Saturday, January 25, 2020

Descent into Avernus - Dungeon Keying

In the last post All Dead Generations looked at the general design principles in Wizard’s of the Coast’s new campaign tome Baldur’s Gate: Descent into Avernus, and compared them to the classic style of open world, location based dungeon crawls. Descent into Avernus is not a classic adventure, it is not meant to be played as an open world and even its locations which have some of the trappings of dungeons, or are named dungeons, aren’t in any mechanical sense. Rather the majority of Descent’s dungeons appear to either be small lairs, arenas to facilitate a specific encounter, or a series of linear scenes sometimes laid atop a map but largely unconnected during play.

The Dungeon of the Dead Three is the last of these: a selection of encounters partitioned off from the location largely as a means of introducing or ending them each with filmic or novelistic flair - to create a predesigned “moment” of gameplay. This is obviously a very different play style then the classic dungeon crawl, and it seeks to produce predictable narrative moments at every opportunity - willingly sacrificing many aspects that define the classic dungeon crawl to do so. Still, The Dungeon of the Dead Three and Descent into Avernus in general show care and creativity, and the contents of the individual keyed areas within it can be evocative enough that even a reader who doesn’t like the encounter based playstyle must recognize that Descent’s design choices are intentional.

The Dungeon of the Dead Three in particular deserves a closer look, because of anything within Descent it is most like a classic dungeon crawl, and seems to want to evoke the feeling of one - even if it pays no attention to the exploration elements of timekeeping, supply or risk management. Yet, despite disfavoring an exploration playstyle (which is hard not to with the 5th Edition of Dungeons & Dragons mechanics) Dungeon of the Dead Three includes many aspects that superficially make it appear to be a classic dungeon: a looping map that includes empty or nonessential rooms, traps and secret doors as well as a traditional feeling of the dungeon crawl - the infiltration or exploration of close corridors in an alien underworld. Here of course that’s focused on a sewer, which unfortunately is also a popular video game cliche, but at least it avoids including wererats. Because of these inclusions, it’s easier to conceive of the The Dungeon of the Dead Three as a classic dungeon, and despite its designers clear intent for it to play very differently, one can interrogate it in the context of running a resources, risk v. reward dungeon crawl.

Beyond modifying or including rules that better encourage exploration play (e.g. random encounters, timekeeping, meaningful encumbrance, and lighting) the question of “how does one design a dungeon that facilitates exploration play” remains. With the larger elements of the adventure: map, concept and basic structure in place, or at least not actively working against the dungeon crawl play-style, the core of the design process is in the individual location keys.

Keys are the basic building block of adventure design, information that the designer believes most important to understanding the location, provided in a way to allow the reader to run the adventure.

There's a variety of techniques to keying locations, from the ultra minimalism found in some of the first published adventures, to boxed text designed originally for tournament adventures, and bullet points or other, mixed types of formalism. The style used in Descent is mixed one: short boxed text, sometimes preceded by and always followed by GM directed text about room contents. Areas without encounters (combat in Descent) lack boxed text and have only short paragraphs. The writing itself is serviceable, but it doesn’t appear to have been intended as writing for a location based adventure, and it’s not well focused on usability. The organization that exists is a haphazard use of bolding to set off paragraphs about treasure or traps in some of the locations. This sort of effort is good, but without consistency it doesn’t help a GM run the location by highlighting the most important information in the key so that it stands out. Descent does make laudable effort to limit the length of its locations, but because of uninspired writing may do so at the cost of dulling down the play experience of the dungeon as a whole.

Boxed text is always a concern, it exists to regularize play experience, an understandable goal in the tournament modules that pioneered it, but unnecessary for players and GMs that aren’t in a tournament. Like all design choices, it has a cost as well as an advantage and that cost is generally a risk of confusion for both players and GM. There’s a lot of potential sins for a designer writing boxed text (or simple keys without read aloud text), and below is a list of some of the most obvious with notes on how well Descent’s Dungeon of Dead Three manages them.

Brevity - Boxed text especially needs to be short because read aloud it risks being a dry monologue that distracts both GM and players from the location and the game. Each sentence increases this risk, more than simple text, because it’s not in the usual words of the GM, instead it inserts the designer’s voice. Boxed text becomes less a conversation between players and GM and more story to be passively consumed. Read aloud almost precludes player questions and requests for clarification and it hampers the GM’s ability to answer these requests because the GM hasn’t had to visualize or understand the location they are describing, only read the prepared text aloud. Dungeon of the Dead Three does well here, its boxed text is short, usually only a couple of sentences. Indeed it’s short enough that a GM could easily paraphrase or adapt the boxed text into their own words without too great a risk of lost content.

Beyond the issue of brevity in boxed text there’s a larger discussion of key length and style, including the ways that formalism can help maintain clarity in longer keys or complex locations.In general however the goals of the key are brevity, clarity and evocative writing.

