Sunday, January 14, 2024

The Underground Maze or Primordial Stack

Crawling Down From 1974
“Dungeon crawl” has entered the popular lexicon as a description of any sort of adventure in an underground or ruinous space. It’s a common way to describe video games and occasionally other forms of media such as a part of novels or movies. Of course it’s most common in Roleplaying games, because the concept comes from Dungeons & Dragons, specifically from the earliest iterations of the game  - 1974's Dungeons & Dragons boxed set. What is it exactly though?

The term derives from the redefinition of the word “dungeon” by Dungeons & Dragons. Prior to the RPG hobby’s explosive growth in the 1970’s and 1980's the term's popular meaning (itself not the original 14th century meaning) of an underground prison was almost the only one. One can offer theories as to why Gygax & Arneson chose to use the word dungeon in both the title of their game and as a descriptor for the primary arena of play*. Interestingly within the first 1974 edition of Dungeons & Dragons, while the word dungeon is used more often than other descriptors for the place where adventures occur, it is largely in reference to the name “Dungeons & Dragons” or the title “Dungeon Master” for the referee. When Gygax & Arneson are serious about discussing the concept of the fantasy space the characters explore they most often use “Underworld” and sometimes “Labyrinth” or “Maze”. Dungeon is the description that stuck, and the word’s meaning is now far more likely to be the one derived from Dungeons & Dragons.

In the context of role playing games for “Dungeon Crawl” is most helpful as a term if it means something beyond a light aesthetic gloss connotation a particular type of D&D-like fantasy (or an element “Gygaxian vernacular fantasy” aesthetic to be more precise): a place of grim stone corridors, screeching not-men to murder, and the occasional treasure chest that bites. This sort of view of the dungeon and dungeon crawl aren’t a problem, it certainly captures something, but it also tends to create a lot of dispute, because it’s a surface definition. A handful of aesthetic cliches, this idea of the dungeon crawl is however immediately and intuitively easy to grasp -- it becomes the first impression of what a RPG dungeon consists of. It’s not especially helpful though, because it says nothing about how the adventure will work with rules or what sort of play it aims for.

To make the concept meaningful, as always I want to look at dungeon crawling and dungeon design specifically from the perspective of how well an adventure encourages or supports “the procedural exploration of a fantastic space”. This is my definition of the “Dungeon Crawl”. It’s what is often referred to as a “location based” adventure, but I would add the additional qualification that a Dungeon Crawl also emphasizes exploration by connecting it to risk mechanics.

Likewise, this sort of design is sometimes considered the product of the early phases of the RPG hobby, especially of early Dungeons & Dragons. To some extent this is true, Dungeons & Dragons started to define this style of adventure design beginning in the 1974 edition, but I would argue that the game quickly grew away from it, with early D&D communities rapidly pushing the rules towards more character and scene-based scenarios such as wilderness adventure and narrative paths as early as the late 1970’s. The Dungeon Crawl was sidelined for some time, and its development has been fairly slow since, or largely about moving away from the granularity of room by room exploration and risk management towards narrative structure or improving tactical combat. One could even say that every edition of Dungeons & Dragons since the first - starting with Greyhawk, has increasingly focused on character and combat options at the expense of exploration - but that’s an argument for a different time. I like dungeons and the Dungeon Crawl play style though, and so I find it useful to look at how they can be written, what past dungeon designers have managed, and how one can better design dungeon adventures today. 

Patterns of Dungeon Design
When I look at Dungeon Crawl adventures, I see a few patterns of design. these are ways that the author of a dungeon adventure chooses to create a space for exploration: the size and "shape" of the dungeon, where its challenges are, what sort of play will predominate, styles of keying and assumptions about how the adventure will be used that are incorporated into the design itself.  While not exacting there are many elements of dungeon design that repeat both in specific authors works but across entire communities and play styles, creating reoccurring patterns or perhaps standards of adventure design.  

