Wednesday, June 23, 2021

Spectral Interrogatories - Stonehell Dungeon I: Down Night Haunted Halls


Stonehell Dungeon I: Down Night Haunted Halls is one of the first Old School Renaissance Megadungeons, published in 2009 when the idea of playing Dungeons & Dragons in a classic style was shiny, exciting and novel. Written and published by Michael Curtis on Lulu, it is in every way a glimpse into another time, and that is not a bad thing. It might seem odd to inaugurate a new reviews project with a 12 year old adventure that’s been turned over, scrutinized, well loved, and copied, but Stonehell is a special adventure, a standard of what is frequently called the “OSR”, and still widely lauded ... so let’s tear it apart ... of course not. While I am not stinting from criticism, Stonehell is an excellent adventure and well worth study, emulation and play. It’s biggest flaws are the, likely unavoidable, result of its status as a megadungeon, its nostalgic intent, and age. At 134 pages, I won’t be able to do a page by page examination of Stonehell, but I have read the whole thing, and I think I’ve even played a game or two set in its halls back in the halcyon days of G+. As a megadungeon I expect certain features: minimalist keying, a strong appeal to D&D’s vernacular fantasy, large amounts of empty space, faction intrigue, unexplained mysteries that can tie into a referee created setting, and scale that requires repeated visits throughout a campaign. Some of these are not things I normally like in an adventure, but they are useful, likely necessary, in a project the size of Stonehell, and even more they’re to be expected in a clearly nostalgic work from 2009.

This is about the best art you'll get.

Stonehell is a particular kind of adventure, one that seeks to emulate not just the mechanics of early Dungeons & Dragons, but its implied and explicit setting. Stonehell is an effort to meet a popular design goal in the early days of the OSR, but still something that can succeed or fail on its own terms. Stonehell succeeds, because it avoids most of the pitfalls of nostalgia. It is referential while maintaining a distinct voice and willing to break with its sources to improve its own quality. Within self-imposed limits Stonehell is one of the best, its faction structure is robust, its existence and themes justified by its fiction, and its locations as innovative as can be for a willfully traditional and predictable setting. Still, for a reader that wants to move beyond the often stifling tropes of Gygaxian vernacular fantasy Stonehell is not a great adventure choice, its numerous humanoid tribes are still essentially evil monsters (though only the human berserkers seem truly monstrous), but they are presented in the classic manner as potential allies of convenience who have goals, rivals and plans. Descriptions are terse and Stonehell’s halls take hundreds of keyed locations to depart from the standard maze of gray stone corridors that define stereotypical dungeon adventure. Stonehell’s overall quality, long held status as one of the first and best OSR megadungeons, and layout innovations mean that even if one rejects playing Stonehell because of its aesthetic and thematic choices, it's still incredibly valuable to anyone seeking to design a megadungeon.


Stonehell may not stray far from Gygaxian vernacular fantasy, with some unavoidable problems, but it is also constrained by its megadungeon format. Here, Stonehell does wonderfully, meeting the strict limits of scope and space with design decisions that still feel new and unexpected even after a decade of emulation.

First, Stonehell manages usability well, it’s laid out primarily as dungeon nodes (four quadrants for each of the five levels in Stonehell book one) of twenty five to forty keyed locations each. Each level has a map and page or two of basic overview, and each quadrant the same, followed by wandering monsters, followed by stat lines, and general information. These introduction pages also include notable NPCs, factions, and more complex puzzles for each quadrant. The keys themselves are minimal, each node's fitting on a page or two, with individual keys only a handful of lines. This style has some negative consequences, and would likely be impossible to utilize with a more unique aesthetic, but it is extremely functional for a megadungeon and allow Stonehell's grand scale.

Second, Stonehell also succeeds at building a location which feels like it has a long and deep history that’s visible to even its minimal descriptions. It is not the backstory of referee facing narratives, but one of small details built into the room keys and nodes of the dungeon. The idea of a living active space, an ecology and history beyond the characters is obvious to players and referee alike.  Stonehell is a location well lived in, where many previous adventurers and various groups of residents have left behind clues and treasures, met terrible fates, or occasionally prospered. More ancient layers of history are evident in architectural variation between the nodes, and generally add variety to the obstacles and inhabitants in specific areas of the dungeon. These are not as naturalistic or coherent as the adventures smaller signs of use such as graffiti and dead adventurers, but there's enough attention to Stonehell's ecology and history to provide a gloss of believability so that Stonehell never totally reaches a funhouse level of randomness yet manages significant variety with increasing strangeness.