Subjectivity - Location keys describe a location - ideally they will provide to notable sights, sounds, smells and feelings of the site. What they can’t do, or at least shouldn’t do, is make claims about the characters' subjective experiences of the site. At the worst these sorts of keys become little vignettes describing how the characters enter the room, how they are startled by something within, their physical and emotional reactions to the room contents and what actions they prepare to take. It is impossible for the designer to guess the subjective experience of characters entering a location. A party of fur wrapped barbarians from the Ice Hells will have a very different conception of ‘cold’ then one made up of lizard folk from the ever-balmy Garden Isles, and those same lizard folk’s reptile brains or messy eating habits may mean that scenes of slaughtered mammals aren’t emotionally moving to them. Beyond the variety of character experience, it’s always good to practice providing player autonomy, even where it has no effect on the game - players are in control of their characters and the GM or worse designer should avoid pushing them into the passive role of observer or consumer of their character’s story.

There’s obviously some exceptions to this rule in a world where magic exists and can affect emotional and physical states regardless of character. In these situations though the effect should be part of the location’s nature, a clue to the players that something is different, that a magical creature or artifact is threatening the minds of the characters, warpping perception or emotion. Likewise conditions of a location are entirely appropriate and useful in a room description. It’s not that a designer can’t say a room is hot or cold for example, just that the designer can’t say a character in the room is hot or cold. The advantage of being rigorous about this rule, in addition to discouraging player passivity is that magical effects and charms become more noticeable - giving the GM another tool to hint at environmental puzzles.

As with brevity, Dungeon of the Dead Three does an excellent job here, none of the boxed text describes the characters emotional or physical response. There are a few minor, incidental descriptions that hint at making value judgments for the characters - a necromancer described as “frighteningly thin” (who says thinness frightens you characters?), but this sort of thing falls under bad word choice far more than excessive subjectivity.

Action Narration - Closely related to subjectivity, but different, a text description treads on dangerous ground when it describes what action occurs involving the characters in a new room. As with subjective experiences this is a problem because it forces players into the role of passive observer of their character’s story. Ambushes and traps still exist of course, and still surprise players, but presenting boxed text of the ambush makes it the only option in the key, the designer disdains and asks the GM to disdain player preparation. This appears a couple of times in Dungeon of the Dead Three. First in an early room where the party is ambushed by cultists playing dead.

Lying on the floor of this otherwise empty room are the pale bodies of three humans in filthy black robes, arranged in a triangular formation. A lit torch lies between them. A rough-hewn staircase to the left leads down to another torchlit chamber.”

The pale bodies are of course cultists pretending to be corpses and preparing to ambush the party, but even the text above this boxed text offers that the party may approach the room unnoticed by the cultists within. There are other possibilities as well - perhaps the party approaches quickly and cautiously to hear the cultists preparing their ambush, or they are disguised as cultists themselves. The boxed text offers one scenario, effectively forcing the ambush. It does this one suspects because the designer is enamored with the image of cultists rising up from false death to surround the adventurers in at the edge of an eerie circle of torchlight.

It’s a good image, but forcing this moment into the game via boxed text (along with various other pre-programmed encounters) prevents the Dungeon of the Dead Three from being a dungeon crawl experience, discarding the defensive plans of an organized enemy for a series of set encounter scenes. A rigid structure that doesn’t provide the GM information on how to run the location if the players make unexpected choices or use stratagems more clever then frontal assault.

A second and similar key involves a final encounter with dragon cultists as the party leaves the dungeon.

“As you emerge from the bathhouse, five figures leap down from the top of the courtyard wall to confront you. In addition to their black leather armor, they wear strange masks and cloaks that give each of them a vaguely dragon-like appearance. All five brandish curved steel blades reminiscent of dragon claws.”

Here the party will fight dragon cultists if they’ve recovered treasure from the Dungeon (a decent link to a much worse campaign - but it’s nice to see WotC pointing out ways its campaigns can branch off and tie together). Notably the cultists here don’t attack, and they won’t, assuming the party isn’t carrying treasure, but again there’s no accounting for PC caution. Both of these situations are caused by the use of boxed text, and while they may feel minor because a good GM can always adapt them to the events of her game, there’s no reason that good GM should have to, and there’s nothing gained in adding this boxed text. Boxed text can work decently for room description - if it remains short and well written, but it becomes troublesome when used to describe encounters with potential foes. In both the above situations the text covers the most likely event, and supplementary text indicates other possibilities, but the text doesn’t add anything, it wastes space and to a less accomplished GM it may read as if it describes the one way the encounter should occur and should play regardless of player action or the past events of that GM’s specific game. Perhaps WotC is using it merely as a way to highlight the tactics or descriptions of these foes, but again this can be accomplished without boxed, read aloud text.