The most common patterns in contemporary dungeon crawl adventures are Philotomy’s “Mythic Underworlds” or variants on the idea — large, relatively minimally keyed adventures that are almost always dependent on referencing rules manuals for setting and detail. Another common design pattern is the “Thracian Ruin”, after the style of Jenelle Jaquays, dungeons with layered history and greater internal detail to facilitate player interaction. Both of these design trends, the dominant forms of Dungeon Crawl, come directly from the same source: the advice and examples in the 1974 “original” edition of Dungeons & Dragons (“OD&D” or the “LBBs”).

Yet the Mythic Underworld and Thracian Ruin are quite different design patterns. They may derive from the same source, but obviously Jaquay’s late 1970's reading has very different influences from Cone’s early 2000's one and this leaves questions…

  • What is the design advice in the 1974 edition?
  • Does any pattern of dungeon design follow directly from the 1974 edition's advice?

Cover Art from Blackmoor - 1975
What OD&D Says...
Dungeon design is of course an essential part of OD&D, or at least the dungeon is described as the primary arena for play. Gygax and Arneson may provide a few tables and basic rules for overland travel, but the first paragraph of “Underworld & Wilderness Adventures” (the third little brown booklet in the 1974 edition of Dungeons and Dragons) is:

Before it is possible to conduct a campaign of adventures in the mazey dungeons, it is necessary for the referee to sit down with pencil in hand and draw these labyrinths on graph paper. Unquestionably this will require a great deal of time and effort and imagination."

Even this introduction raises a few points about dungeon design. Most conventionally for Classic, OSR, and similar play styles the reader is told that the process of dungeon design starts with a map drawn on graph paper. Gygax & Arneson also note that the process and the keying that follow require a “great deal of time and imagination” - obviously this isn’t a controversial idea but it highlights a few essentials, including the strong impulse towards referee control and creation in OD&D. Elaborating on the key points here we see:
  • The dungeon is a mapped space.
  • The dungeon is a measured space.
  • The dungeon is an imaginative space.
The dungeon is mapped, and so requires navigation. The referee doesn’t just set the scene and the dungeon isn’t simply scenery. Instead, with a map the referee has the tools to make how the characters traverse the dungeon into a component of play itself that is fixed and specific. With the map the dungeon can be explored and revisited, and its areas have both individual specificity and a spatial relationship.

Not only is the dungeon mapped, but it exists atop a grid, allowing for even greater specificity through measurement. With measurements spatial relations become time relationships as well, because the dungeon has enough specificity for the referee to consistently document speed. The Dragon’s Lair can be both West of the Great chasm, and require five turns to reach.
  • The dungeon is the location for the campaign.
An often overlooked point here is that the dungeon and campaign are synonymous. A dungeon is the arena for a “campaign of adventures” not simply one part of a larger campaign structure. This idea is one that Philotomy’s Musings and the OSR (at least until it’s later years) gave a great deal of weight, and so championed the mega dungeon. However, the exact nature of the Little Brown Books’ mega dungeon proves somewhat different from that of OSR conventions.
  • The dungeon is to some degree a maze
This is the point where one starts to see the distinction between OD&D and its subsequent revisions or reinterpretations, Underworld & Wilderness Adventures describes dungeons as both “mazey” and a “labyrinth”. It might be easy to dismiss these words as common Gygaxian thesaurus abuse, but what we’ve see of early maps for Castle Greyhawk, the lesser known games that appear immediately after D&D’s release and in the LBB’s example suggest that maze is an appropriate description of the dungeon design contemplated by Gygax & Arneson. To what degree and in what way the OD&D dungeon is a maze is a worthwhile question, especially because mazes have such a poor reputation in dungeon design for being "
large, samey, and confusing environments"**.

After the introduction Underworld & Wilderness Adventure provides the reader with a diagram of a multi-level dungeon and a bit more advice on getting started:

"The dungeons should look something like the example given below, with numerous levels which sprawl in all directions, not necessarily stack neatly above each other in a straight line."