Third, and Stonehell’s greatest accomplishment, it builds a wonderful structure for faction intrigue, almost every intelligent monster within has enough personality for a referee to easily roleplay and almost every group or powerful individual has plans, goals, enemies and relationships to the other factions. Evil wizards, medusa information brokers, goblin refugees, mushroom dryads, a paranoid dragon, and ogre bachelors all are entities that adventurers can interact with and whose goals they can thwart or assist. Most dungeon nodes are balanced around an ongoing conflict or conflicts that define the more powerful groups of creatures within an environment, in addition to a set of vermin or lesser encounters. It’s a very functional system, and one that’s absolutely core to successful megadungeon design, or really any form of classic dungeon adventure. One of the ways Stonehell does so well with its factions is that it makes its denizens’ goals largely comprehensible, sometimes sympathetic but rarely clearly aligned with the players. The players may find allies easily enough, but it’s likely to be a tenuous and temporary relationship that generates more intrigue and conflict rather then simply making one area of the dungeon safe.


The first quarter of Stonehell's first level...
and likely the weakest

Stonehell is an excellent adventure, but it is also 12 years old, and it has its share of minor issues. The first, and perhaps largest issue (because it's nearly impossible to fix) is found in Stonehell’s maps. Stonehell’s individual nodal maps are often solid, and increasingly good on the lower levels, with multiple looping paths and both scattered room placement and predictable symmetrical spaces when they make fictional sense. The maps tend towards individual rooms off branching corridors, but this isn’t always the case and there’s considerable variety, especially deep in the dungeon. The way the nodal maps connect to each other to form levels however is not as well considered. The nodes are very distinct and transitions would be abrupt under any circumstances, but most have only a single access point to their neighbors, and this is repeated. Moreover the maps themselves, while they vary internally to some degree, have very consistent external dimensions and all five levels are arranged in squares, with each quadrant a node. This is a small complaint, but a megadungeon’s nodes shouldn’t just have different content, they should be different sizes, with different amounts of keys within and different spatial arrangements across levels. The vertical aspect of Stonehell is also mixed. Verticality and vertical connections between the levels of a megadungeon are key elements of functionality because parties either need to leave the dungeon between sessions or retreat to a haven within. Numerous and varied ways to access various parts of the massive structure allow for play to advance beyond an initial level and entrance. Spatial navigation between levels in Stonehell seems fairly decent, there are elevators, chutes and other ways to travel between them unexpectedly and quickly. However, Stonehell has few entrances and it would benefit from more.

Ways to access each of its levels, though maybe not immediately, would dramatically improve the ease of running Stonehell both as a long campaign (one that explored the lower levels), keep with its long strange history, and help mitigate the predictable level layout as parties might enter at various points. This problem compounds when one considers how Stonehell (and classic adventures generally) are designed to be run vs. the actual conditions of most contemporary play.  Most people won't be able to send a party of 6-10 characters into Stonehell week after week for half a day at a time.  Instead contemporary play encourages two to four hour expeditions with far smaller groups and shorter campaigns then the older Gygaxian ideal of weekends given over to Dungeons & Dragons.  With this in mind, access becomes far more important then they might first appear.

Other small issues are the somewhat unexciting descriptions of treasure, which is still far better then the majority of D&D products past or present, the sometimes odd placement of puzzles and traps, as well as a lack of useful detail in their descriptions, and the way the format sometimes requires a messy (but mercifully short) flipping back and forth between the pages. This last organizational issue, and Stonehell’s repeated style of minimalist design, creates an additional small concern and compounds a second.