WotC's new collector's editions have
excellent art.
Indeed, enemy tactics and goals are one of the best things a combat heavy adventure can provide. Knowing that a band of cultists likes to ambush their enemies, or that another group is single minded in its pursuit of stolen cult treasures opens up possibilities of play - options for the GM and even potentially for the players, at least when these goals and tactics aren’t locked into a specific encounter and behind read aloud events. In Descent this problem is part of the larger design choice to focus even its putative “dungeon” on relatively linear encounters that create moments.

Detail and Word Choice: Besides brevity the read aloud in Dungeon of the Dead Three is functional if uninspiring. Excessive and useless words are frequent and repeated, while useful or evocative detail is avoided. Descent wants the players to know that every torch in the Dungeon of the Dead Three is held in a wall scones, but it doesn’t bother to describe (even with an adjective or two) these recurring bits of dungeon decor. Word choice and detail might seem like a petty complaint, or perhaps just something that depends on the writer’s skill, and they certainly aren’t specific to boxed text. This is of course true, but it doesn’t take an amazing writer to write a good adventure, it only takes an understanding that adventure design, especially key design, is both functional and artistic writing. Likewise Boxed Text exacerbates the failures of the designer as a writer because it insists on providing the description of the location, rather then offering a description of it that the GM can put into their own words.

Practically, key design is an odd combination of Poetic Writing - specific words within a restrictive structural limitations to evoke emotion and imagery - and Instructional Writing - using organization and simplicity to provide clarity of message efficiently. Both notably emphasize brevity and form, while there’s a potential conflict between evocative poetic word choice and clear directive business/instructional writing. Resolving this conflict is a place for writing skill, but it’s only necessary once the basic requirements of structure and clarity are addressed, it’s also not something that requires mastery when a basic form is in place.

Basic form is perhaps too strong a concept - there’s more than one option: boxed text, bullet points and introductory highlights, paragraph organization, and even outliers like an information matrix or procedural generation all have long histories and all seek to address the essential difficulty that separates adventure keying from other forms of creative writing. Adventure keying, at least in a classic exploration game, is an effort to provide sufficient information to the reader to envision a fictional location without controlling the narrative importance, effect or action of and in the location. At the same time this information needs to be in a form that the reader can both relay to others (the players) and sufficiently evocative and/or detailed to build or improvise on in response to questions. Obviously confusing, convoluted description and complexity make this harder, but so does vagueness, blandness and lack of content.

The writing in The Dungeon of the Dead Three tends towards the second of these, wasting space with unnecessary and repetitive references (like the torch sconces mentioned above), but also a few incidences of Inaccessible Lore and a general failure to emphasize room elements that will be of interest to players. The keys are instead optimized for the mechanics of the combat encounters they include, and while its an excellent design choice to note tactics, there’s little else offered in many of the rooms - nothing to examine or wonder about, just similar combat with small numbers of disorganized foes who each have a minor mechanical variation in their tactics.

Even these enemies aren’t well described, and while WotC has made a sometimes inexplicable design decision to place most monster stats in an appendix, the descriptions in it rarely offer much detail about the appearance of the creatures. While this monster appendix includes illustrations, they are few and they don’t tell much. The failure to provide detail or description that sets various foes apart creates a risk that a GM will simply refer to them by name “Necromite of Myrkul” or “Iron Consul” making them into known entities and reducing the monsters quickly to bundles of combat statistics, rather then unknown qualities for the players to approach with caution, consider as full fledged fictional characters, or attempt their own classifications and develop their own understanding of. It’s good to know what Bane’s cultist think and how their organization views itself, but more individual description then their love of black clothing and heavy armor makes them distinctive and interesting, especially in a scenario that focuses on repeated combat encounters with the same cultist types. These feel less like Tabletop RPG foes, with options to negotiate and goals or plots beyond direct conflict, and more like enemies in an ancient roguelike video game - rushing down corridors to attack.

Distraction and Boxed Text: Boxed text makes bad detail and uninspiring design worse because the designer presents it not as a utilitarian document for the GM’s use but as player facing description. When there’s no read aloud the GM must make up their own description, and even if this doesn’t possess literary or poetic merit the GM’s text will be appropriate to the GM’s table and it will be clearer to players used to their GMs style. Good key text can of course give better images and even words to a GM, so that the GM can pick and choose based on his own interests and understanding of their players and setting.

An Example of Detail: Room keys in exploratory games should help the GM run the room, and even if they can’t offer answers for every potential player question, they are best when they offer details and imagery that allows the GM to easily extrapolate. Despite a focus on encounters, Dungeon of the Dead Three does offer some of these details - but it tends to do so in a flat way that makes them hard to build off of. For example a room containing a rather good puzzle - an evil altar that offer skeleton minions to characters who foolishly dare to read its inscription aloud is described as:

“This dry, partially collapsed room contains a stone altar with humanoid skulls and bones piled around it. The top of the altar is covered with dozens of half-melted candles made of black wax, all currently unlit.”