This is followed by more information:

“In beginning a dungeon it is advisable to construct at least three levels at once, noting where stairs, trap doors (and chimneys) and slanting passages come out on lower levels, as well as the mouths of chutes and teleportation terminals. In doing the lowest level of such a set it is also necessary to leave space for the various methods of egress to still lower levels. A good dungeon will have no less than a dozen levels down, with offshoot levels in addition, and new levels under construction so that players will never grow tired of it. There is no real limit to the number of levels, nor is there any restriction on their size (other than the size of graph paper available.“Greyhawk Castle,” for example, has over a dozen levels in succession downwards, more than that number branching from these, and not less than two new levels under construction at any given time. These levels contain such things as a museum from another age, an underground lake, a series of caverns filled with giant fungi, a bowling alley for 20’ high Giants, an arena of evil, crypts, and so on.”

Before looking at the diagram and its implications, consider the advice on design. A dungeon should have many levels (we'll get to this), but again, it exists as the arena of play for the game.
  • The Dungeon is singular - there is a one location based adventure in a given campaign or created by a given referee.
OD&D doesn’t suggest creating multiple small dungeons, but a single, ever-expanding one that contains such variety that the “players will never grow tired of it.” Designing a dungeon is the referee’s single creative outlet and duty for the 1974 edition (overland adventures having been largely handed over to a board game for a map and mechanics, with new random tables for contents). Interestingly the dungeon is also the referee's signature and a marker of their ability that players can compare as the referee runs multiple groups through their unique dungeon.

The dungeon as singular expression of the individual referee appears to have been popular, if not dominant, in early RPG play. Both Arneson and Gygax’s early games followed the practice, as did others in their circles. According to Cal Tech and other Los Angeles sources, there were at least twenty such distinct individual dungeons being played in the local scene by summer 1975. In the fan writing of both of the West Coast and Midwestern early RPG communities it was common to refer to these singular dungeons by the referee's name "Lee's Dungeon" or "Dave's Dungeon".  One potential difference in this area between the Midwestern and California scenes, and something that prefigures the 2010’s embrace of online play by the OSR, was the way West Coast design and culture tended to allow characters to move freely between these dungeons, compared to Blackmoor and Greyhawk’s seemingly more consistent casts of characters. However, despite this potential difference, both mid-1970’s RPG communities held firmly to the idea that a campaign and single multi-leveled dungeon were synonymous. This idea is one of the basis of the “mega dungeon” and one that the OSR took as its own in the 2000’s and 2010’s.

As intriguing as this idea might be, it’s well enough known and fits within the popular ideas of OD&D’s play style that largely informed 2010’s OSR design. However, the most distinctive and interesting aspect of OD&D’s design notes is where it splits from these sorts of general assumptions, and to a degree from what we know of D&D’s development after OD&D publication. This is the text’s emphasis on navigation and “mazey” design -- here OD&D breaks both with general conception of the mega dungeon and the apparent design practices of both Gygax and Arneson’s home games.

Mazes & Monsters
Early Dungeons & Dragons is fond of mazes, which held a strong place in the 1970’s and 1980’s game culture, while dungeon design in the sense of exploration focused location based adventuring,  means providing risk and tension that make navigation decisions impactful. However, I don’t think traditional mazes aren’t a very good fit for dungeon design. Complex, maze book style mazes are fun, largely because they are an almost meditative, individual activity.

Drawing through a maze with your pencil isn’t a game to play with friends, the decisions involved are frequent, but they are simple questions of turning right or left informed by guesses or checking the path ahead. The key to “winning” a maze is visualizing the entire path through it at once, and this is a visual activity -- unraveling the fascinating artistic tangle of lines involved in a traditional paper maze. Later with early CRPGs like Bard’s Tale, mazes are applied successful to the RPG format, but again, these a solitary activities and depend on digital visuals for much of their impact. Again, the maze doesn’t lend itself to RPG play where the referee describes an environment in response to player questions and changes it based on player decisions. Visualizing mazes collectively is extremely difficult because the group doesn’t have a shared reference for the space, and it’s already hard to navigate using only verbal cues. It becomes even more difficult when the space being described is by necessity a web of nearly identical corridors. It’s also boring.