Detail often lapses where it would be especially useful, a wasted opportunity. As mentioned, many traps simply inflict damage or effects, without a mechanism described that players can disrupt. Likewise, in multiple places the GM is instructed to provide “2 random cleric spells” or something similar. This is never a great idea because leaving out the name of three scrolls or a couple of descriptive adjectives won’t save a lot of space, but it does create a sometimes jarring need for unexpected improvisation. Furthermore placing scrolls and other similar tools for players can also offer solutions to specific puzzles or traps, and description always allows the designer to reinforce the themes and aesthetics of a dungeon. The second issue related to layout is the waste of space by a sometime insistence that magical effects in the dungeon exist as new spells in the D&D format, and that some dangers or strange sites must be justified by a magical item used in an unexpected way. A healing mud bath works because a lost staff of healing is buried beneath it (and very very difficult to extract), and a bridge trap includes a long description of how a decanter of endless water is mounted so hobgoblins can use it as a water cannon. This is an annoying Gygaxian D&Dism, a sort of magical positivism, where magical items and spells exist but only in a canonical form that is then rigged into something else. It’s unnecessary, wastes space, and Stonehell clearly knows better
showing multiple uses of unbounded magical effects and traps. Again, this clumsy reliance on repurposed ‘official’ magic items is minor, but it becomes a layout issue which makes it noticeable in a work that is otherwise incredibly successful at layout. Wasted space here could have also been used to make puzzles and traps more mechanically detailed, allowing for greater approachability, and a variety of disarming methods.

Finally, and with no charity or fairness, Stonehell looks antiquated. Its layout, despite utility, is cluttered, paragraphs dense and conversational, art nearly nonexistent, and maps extremely plain. This isn’t really a critique, as much as an acknowledgement of how far the aesthetics of amateur dungeon design has come in the past years. While copious, professional, purpose-drawn, and consistent art could dramatically improve the sometimes blandness of Stone Hell’s description, and a greater application of side bars, bullet points and other usability tricks would make it more accessible, this sort of change can’t reasonably be expected in a project from 2009. The amateur RPG publishing space simply lacked the design talent and practice that it has today when Stonehell was published, and that it holds up as well as it does is an example of how well thought out its usability elements really are.  It is also a warning to designers who might be overly proud of their aesthetic, because the good fundamentals of design and usability in Stonehell will take one a lot further then mere artistic sense. Still one hopes that Stonehell will get an upgrade some day with a more contemporary approach to layout and far more art.


The intent of this review series is more then just looking at products that interest me or cross my desk, it's to interrogate them for what I can learn and pass on about adventure writing, and part of the reason that I chose to begin with Stonehell is because it spotlights two issues very close to my own interests: vernacular or vanilla fantasy and megadungeon design.


Stonehell is as good as one can expect from a nostalgia constrained adventure, and it’s not quite fair to attack a product because it succeeds at its creator’s goals.  Yet all adventures written as Gygaxian vernacular fantasy can suffer in play, and a referee planning to use Stonehell should first decide if they see these elements as negatives, positives, or unimportant.

By Gygaxian
vernacular fantasy I mean a sort of setting built from Tolkien and Conan, sieved through the particular wargaming obsessions of Gary Gygax and early TSR. Fantasy rationalized by a sort of semi-historical taxonomizing obsession where Osprey Publishing’s books of medieval armor are the controlling authority for Conan’s fighting abilities. There’s nothing inherently wrong with this genre, though it can hold some disconcerting reflections of mid-century America identity, and it served a purpose in 1975—allowing wargamers to appreciate roleplaying games and creating a sort of referential “reality” to buttress the scattered rules of 1970’s Dungeons & Dragons.

However, in 2021 Gygaxian vernacular fantasy is a collection of cliches and that presents difficulties. Vernacular fantasy allows “system mastery” by players. Most notably players can memorize the stat-lines of monsters and so drain all wonder from encounters. This has only worsened with time, as “goblin” and “giant rat” become synonymous with the tutorial level of fantasy video games. Players well versed in popular fantasy culture don’t even need to have memorized a monster manual, but still cannot engage with the creatures of Gygaxian fantasy except by reference, and those references are often wrong. Popular conceptions of fantasy worlds, themselves formed by Dungeons & Dragons, have evolved and regularized so that while goblins are dangerous in classic D&D, the modern cliche they embody is not, and it’s so powerful that players can rarely envision anything else. This not only removes a sense of wonder and discovery from the game, but it can create expectations wildly in conflict with the mechanics. For some players this creates frustration and a sense of inhibition or enervation, where one’s character is incapable of triumphing over creatures of the most base sort. Yet vernacular fantasy remains popular among some, perhaps precisely because it’s predictability gives a sense of mastery, which can also create its own problems. The disjunction between Gygaxian and modern cliche may feed into misunderstandings among other styles of play that classic play is exceptionally lethal, itself inflaming gatekeeping behavior and posturing among frustrated fans of Gygaxian vernacular fantasy. 