There is nothing wrong with this description, I picked it based on its length and at random, but the details it offers (black wax candles and piled bones) aren’t especially important to the location, while it misses even a hint at the most important detail in the room. The same glance that catches the candles should reveal that the altar has a words engraved around it. While brevity might be central to a good key, there’s enough space to create concrete details that highlight the important aspects of a location. With an interactive object or puzzle in the room it isn’t useful to hide the important details behind the mundane ones, unless there’s an in game reason (a concealed door say). Nor is there any other descriptive detail about the altar beyond its contents and effect trigger. Generally important piece of dungeon furniture act as landmarks and memorable part of the play experience, meaning they deserve memorable description and expressive language.

“Dark, silent, dry, and half choked with tumbled stones, a narrow path between mounds of brittle bones and gleaming skulls winds through this low ceiling chamber to a high cracked marble altar. The drippings from black sagging candles streak the prayers engraved on the face of the altar and scent the air with tallow’s acrid reek.”

I’d call this a better description even if it’s a line longer. It’s not better because it’s more literary - it’s better because the description has more specific details and so both a more vibrant imagery and more information (including a reference to the puzzle/interactive element of the location). The players both have a greater understanding of room layout (a path to an altar between bone piles) and the details are set off with more information that helps envision them. Likewise the smell, sound and feel of the room (tallow, silent and dry) are engaged in addition to the visual. The extra length is even justifiable as it allows some of the details about the engraving to be removed from the GM facing text.

Beyond the issues I’ve associated with boxed text (which also apply to any form of adventure keying) there are other more general concerns - some small and easy to improve and others large and touching on the core questions of adventure design. I’ve just laid them out briefly with examples of how Descent and the Dungeon of the Dead Three succeed or fail at them. Like everything else here, nothing is absolute - the best location key is the one that works to aid a GM in running a game that they and the players enjoy while also making clear the designers intent.

Room Names: It’s fun to give literary names to locations, as one would to the chapters in a book, or a poem. Jokes, references and subtle riddles about the location make the designer feel clever - clever or not they are a bad idea. A location key is a functional document as well as being a poetic one, and the reader has (sometimes very quickly) digest its contents and then explain them to others. A clear and descriptive room name titles the contents of the room or describes the events within can set the tone and context for the rest of the key. For some rooms a name is almost all that's needed. "Cesspit" or "Musty Bedroom" do a great deal of descriptive work, leave a clear set of ideas for the GM to elaborate on and allow the designer to note only important and unique features in a key. With a more prepared GM the room name also offers a hint to jog the memory of the past reading.

In addition to the literal name of the room, the room number itself is a tool for the designer. Numbers should flow naturally, as sequentially as possible on the map, with room 1 the main entrance, and the rooms closest to it in sequence. Regions in larger dungeons can be broken up into lettered subsections for further recognizability.

In both of these cases The Dungeon of the Dead Three is fairly successful. While some of the rooms have confusing or meaningless names, "Welcome to the Dungeon" or "Bhal's Rest" the majority are clear and descriptive. The room numbering scheme is a somewhat confusing but largely clear. A dead end is room six, while the rooms leading to it are 7,8 and 9 in reverse order - though a secret door leads directly from 5 to 6, so even this is comprehensible.

Inaccessible Lore: A regular compliant in adventure reviews is the inclusion of too much history, lore or backstory about locations and adventures: the past uses of rooms or historical events that led to a location's decline, ruin and transformation into a dungeon. Designers padding room descriptions with unnecessary historical details is a common problem. Perhaps it's part of many designers process - thinking through the location's past, but only rarely is there a need to provide this to the GM running the adventure, and even more rarely to the players.

Most lore is simply inaccessible. Characters have no reason and rarely the means to figure out the histories of the dead whose tombs they loot - unless those histories directly impact their ability to loot. This is of course where the general prohibition on inaccessible and excessive lore runs into an important element of exploration play: secrets and discovery.

One of the most important aspects of exploration is the players unpuzzling the location. Learning its layout and the best paths through it - what I call orienteering, but also learning the way the location works, what symbols and objects within are significant and its significance in the setting as a whole. Often this significance includes history and hidden secrets that help the players navigate and survive the dungeon. The original uses of a location can define its rooms and their spatial relationships. A ruined fortress will have few entrances while a palace has many.

Understanding the history and present of a location should be useful to the player characters. Furthermore, many players find unraveling secrets, and discovering the connections between a location and the setting as a whole is rewarding. Thus a simple prohibition on history or lore in dungeon keys is shortsighted, but lore must be accessible and useful.

Players learning that a crow symbol on a sarcophagus indicates the body within is an undead guardian is useful. Knowing the personal history of the crow guard and the names of its five founders, wight kings destroyed eons ago, is not.