One could look at the frequent suggestions to create mazes and labyrinth in the Little Brown Books as a mistake, but OD&D had been play-tested when it was published, at least to the degree where something as obvious as the way traditional mazes fail as dungeon design would have been obvious. Instead, it’s worth considering that maybe the maze of an OD&D dungeon isn’t the same as a paper maze -- it’s one optimized for the largely verbal description of RPGs. Looking at the second map provided in Underworld & Wilderness Adventures, one can catch a glimpse of what Gygax & Arneson might have intended. This map shows a “sample dungeon level” -- though obviously a somewhat strange one.

1974 - OD&D's Sample Dungeon Level

This map is likely to look “wrong” to a modern designer’s eyes, and the keys are worse. The modern conception of the “Old School” dungeon as endorsed by OD&D is a mega dungeon, or at least a tentpole dungeon (meaning it supports the entire campaign). Again, this conception is often connected to Jason Cone’s Mythic Underworld reading and these days follows the Mythic Underworld design pattern. Such dungeons are many-leveled and sprawling. They are also often created with at least initial random steps (the Mythic Underworld is less insistent on this then early D&D practice appears to have been), meaning that their challenges tend to consist primarily of monster encounters. Puzzles and traps where they exist in the mega dungeon are scattered and because of the large map by necessity relatively self-contained. The OD&D sample dungeon only matches this pattern in that it has many levels otherwise it suggests a very different dungeon.
  • The map is small, 8 keys and a fairly limited network of corridors.
This map appears to fill a sheet of graph paper in both the original 1974 version, and the less sketchy version added to subsequent printings of OD&D. Yet, the actual size is a bit difficult to determine as the map shockingly lacks a grid - though one wonders if this is intentional or a product of OD&D archaic and somewhat amateurish printing and layout design. Without the grid I’d estimate that the maps corridors run roughly 310’ from East to West, with a winding North South corridor that covers approximately 600’ of distance. In OD&D, with its potentially double movement turns, this means a party with armored characters (at 120’ distance per turn) should be able to cross the map in three turns and meander North or South from the main corridor in four or five turns -- assuming they aren’t distracted by random encounters, traps, or keyed locations.

By dungeon standards this web of corridors is relatively limited, though the branching structure and lack of loops means a party will have to backtrack — the various traps and puzzles aimed at confusing the players mapping will do far more. Even counting these complexities as size, the sample dungeon still contains only eight keyed locations, two set piece monsters, and nothing that might be considered a faction. There is no way to consider this dungeon level expansive.

Corridor length and density matter, because corridors act to increase the number of random encounters in proportion to their length and number. Likewise, more turns spent in corridors mean that light sources are more likely to be exhausted. For the sample dungeon these baked in risks are especially important as it has only two set encounters, and its traps are rarely injurious. The sample dungeon is dependent on random encounters and supply depletion to create tension and threaten character survival.

Interestingly, when we compare the sample map to the fragmentary mapping information we have from Gygax’s own Castle Gygax (see map below), the Greyhawk map fully fills the page, largely with maze-like corridors and many repeated key numbers. Gygax’s Blackmoor maps (even the ones we see vary somewhat) are similar to those of other early and proto D&D dungeons such as Blackmoor (per First Fantasy Campaign), Tonisborg (by Svenson)and the West Coast dungeons of 1974 and 1975 appear to share this style of dense, page size maps. Within Robert Hollander’s June 1975 installment of his referee advice “THOU ART GOD” in Alarums & Excursions he provides details for dungeon design. Hollander’s first step to adventure design and game prep consists of mapping out OD&D’s suggested three levels and using “at least one sheet of graph paper per level”.

A redrawing of Castle Greyhawk's original first level
by ODD74 user Sebastain
This dichotomy between mid-1970’s dungeon design and the sample dungeon is curious. I can’t be sure it’s intentional, but given the still limited number and minimal writing style for keyed areas on the Greyhawk and other early dungeon maps there is no reason that the sample dungeon couldn’t have been similarly dense.

Perhaps it is meant to offer only a few ideas and inspiration, but one would expect the authors to note that it’s incomplete. The authorial intent is of course unimportant here because it doesn’t impact the ideas offered by the text — either to us or to 1970’s readers outside of the Midwestern war games community who had only the text to teach RPGs.