Beyond these social concerns, the predictability of both sorts of vernacular fantasy can lead to boredom. Knowing the reputation and nature of the majority of creatures, items, and magics encountered tends to create a sense of general blandness that is especially evident at the start of a campaign. It may not be as serious an issue with the long campaigns envisioned by 1970’s and 1980’s Dungeons & Dragons designers, but contemporary play among adults tends towards shorter sessions and shorter campaigns. In a long campaign the way the characters and the story of their adventures (as well as the setting itself) will evolve complexity, strangeness and uniqueness is a well known aspect of classic play, or at least one that’s often lauded as “emergent narrative”. While this idea is intrinsic to most types of roleplaying games, and worth its own discussion, here it suffices to point out that shorter and fewer sessions of play will give less time for narrative to emerge. When coupled with a bland starting point (and Stonehell's early levels are exceptionally bland), there may never be a chance for the characters to evolve beyond blank slates, or for the intrigues of the setting to weave complex faction relationships around the party. There will never be a chance to do interesting and meaningful stories that transform the setting, discover the Infernal Machine of Lum the Mad, or even encounter creatures more exciting than giant rats and kobolds, because the campaign will only last seven sessions where one character reaches level 3.


Stonehell is one of the larger published megadungeons, and a very successful example of the form. It’s inspiring to read, making the process of megadungeon design look easy … but megadungeon design for publication isn’t easy. Like most adventure design, the gulf between the notes a referee needs to run their own creations and what one needs to include to make a published work playable is huge for a megadungeon, but unlike smaller adventures, megadungeons have to be extra cautious about key size and usability.

There is simply no way to include a significant level of granular detail or complexity in each key of a megadungeon (though many try) and still expect a reader to use it without enormous amounts of adaptation and preparation. Stonehell has an estimated 600 - 700 keyed locations in its first book (and the first 5 of 10 levels), leaving ⅕ of a page for each location if the adventure only contained keys. Yet a megadungeon, more then most adventures, need a foundation for those keys, a way that a reader can look at the levels and see the spatial, social, and narrative connections between them and within each node. Neither keys nor this relational information can be especially dense or complex either, as the volume of information that the reader is expected to digest and hold in their head to run the megadungeon is already huge.

These limitations mean that to succeed a good megadungeon needs (or I suspect it needs) to be very careful in its approach to usability and rely on some degree of minimalism. Generally keyed areas will include no more then a single interactive element, and these will need to be described with a minimum of detail. Like One Page Dungeons (and Stonehell is often described as using a one page dungeon format - it doesn’t, but it shares some elements), megadungeons must be extremely concerned about the use of space, squeezing description into as few words as possible and including elements of design that work harmoniously to produce a clear overall feeling, tone and structure. Because of this requirement of brevity, and the need for greater usability inherent in megadungeon design, it seems difficult to include significant amounts of creative setting building within a megadungeon, it must primarily be useful in play, and largely self referential, because the megadungeon’s scope, both as a physical project and as an in game location is so great. The megadungeon more than most other types of adventures needs to be placed within an already existing setting, but the amount of work involved means that its author, especially an amateur author, will rarely have time or energy to produce such supporting materials. Thus, the need to save space and also remain interesting (usually accomplished by unique imaginative elements) becomes acute in a published megadungeon. In Stonehell’s case (and many other megadungeons) it’s reliance on Gygaxian vernacular fantasy helps save space and alleviate this problem, making room for a fully keyed location and description where one would otherwise need to add the necessary descriptions of monsters and their behavior.

I think it’s an open question to what degree this reliance on cliche or predesigned setting is a useful or even essential element of megadungeon design. Certainly there are other classically inspired megadungeons, and some push more aggressively beyond the vernacular fantasy of Stonehell. How and how well this works though and if they can accomplish it while producing a fully keyed, adequately complex adventure is an interesting quandary worth examining in future reviews.


  1. Excellent, in-depth, thoughtful review. Also very cool to see material from the earlier period of the OSR reconsidered.

    1. Thanks, I think Stonehell still stands up as a very useful project for anyone attempting to design a large dungeon - its factions, layout and the way it manages a truly staggering scope have lessons worth repeating even 12 years and thousands of adventures after Stonehell's publication.

      I intend to look at some more contemporary methods of abstracting megadungeons in the future, and Stonehell is a far better example of the classic form then most megadungeons.