Hooks and rumors can provide knowledge of history or other useful fragments of lore in game, and exist outside of location keys. Within keys lore is best limited to what's immediately observable by the characters and so left fragmentary - at most offering space for speculation. The designer might know the whole history of a location to define its internal logic, but the key doesn't need to relay this except as architectural or other passive detail.

The Dungeon of the Dead Three largely avoids unnecessary and inaccessible lore - running counter to stereotypes of WotC campaign books and adventure paths generally. It's perhaps too reluctant to provide historical or lore context, though this is largely the product of Descent's adventure path design which has no use for player investigation or research as a means of discovering dungeon locations. Rumors aren't especially helpful when propulsive plotting decides where and why the party will next adventure.

Level Organization and Room Complexity: A sign of the enormity of Descent’s departure from the classic dungeon crawl and exploration play is the density of encounters on its map. Dungeon of the Dead Tree has 33 keyed locations of those 12 are combat encounters (red on the map below), 7 traps (blue on the map), 4 secrets or puzzles (green) and 4 treasures (yellow). Some of the encounters mix trap or puzzle and combat - including a potential ally, a betraying prisoner, and guarded secret door. However both by looking at the map and the numbers above, it’s clear that there is almost no empty space in Dungeon of the Dead Three - almost every room requires action, with 7 empty rooms. In a traditional dungeon crawl the percentage of rooms with no active elements (resident, trap, puzzle or treasure) is around 30%. This ratio of empty rooms is high for contemporary design, and unnecessary even in a resource management focused game where the rules have been adjusted for shorter session length (slot encumbrance, exploration/overloaded encounter dice etc.) However, empty rooms are an absolute necessity in level based design and for resource management play. Empty rooms extend the length of time required for exploration - depleting resources and building risk. Empty rooms also add spatial complexity, making more complex maps and their unpuzzling possible. Finally empty rooms simplify key design as they give the designer extra space to include detail and offer access to information through dungeon dressing without cramming too many elements into a single key.
Red: Combat, Blue: Trap/Trick, Green: Puzzle/Secret Door, Yellow: Treasure

The Dungeon of the Dead Three isn't a level designed, exploration adventure, but instead a series of encounters without resource management mechanics. Its challenges and obstacles each exist independently, so space is a pointless delay, and unpuzzling the overall spatial or social arrangements of the location unimportant.

Keys, Interconnectivity and Encounter Design: Keys should relate to the adventure as a whole. While it’s tempting to only discuss them individually, keys act collectively to create a location as a whole. If classic level based design has a story or narrative element it’s found in the relationship between keyed areas: the tensions between factions, implications about lore, and flavorful details. In a scene based and encounter designed adventure like Descent into Avernus the keys however are literal scenes in a set narrative. It hardly matters that Descent avoids the pitfalls of forcing subjective character experience of narrating action in its keys when the goal of them is to create a specific story experience.

The previous post here discusses how the keys in The Dungeon of the Dead Three insist on creating “Moments” rather than “Locations”, meaning that the goal of its keys isn't to evoke the visualization of a space for interaction and exploration but instead to create a dramatic scene and moment of excitement. Ultimately this shouldn't change key design much - the same challenges of brevity, clarity and arresting or inspiring imagery remain. The emphasis just changes and with it the interactions between keys. Encounter based design needs to link the encounters (and keys) together into a credible story, while level based design needs to organize the keyed areas in a way that makes spatial sense. There's surprisingly little crossover. With encounter based design the space between encounters is elided and mechanically dismissed - the party moves from moment to moment and narrative event to narrative event with spatial concerns (and the map) acting predominantly as scenery. In a level based game story is formed largely by the goals of the players, how they conflict with various npc factions, and other obstacles or puzzles. Predetermined story exists only as probability (if the PCs do X, Y will happen) or as history and rumor. In either level or encounter based design the key has to focus on different aspects, while the need for brevity sidelines others.

A well written key will uphold the structure of the adventure it’s part of, working with the style of the adventure to organize the larger location/story and connect disparate keys together in a way that is logical.

Dungeon of the Dead Three’s keys, even accepting that it’s seeking to create a narrative rather than a location, don’t feel especially well connected. A few tie into outside events (kidnapped prisoners) but the individual keys and the moments they create, while maintaining a constant theme of evil cult lair, are independent, unconnected by time or relation to each other. Prisoners can be freed and betrayers allowed too close, but they don’t do much in later scenes, or provide valuable or misleading information. The alarms can be raised, but it only effects the immediate encounter. While the classic solution to this is to treat the dungeon like a fortress under siege - with its inhabitants a faction whose tactics, order of battle and weaknesses combine with the dungeon layout to create the complex problem that the adventure asks the players to solve, less complete revision are possible. Keys can interconnect, simply by implying relationships between the foes encountered in them, or by making references and providing clues to each other. For example, the cultists undountedly know the traps in their own dungeon, and it makes sense that they might use them or reveal them if captured.