  • The keyed locations on the map are largely traps or tricks (Puzzle Challenges).
All eight keys on the OD&D sample map represent tricks or traps, though some include monsters (ogres and a basilisk) -- again anomalous from early D&D dungeons and very different from more modern mega dungeon design. However, the prevalence of traps is in keeping with a general principle of dungeon design that is also noted by Hollander: “Specials” such as traps, puzzles, and set piece encounters require greater design effort and explicit keying.

This in turn is indicative of an entire structure of dungeon design that depends on random generation and minimalist keying to fill the majority of the dungeon level. While less absolute about this technique then contemporaneous fan commentary (such as “THOU ART GOD”), extremely early adventures like Temple of the Frog and Palace of the Vampire Queen, or some of the more "revivalist" or reactionary parts of the OSR and Post-OSR, Underworld & Wilderness Adventures follows the sample dungeon with a set of advice for randomly filling in the bulk of one’s levels that includes this advice on the technique:

[Keying] can become burdensome when faced with several levels to do at one time. It is a good idea to thoughtfully place several of the most important treasures, with or without monstrous guardians, and then switch to a random determination for the balance of the level.

The actual specifics of OD&D’s procedural generation are fairly in line with subsequent methods, though lighter -- a 1d6 roll creating 30% empty rooms and others with monsters, treasure caches or some combination of the two. Notably the treasure caches here do not use the treasure tables and types, but a distinct table that produces smaller hoards with the automatic presence of silver and a total amount based only on dungeon level. Monster placement is more intriguing being complex and directly connected to the way OD&D balances challenge. The tables using both monster and dungeon levels to create varied opposition offer a wonderful system that highlights how a small palette of monsters and extremely simple combat mechanics combine to create broadly applicable monster classifications and archetypes.

OD&D Encounter Chart
from Alarums & Excusrsion '75
(It's a better chart)

That OD&D endorses the fixed and fully keyed placement of puzzles and tricks while suggesting the use of random generation to simplify treasure and monster placement suggests the importance of these sort of traps and puzzle challenges to the dungeon design pattern. Likewise, the amount of space devoted to trick and trap explanation in Underworld & Wilderness Adventures is considerable - they are presented by Gygax & Areneson as the centerpiece of the dungeon crawl experience. Puzzles and traps are “one of the most stimulating parts of the game” and the referee is encouraged to “include as many mystifying and dangerous areas as is consistent with a reasonable chance for survival.” From this I'm willing to conclude that the OD&D dungeon was intended when it was written to be foremost a “Puzzle Dungeon”. While monsters are a risk to characters, navigating the environment itself and the traps that make this difficult are a coequal or the dominant locus of play.
  • The Puzzle Challenges offered are designed to disorient players and confound mappers.
While traps, tricks, and puzzles may be OD&D’s primary concern for dungeon designers, the specific sort of traps and tricks that Underworld & Wilderness Adventures offers is even more unexpected to the contemporary dungeon designer. A “a shallow pit filled with poisoned spikes” is offered (as are a variety of other pits), and the potential deadliness of traps is mentioned more than once, but the sample dungeon and the advice that follows it are far more interested in non-lethal but disorienting traps and trick.

Rotating rooms, stinking rooms, shifting walls, sloping passages, "door mazes", teleporters, false doors, one-way doors and infinite corridors -- these are the traps and tricks offered by OD&D. They aren’t intended to kill characters or even reduce hit points, but to confound mapping the dungeon and exploring it quickly.

Circuitous Paths
While the top down map of the OD&D sample dungeon looks odd to modern eyes, the first map provided in Underworld & Wilderness Adventures, the side view of the OD&D dungeon is far more familiar and a tradition that has been followed in almost every edition since.  The OD&D map is simple and lacks the evocative touches of some later versions, but it offers the same: irregularly stacked levels and sublevels with a variety of connections between them.  It may seem odd to make this the introductory image of the dungeon, but for OD&D, and other early editions it is not -- the classic dungeon, from the OD&D sample dungeon to Caverns of Thracia or the OSR Mythic Underworld, is a collection of connected levels that perform several key functions. First and most important dungeon level is a marker for difficulty, Underworld & Wilderness Adventures is rigorous in apply this rule, but later classic editions such as Holmes and Moldvay Basic usually follow suit. Encounter tables (which are also often used as stocking tables) are based by level with the rough rule that monster HD correlates with dungeon level.  Of course OD&D's stocking tables break with this rule more aggressively then later editions, but the basic structure remains, at least for the first few dungeon levels.
  • Dungeon level is both a broad marker of difficulty and a way of regulating character access to special areas of the dungeon.
A sample Dungeon Side View/Elevation Map
Moldvay Basic 1981
Secondly, and perhaps more importantly and certainly more interestingly, numerous dungeon level complicate and enliven exploration. Underworld & Wilderness Adventures puts it like this:

“In laying out your dungeons keep in mind that downward (and upward) mobility is desirable, for players will not find a game enjoyable which confines them too much. On the other hand unusual areas and rich treasures should be relatively difficult to locate, and access must be limited. The layout of a level will affect the route most often followed by players. Observation of the most frequently used passages and explored rooms will guide the referee in preparation of successive levels, which, of course, should be progressively more dangerous and difficult.”

Vertical movement between levels is a method for controlling access to special areas and treasure, increasing or decreasing risk and increasing the scope of the "maze".  Underworld & Wilderness Adventures recommends both intentional methods such as stairs and obvious ramps and of course tricks that move the party up and down without their knowledge: chutes, teleporters, pits and of course the subtle sloping passage.  It's notable perhaps that the Dwarf class in OD&D (to the degree it exists) is not hardier or better in combat then human characters, but does gain the ability to notice sloping passages, new construction and other architectural details. It should be clear that these abilities are important enough to make them the signature of one of the more iconic non-human fantasy heroes and that the OD&D dungeon is meant to be a place where spotting a subtle downward sloping floor is a victory.

For the sample dungeon a larger number of levers also encourages the creation of looping and alternate paths. Looping is of course an overused term (almost another empty OSR maxim) taken from the OSR interest in mega dungeons and the Mythic Underworld, while secret passages are a dungeon design standard that was used extensively in 1970's dungeon design but fallen out of favor since. OD&D approaches these ideas in with a clear insight and when Gygax & Arneson emphasize that "[t]he layout of a level will affect the route most often followed by players" or how "[o]bservation of the most frequently used passages and explored rooms will guide the referee in preparation of successive levels" they emphasize of navigation.
  • Smaller levels require vertical connection to avoid constraining player navigation options.
Gygax & Arneson are call on referees to design their own dungeons with careful attention to how characters will move through them, and specifically how to avoid a design that "confines" characters while also makes "unusual areas and rich treasures [...] relatively difficult to locate." This isn't the tiresome demand to put a loop into every corridor and create the all too common contemporary dungeon where every room connects to almost every other room - rather it's about designing the dungeon so that player navigation choices matter. As the previously mentioned hyper-connected dungeons show, this becomes difficult on smaller dungeons.  When one has a hundred rooms per level it's easy to avoid both linear design and to keep some areas separate, but with small 10 - 20 room spaces this can be difficult.  The OD&D sample dungeon provides a solid answer to the issue, add interconnectivity to the vertical as well as the horizontal.  A dungeon map might look fairly linear: a main trunk with a few branches, but it these branches are connected by a path on the level above or below, they can be as connected by as any map that loops from above.  Stairs, chutes and pits also have the advantages of being easily connected to multiple levels, providing limited access to special "sub-level" areas, and offering the players a clear liminal space -- a break between two regions or sections of the adventure. With the limited level size of the OD&D dungeon, vertical circuits offer the chance to make the space less constrained, but still broken into discrete regions. 
  • Verticality allows for limited size levels making it easier to include entrances and exits that go directly to specific areas.
Secret entrances and shortcuts are all important for making exploration play work, and verticality offers a greater opportunity for secrets. Secret doors in the sides of pits, hidden trapdoors, and player utilization of many of the same puzzles and traps the sample dungeon presents are all available for the players to discover, unpuzzle and utilize to find faster, safer and more optimal paths through the dungeon -- on a smaller level footprint.

