations of this approach to dungeon creation is the increased level of time and commitment that it demands from the game’s referee.
The referee who ran the game that I’ve been discussing in the previous blogpost mentioned that it took quite a lot of time to create the scenario that we played through, owing to the high level of refinement and detail that he put into the scenario.
The referee also noted that if he was running a weekly game, then he’d need to create a great deal of material for the players to engage with, provided that the players were making progress in the game world at a reasonable pace.
When these two things are considered together, it seemed apparent that a referee who solely relied on a “hand-curated” method of dungeon design would be unable to produce the amount of gameable content required to sustain a weekly game, unless the referee significantly reduced the level of refinement and detail present in the referee’s game world.
So, what’s the alternative? Many old-school referees champion the use of procedurally-generated material, created through the use of resources such as AD&D’s Appendix A, or the “Solo Dungeon Adventures” tables found in the first volume of The Strategic Review. In the 3 LBB of OD&D, we see that there are rules provided for the random placement of monsters and treasure in a pre-constructed dungeon environment, as well as rules to determine which type of monsters are encountered in the course of a wandering monster encounter.
The use of these “procedural-generation” methods allows the referee to prepare a large amount of useable gaming material quickly and easily. Personally, I’m a big advocate for the use of this kind of random content generation. I feel like this this way of creating material helps facilitate spontaneous gameplay that is likely to take a different direction than what the referee or the players may have planned, and so I encourage all referees to embrace the “oracular power of the dice”.
However, relying entirely on procedurally-generated material is not without its pitfalls. Without a referee’s guiding “hand on the tiller”, randomly generated content that is inserted into the game without adjustment can lead to a game world that appears completely incoherent.
I’m certainly no stickler for “Gygaxian Naturalism”, and I enjoy a funhouse megadungeon as much as anyone, but when the entire game world lacks all sense of internal consistency and thematic cohesion, the player’s ability to meaningfully engage with the game is greatly eroded.
This is obviously undesirable, and not just because of the way it disrupts the player’s immersion. Considering this issue on a purely mechanical, tactical level, it is evident that an entirely-random game world completely compromises the “player skill” element that is so heavily emphasised in old-school D&D.
When the fictional reality of the game world is left up to the throw of the dice, the player loses the ability to analyse their character’s surroundings, and to use environmental information to try and predict what obstacles and hazards lie ahead. As a consequence, they player is unable to make meaningful decisions in the course of the game.
When a referee considers these factors together, they may reasonably conclude that it’s appropriate to adopt a balanced approach, that combines the best elements of “hand-curated” and “procedurally-generated” dungeon design. When designing dungeons for my own games, I make use of this “balanced” style of dungeon design.
When creating a dungeon, I typically start by thinking about what the overarching theme or “vibe” of the dungeon is. After that initial consideration, I assemble a curated random encounter table that fits the vibe I’m going for.
Then, drawing inspiration from what I’ve included on the random encounter table, I put together a few pre-created “set-piece” rooms, created with the aim of further reinforcing the theme and aesthetic of the dungeon. Afterwards, I scatter these pre-created set-piece rooms throughout a larger dungeon complex, which I create through the use of procedural-generation.
Something to keep in mind here is that empty rooms are perfectly fine to include in the dungeon, they’re not wasted space. There’s a number of reasons why this is the case, but I won’t go into detail on that matter here, as this topic has already been superbly addressed by the blogposts “On Empty Rooms” by Delta (Delta’s D&D Hotspot) as well as the “Empty Rooms” by James Maliszwewski (Grognardia).
One piece of advice that I’ve found to be valuable is to avoid trying to “wrap things up” too neatly, with everything making perfect logical sense. It’s important to allow yourself be surprised by the random outcome of dice rolls, which can organically generate unexpected connections between disparate elements of your dungeon.
I understand that there can be an urge to ignore procedurally-generated dungeon generation results that “don’t make sense” at first glance, but I’d advise that referees resist this urge. The referee should instead approach these randomly-rolled outcomes with the aim of “refining, not reforming" what was generated.
The referee should look over the map of their dungeon as a whole, and consider how they can apply their own creativity and intelligence in order to expand on the procedurally-generated elements. The referee should seek to identify connections that could be made between the various inhabitants of the dungeon, whether that be organised factions, individual NPC’s, or even unintelligent monsters.
Further analysis and advice regarding old-school dungeon design can be found in the PDF below. As always, I hope that readers find something of value amongst this collection of information, and are able to implement any useful aspects into their own games.
Once again, using procedural generation before a session is a legitimate tool in the referee's toolbox. My contention is that procedural generation should be done *before* the session actually starts and is locked in when the players show up. Once they show up, generating as they go leads to weird outcomes where player's choices ultimately don't matter, since every single do is random. If they are locked in, players can make reasoned choices about whether they should go down the left hallway with the blood streaks or the right one with that weird glowing rock. Since the ends of each hallway are already detemined.