Yet Descent is too enamoured with the individual moments it creates to make anything but thematic connections between  encounters: a brutish giant battling an assassin in a pool of dead cultists, cultists playing dead who rise from the dark to attack, or a scene of torture. Each key is largely independent, each key a singular moment and image relayed by boxed text. In any style of play this is a failure.

Formalism: With brevity and clarity two of the most important parts of writing an effective key there’s always been an interest among adventure designers in formalizing location keys. From simple structural guides to more rigid types of formalization like grids or bullet points, these efforts have mixed success. On one hand formalization is helpful because it systematizes each key, theoretically making it easier for the GM to use. On the other formalization can stifle detail, description and creativity and can emphasize unimportant aspects of specific keys.

Some formalism is always necessary, almost synonymous with organization, and Dungeon of the Dead Three takes this minimalist approach - its keys aren’t simply rambling narrative descriptions of the location. First it breaks the Dungeon of the Dead Three into discrete, and rather specific areas - room by room description. Then of course it uses boxed text to highlight important descriptive passages. It’s use of boxed text is more clever than most adventures that choose to provide it, saving boxed text for important locations with more interactive elements, and leaving other, less important rooms with only GM text. Other formalist elements the Descent designers include are bolded text for NPCs and monsters (alerting the GM that there’s a stat block somewhere for an entity with that specific name), paragraph structure, and general description.

Paragraph structure is a useful and intuitive type of formalism, it's one we're all familiar with, because it's a cornerstone of coherent writing. Organizing paragraphs is also an art of sorts, and largely dependent on the nature of the writing - your average legal brief will use paragraphs differently than your average novel. As discussed above, adventure keys are a special sort of instructional writing with a sometimes contradictory poetic flourish. This means paragraphs should aim to be short, descriptive and self contained. Some designers break them up with subheadings - a first paragraph to describe the room as a whole, the second its inhabitants, and subsequent ones each describing important interactive aspects of the room.

General descriptions are another formal way to organize keys, avoiding repeated description of basic location elements: ceiling height, wall material, doors, and other common environmental elements shared by the entire advetnure locale. This can be a wise choice, as it focuses keys on the location they alone describe, and helps keep them brief. It's also work (along noting with specific room dimensions and exits) that GM aides - specifically maps and illustrations often manage more effectively. Another issue with the use of general descriptions in Dungeon of the Dead Three is that what’s provided isn’t simply general description, but important combat and exploration mechanics - the fragility of the dungeon’s supports and a mechanic for roof collapse. While this is meant to offer the players a general way to do exciting things in combat (though perhaps some limit on the amount of supports one can break and roof falls one can trigger without precipitating a general collapse of the tunnels might be worthwhile) but it’s a video game style universal level mechanic, rather then something considered as a coherent part of a fictional space. Level mechanics need to be highlighted, and the structural collapse mechanic should at least be boxed off in general description and highlighted in areas where it’s appropriate (as opposed to a general mechanic that’s always active).

More complex formalism is also possible - usually in an effort to reduce the size and complexity of the individual key. It can work, especially as an ancillary to good writing and details, but it can be counter productive - building a matrix for each key that repeats or contains meaningless and basic information simply to fill in the blanks set by the form. As with most things in adventure design, the more one leans into an extreme, the more likely it is that the negative aspects of that extreme will come to dominate. See the note below about formalism and minimalism in early adventures for an examples.

GM Aides: One of the notable keying decisions for all of Wizard’s 5th Edition adventures is the failure to include monster statistics in the text. Instead large statblocks with minimal descriptions and occasional art wait in an alphabetically ordered appendix at the back of the adventure tome, or the GM is referred to additional books, such as the Monster Manual. For such relentlessly combat focused adventures this seems like a poor design choice, but 5th editions start blocks are both long and formalized so putting them into each key, especially repeating them, would also be disruptive and inefficient.

This may seem an odd subject to raise in a discussion of "GM aides", but it’s symptomatic of a failure in Descent, WotC products, and published adventures more generally - a lack of effort at usability. In a combat focused adventure, monster statblocks are very important to the GM, who will be running frequent combat encounters, and the adventure's designers should both recognize this and provide what help they can. Simplified stat blocks need to be at hand for running the combats in The Dungeon of the Dead Three, and asking that the GM either prepare a digest for them or flip back and forth between the adventure text and an appendix full of monsters is bad design.

A simple GM Aide - a sheet with monster stats by keyed location to tear out or reproduce would make running the Dungeon of the Dead Three much easier, likewise a map that notes the locations of traps, monsters and notable features of each room. If the adventure were more complex other types of aides could also be a great boon to the GM without much effort on the part of the designer. Orders of battle with stats and tactics to track a faction’s available resources for an actual siege/infiltration scenario, a web of relationships showing loyalties, hierarchy and plots would help for adventures with mystery or negotiation, a calendar, clocks or timeline if events unfolded as something other then a linear story, and random encounter tables for exploration scenarios.