    2. I’d also recommend your readers look at Michael Curtis’ blog (the writer of Stonehell) from that era too. Tons of great posts about the development of Stonehell, and thoughts about RPG design in general.

  2. I think the Vanilla of the firsts levels is the point.

    1. It almost certainly is the point, as repeatedly mentioned in the review, Stonehell is a nostalgic project attempting to create an adventure similar to those of the early years of D&D. The issue with the first level is not that it's Gygaxian vernacular fantasy, but that it has too few entrances, sub-optimal linkages between the nodes, and that even within its genre it is uninspiring and relatively uninteresting. Given that it will take several sessions to get through this means it will likely discourage even players who like the genre.

    2. To support Gus' point here, I played in Stonehell in 2010 (I think), run by Scott Driver (of the now defunct Big Ruined Pile blog, before that the Wilderlands of Darkling Sorcery). I played a number of sessions in the first level, and my impression was that it was very boring fare with a long funnel of an entrance, all grey corridors and orcs. So I never read it or looked at it again. Only recently I learned that the lower levels were more interesting from a group that is running it now. I thought, "Oh, I guess I got the wrong impression from the early parts of it. Maybe I should look at it after all" Pablo if your point is that you need to earn access to interesting dungeon by penetrating boring early levels, my burning question would be: why? Why not make it interesting also at the beginning? What is stopping a designer from doing that? Gus' answer is that the motivation is in part nostalgia, which I get. But then the point is, can't we combine nostalgia with a more interesting take on the classic D&D tropes?

  3. I think it would be useful to flag which contributor has written a review at the top of the post and/or as a label. Thinking of other kinds of review sites as a comparison, it's nice to get a feel for a critic's tastes over time and I could see it being useful to be able to quickly filter for a specific contributor's work.

  4. I have been running an open table OSE campaign for 20 sessions now, with Stonehell as the theoretical tentpole and a few other dungeons (mostly Necrotic Gnome stuff) scattered around the world.

    My players never want to go to Stonehell if they have another option. I think it's the treasure hauls... they've gotten the (correct) impression that the risk/reward is much higher in Stonehell than in most any other adventure I could throw their way. I'm not sure if this is part of "Gygaxian fantasy" or just the author's preference, but that dungeon is *skint*.

    1. Stonehell has quite lower amounts of treasure on its upper levesl, something I find fairly common in "OSR" work of its era. I don't know if it's the treasure placement ideas of later editions, personal choice, or bad luck rolling on the random tables. I'd figure one would want at least 48,000 GP on the outdoor and first levels combined (enough to get a party of 6 PCs to 2nd level for each quadrant). It seems to have about 1/2 that.

      The classic treasure tables are pretty generous at times, but also tend towards feast or famine based on a high degree of randomness, though I'd consider treasure (and/or XP) stocking less an aesthetic issue and more a design one.

    2. I'm prepping Stonehell for an open table using OSE, and the treasure was looking very thin to me, so I made the calculation (thanks to CTRL-F), averaging value where needed. Not counting the random treasure from crypts in 1B (which would probably add a couple hundreds gp to the haul), it adds up to 22,610 gp.

      About half of that is located in 1C; and the vast majority of it is in a single room (either in a chest or on the three orcs keyed to the room.)

      So, not counting the orc's hoard, we're talking around 13k distributed in ~187 keyed locations. That's 6 PCs to 2nd level. Ouch!

      I think that I will multiply most treasure by 10, and I don't think that it will feel like a generous dungeon!

    3. Came back to this review after skimming (sorry, it's very lengthy!) your Dwarrowdeep review. I'm still running Stonehell at my table - we're on session 20 or so and they're still just scratching the surface of the 4th level.

      I did finally break down and do a full treasure rebalance, going through and totting up the available treasure on each quadrant and rebalancing to (in most cases) have enough to allow a party of 4 fighters the level of the dungeon level to advance. Feedback from the group since I did the treasure rebalance has been pretty positive.

      When I eventually finish the treasure rebalance (I'm only at level 4-ish right now) I'll probably put it out publicly. Love that dungeon to bits, but it just needed more treasure to support our playstyle!

  5. As usual, very good and immersive review. I like your interrogations about vernacular fantasy (also the concept vernacular instead of vanilla) and megadungeon design.


A Pinch of Salt: Churn Rate

Author: Ian Yusem Artist: Sajan Rai Reviewer: Dan D. System: Mothership Run as Referee General Disclaimer: I share a couple discord servers ...