For episodic online play secret of discoverable entrances leading deeper into the dungeon are also a useful design tool.  Without them characters will move long distances from the entrance as they advance through the dungeon and so be forced to end the session inside.  "Camping" in dungeons brings a number of technical problems around justifying the presence or absence of party members from one session to the next and determining character recovery.  Dungeons with a variety of entrances to a variety of areas within provide a solution to these issues.

Adding these entrances and secret shortcuts can be difficult as a designer, and smaller levels make it easier because it's always easier to add an exit on the outer edge of a map then to incorporate it into the center where it will interfere with the levels above and below.  This makes accessing areas in the center of larger level maps more difficult, because the designer either needs to start designing with central access points in mind and work it into the other levels, somehow account for the long trek through the level eating up session time, or depend on portals or teleportation. Staggering level position can also help, but smaller levels also make this easier.  Finally, given the useful admonishment in OD&D to start with three levels and expand your dungeon as you play, the problem of access to deeper levels becomes more serious because new entrances are harder to push through the upper levels.

The Primordial Stack
Ultimately the OD&D sample dungeon and advice in Underworld & Wilderness Adventures imply a novel dungeon design pattern, distinct from both the narratively coherence, textually dense Thracian Ruin and the sprawling minimalism of the OSR's Mythic Underworld. OD&D may or may not succeed in providing its promised "mazey dungeon", but the combination suggested within Underworld & Wilderness Adventures is distinct, perhaps an extinct common ancestor of more common design patterns. The OD&D dungeon doesn't meet the hundreds of rooms per level ideal of the OSR mega dungeon, or even the sort of tightly mapped web of corridor that appears to have made up early referee signature dungeons like Greyhawk both of which have labyrinthian element ... it uses similar techniques to create a tentpole dungeon. I think of this dungeon design pattern as the "Primordial Stack" or "OD&D Tower".

I suspect the Primordial Stack is a functional design pattern because its vertical linkages and disorienting puzzles slow initial exploration, making it tense even with a small number of keys and limited network of corridors. Later expeditions can proceed more quickly to interesting sites, having untrapped or found efficient (and often secret) paths to new areas. The difficulty of escape, treasure removal and returning to the dungeon each session aren't eliminated, but they are dramatically reduced - and by player ingenuity. I find this opportunity compelling -- a potential victory condition for exploration where new paths of entry and escape make subsequent adventures easier. Exploration play has usually lacked such a metric, and this is one of the reasons it's harder to encourage than combat or roleplay, I believe that with a properly designed Primordial Stack it might become much easier. 

I can't say that this is the intended effect of OD&D's sample dungeon.  It may simply be an impression the text creates against the intention of the authors ... the sample dungeon is potentially one of the many voids or contradictions in the 1974 rules that readers and players were asked to fill or reconcile without readily available guidance. The Primordial Stack also suffers from other potential issues. Puzzle Dungeons in general have a bad reputation as many players find them frustrating, worse puzzles can be hard to contextualize -- even with the core strangeness of a hostile mythic underworld or the excuse that "a wizard did it" it's hard to explain dozens of rotating rooms, sliding walls, and teleportation devices with enough coherence for many players. 

Yet the Primordial Stack is still promising, even or perhaps especially in 2024. The design pattern consists of: a relatively small number of keyed areas, constrained level size, lots of verticality, many levels (with opportunities for exits entrances and shortcuts), and puzzle style obstacles that don't prevent exploration but radically increase the danger of an area until they are unpuzzled. At first glance this makes the Primordial stack ideal for shorter sessions, because of small level size and vertical connectivity the dungeon is easier to enter and exit. Exploration and navigation challenges also offer additional interim goals. Players will both be able to access new content and accomplish something each session as they chip away at the puzzles that make at the dungeon's "maze". 