It’s perhaps unfair to focus on what aides Dungeon of the Dead Three lacks, because Descent’s designers as a whole include plenty of useful aides in the adventure’s appendix’s. None of these seem directly aimed at making the adventure easier to run, or reducing GM preparation though. Even if the extra information and tables to generate Baldur’s Gate encounters or plot hooks are a great addition to the campaign and setting, they aren't diretly assisting one in playing the larger adventure. There’s a place for these informative appendix’s as well - they generate GM excitement about the adventure and place it into a larger context, but they aren’t aides in the sense of bolstering immediate playability, the sort of aides that have the potential to make a good adventure great, and make a mediocre one usable.

Below is a complete analysis of a sample key, picked because it’s a fairly interesting, relatively brief, and interactive location, but also because it highlights several of the problems and the relentless encounter focus of Dungeon of the Dead Three:

D 26: Bhal’s Rest (1)

Read or paraphrase the following boxed text when the characters enter this area for the first time:(2)

Four flickering torches in wall sconces (3) light this partially collapsed crypt. An open sarcophagus (4) in the middle of the area is filled to the rim with blood, with spillage streaking the sides of the sarcophagus and pooling around its base (5). The sarcophagus lid lies half buried under rubble behind it. (6)

Bhaal worshipers like to bathe in the human blood that fills the sarcophagus. (7) A female reaper of Bhaal (see page 233 for her stat block) lurks behind the sarcophagus, drenched in blood from head to toe. (8) At the first sign of intruders, she uses a disguise self spell to appear as a frail old woman named Nebra. (9) In this guise, she claims to be a florist who was captured by the Dead Three and brought to the dungeon as a slave. If the characters fall for the act and bring Nebra with them (10), she claims to know the location of a secret door (area D23) and tries to lead them through it to area D29, where she hopes to find reinforcements. If she leads the party to a location occupied by other cultists, she drops the disguise and attacks. She also attacks if her spell ends before then. (11)

1) The room’s name “Bhal’s Rest” doesn’t reveal anything about the location. It’s cool sounding maybe, but the evil deity Bhal isn’t resting in the room - it’s not a literal description, neither is it a description of events likely to occur in the room. Bhal’s Rest is a bad name, because it tells almost nothing about the room, offering the GM immediate cue, and given the number of references to Bhaal in the Dungeon of the Dead Three it’s not triggering the GM’s memory either. Better names might be: Blood Altar or Bhal’s Lesser Shrine.

2) This is an entirely superfluous sentence, and wasted space on either side of it. Text in a box that is obvious description is a read aloud. Even if the GM is entirely new to the concept, by room “D 26” the GM should know how it works. While I appreciate the inclusion (and repetition) that the GM might want to paraphrase the description - a paragraph talking about how to use boxed text early in the adventure would allow a better discussion of how boxed text works, its limits and when to depart from it.

3) The sconces. This isn’t really bad description, but it's directly copied from several other keys, repetition again. Given the importance of torches mounted in wall sconces this is sort of thing that should be in the general description section. Something like “All rooms are lit by two to five tarred rush torches set in hammered wall sconces unless otherwise noted”. An ambitious map might even include shading and light circles around marks to indicate torch locations.

4) Rather than pointless detail here’s a lack of detail. Room dressing tends to become an immediate focus of player interest, and sarcophagi are in a special category of very important room dressing - tending to contain undead horrors, grave treasures and fiendish traps. Players often want to know every detail of such a potentially important and interactive object: material and decoration for starters.

5) A repurposed sarcophagus filled with blood is a stunning image, but there’s a wasted chance here as well. The description of the blood is fine, a bit more simplistic and less poetic then it might be in a better writer’s hands, but sufficiently powerful. What’s missing from this description is any other sense information besides sight. Smell seems the most important here - the metallic slaughterhouse reek of rotten blood is as disturbing as the image of the blood filled coffin. Don’t forget hearing and smell because they can both offer interesting clues to the environment (the blood is not quite fresh, or definately blood, not red paint or dirty water).

6) The fate of the sarcophagus cover is both meaningless here because it has no interesting detail or clue about the room to impart (unlike an earlier one on a trapped tomb depicting a screaming barbarian) and inexplicable as part of the boxed text. The lid is “half buried” in rubble, so presumably it’s not something visitors will immediately notice. While there’s reason to include details of all sorts, and red herrings, a blank, half buried coffin lid seems far less important than say the blood drenched, disguised cultist crouching behind the tomb and the inevitable trail of dripping blood that leads to her.

7) While it’s a stunning image that the cultists bathe in a makeshift font of rotten blood this is unknowable information for the players. It’s not entirely useless GM directed fluff, as the cultist may need to explain why she’s drenched in blood, and a character with knowledge of Bhaal worship might know about the blood bathing practice, but this shouldn’t be the first GM facing sentence.