The difficulties of creating an the Primordial stack, or the general problems of "Puzzle Dungeons" are not insurmountable, and are similar to those of "hack n' slash" adventure design. While some contemporary traditional games have resolved the issue of combat focused play by making adventures a series of fairly quick or simple branching choices that all lead to to elaborate combat set-pieces using more complex tactical rules, older style design has also found a solution. Older style dungeon adventures prevent the combat grind by offering variety (even the OD&D design standard of having at least 1/3 empty rooms does this), and more importantly by offering faction and roleplaying interactions that can avoid constant combat even when encounters are frequent. With similar solutions I suspect the irksome aspects of the Puzzle Dungeon within the Primordial stack can be tamed.  Starting with variety of course, but more by relying on some of the techniques Jenelle Jaquays famously used in her adventures: historical justifications for the peculiarities of the dungeon, greater interactivity through detailed description, and faction intrigue are all likely to mitigate both the frustration and incoherence of frequent puzzle challenges.

A step further is to compare the function I've assigned to the OD&D sample dungeon's puzzles and traps and the actual challenges presented.  This goal is (again) increasing the tension of exploring a "maze" while using smaller levels with a high degree of accessibility and connectivity so that play over multiple expeditions isn't bogged down too much retracing of steps and isn't forced to stop mid-crawl by players schedules. All that's necessary for this is a dungeon with many entrances and connections between levels and key area (such as faction bases, puzzle guarded treasure vaults and set piece lair encounters) - connections that require player effort to find or make safe after a period of exploration long enough to deplete supplies and risk destruction by random encounter.  This doesn't necessarily mean utilizing the confusing traps and tricks that Gygax and Arneson offer in Underworld & Wilderness Adventures however.  Plenty of options exist that don't seem too bizarre or magical for even mundane dungeons or ruins: earthquake shifting rubble, tidally flooded areas, portcullises, deadly but predictable patrols, or floodgates all serve similar design goals to a rotating room or set of shifting walls, in that the block or open pathways either when triggered or on a schedule the players can unravel. One need not entirely reject the excitement of spinning rooms or even teleportation and non-Euclidean geometry as long as there aren't common and there's a functional, discoverable layer or two of history in the dungeon that gives the spaces meaning and a rationale (not necessarily a rational one - wizards still "do things".

Draft Map 1st level of the testbed ... Mt. Sainte Bec

To me these are exciting possibilities - a potential mega dungeon design pattern that may work with modern online play, will be smaller than the Mythic Underworld, yet less dense then the Thracian Ruin while having some of the ease of creation of the first and the coherence of the second. Best of all, it's a dungeon design pattern that is focused on exploration! A way of making the Dungeon Crawl in its purest form more accessible in 2024. It's an experiment I will have to try, I know from dungeon 23 that I burn out after 96 rooms ... but ... and while writing this post I mapped and keyed two stack levels for the ruinous monastery of the infamous Beaked God... Mont St. Bec. Once I manage to get another level and maybe a rotating room (to honor Gygax & Arneson's vision) I'll play test it and see if it works as well as I suspect it may! 

The cover art for TSR's Ruins of Undermountain - 1991 by Brom.
Ruined by me and the desire for a similar color palette.

*Anecdotally the first “dungeon crawl” played in the Midwestern war games community was a raid on an enemy castle, entered via a secret tunnel into the dungeons (a prison beneath it). Infiltrating the enemy fortress from below is a well worn idea, but somehow RPGs became stuck on these lower levels. I personally suspect that the prevalence of the term “dungeon” is the result of Gygax & Arneson’s desire for another word starting with “D” to use in the title of their game. Part of me also wonders if the term “dungeon master” in the BDSM community predates or borrows from the RPG term, but that seems like work for some eager grad sociology student.

**The link here is especially worthwhile -- Paper & Pencils' set of rules for filling undefined labyrinthian spaces -- perfect for the spaces between interesting nodes of mega dungeons. 


  1. I think what is pretty interesting about the sample dungeon above is how much it resembled Level 2 of T1's Ruined Moathouse and it too has 3 entrances from Level 1 which is a lot for its size.

    1. Your point on size is a really well taken and there is an interesting similarity of map design.

      I note the moathouse doesn't depend on navigational puzzles much (there's a few secret doors) but is pretty heavy on monsters and has a number of deadly traps instead.

  2. How has your efforts with this panned out? I would really love to see a worked example just so that I could grok it fully.


Old Games

Let’s talk about old tabletop roleplaying games - specifically the kind of games played in the 1980’s and recently depicted in the nostalgia...