8) Only now does the key get to the resident of this room - a nameless cultist who the party will likely interact with. This should be in the first sentence of the key, her presence is the most important part of the room, and while she’s hiding, the only explanation for why it isn’t is a limitation of boxed text. Because boxed text can’t include things that aren’t immediately obvious to the players, there’s no indication of the cultist until here, fairly deep into the key. That’s likely to cause confusion, especially for a GM using this key without having read it first, made notes and prepped the dungeon. It’s also easy to avoid, even without abandoning boxed text, by either noting that “A cultist hides in this room intent on deceiving the characters” prior to the boxed text, or otherwise using formalism to indicate that there’s an important hidden aspect of the location. Even introducing the cultist in the first sentence below the boxed text, would be an improvement.

9) There’s no description of the cultist or her disguise, though to the designer’s credit, the cultist’s plans and planned deception are noted. Describing monsters is always an important element of the adventure key, because the unique nature of a monster helps make it interesting and compelling, even if it’s directly from a monster manual. Something to change the monster from a pile of easily learned statistics in a manual into an active part of the setting can really add to player buy in and encourage both reflection/speculation about creatures and solutions other than combat.

10) This sentence feels like an aside, and as a matter of line editing it’s entirely unnecessary - it also shows a bad tendency to center the monster around the PCs goals. If the PC’s are deceived something will happen focuses on the player reaction to the plan, rather then the cultist’s plan itself, her goals and her methods. A GM can’t honestly or realistically control the players response to encounters, but language of this sort encourages that misconception. Better for designers and GMs to focus on what the setting does, regardless of the players response. Better simply to describe the cultist’s general goals and potential plans: “The cultist attempts to use her spell to deceive the PCs and offers to lead them through a secret door in the hope of springing an ambush. If rebuffed or the odds look bad she will try to escape and use her status as a rescued victim of the cult to damage the party’s future plans or provide dangerous misinformation.”

11) Including monster and NPC tactics is always great to see in an adventure key, but here the tactics are stilted an unreasonable. While the idea of infiltrating the party and attacking once allies are present makes sense, the cultist has no alternate plans - because she has no goals - simply waiting to launch her one woman assault when her spell fades. Designing complex foes requires more than just a statline and a single tactic that are related to the GM like a script for the monster to follow. When monsters have goals and personality the designer gives the GM tools to improvise when things depart from the designer’s expectations. A sentence indicating the cultist’s goals, personality and name would allow her to be an interesting part of the rest of the adventure - rather a tactical complication in one of The Dungeon of the Dead Three’s set piece combats.


Cover of Palace of the Vampire Queen
with a great font and color
The earliest adventures, specifically 1976’s Palace of the Vampire Queen by Peter and Julie Kerestan (you will find a lot more women involved in the earliest days of the TTRPG hobby then popular impressions might lead one to believe) and 1975’s Temple of the Frog by Arneson (which may also be the first published adventure) both rely on ultraminimal keys. These are playable adventures, but they require the GM provide all the details and description. Palace of the Vampire Queen is exceptional in its brevity - relying only on a matrix with columns for each “room number”, “creatures encountered” “Max Hits” [HP], and “Room Contents”.

The confrontation with the Vampire Queen herself is thus described as:

Room 30 - 1 Female Vampire - 40 [HP] - “Vampire Queen. Will turn into bat and escape if door is not shut behind party. If she escapes she will go to Dwarf Princess, if Princess is still living and threaten to kill her if party does not let the Vampire Queen go free.”

This is minimalism in its most refined, and useless, form. Such details are easily generated by a set of very simple tables, and give the GM so little to work with that the entire prepared portions of Palace of the Vampire Queen, the shared content found across tables and GMs consists of specific monster stats, maps and a few implied faction relations. The specifics of the rooms and the nature of the creature are left almost entirely to the individual GM - does the Vampire Queen live in a perfumed boudoir of opulent silks, or a decayed one filled with the mummified corpses of her playthings? Perhaps she is a brutish ½ bat thing wearing a crown of bones and the room is a bare stone lair? The adventure key doesn’t help here but there’s not really enough information to even decide how the adventure is meant to be played - a plodding gamified exercise in hack n’ slash where the monsters stay in their individual rooms (a monster zoo) or something more? This of course means that Palace of the Vampire Queen can still function as a module, in almost any mode the GM wants - because the GM and players will be making the decision about how the adventure works: as a combat gauntlet, full of faction intensive negotiation, level based exploration or something entirely different. The keys don’t provide more than a vague inkling of the Kerestan’s vision, though some comes through from the introduction, they’ve cleverly implied relations between various monsters and a lot can be inferred because of the large shared cultural repository of vampire tropes. Still Palace of the Vampire Queen’s willingness to force the majority of preparation onto the GM means that its ultraminimal keying considered a failure by contemporary standard despite the simplicity and clarity it offers - there’s simply not enough character or detail to make it a useful product.

Arneson’s Temple of the Frog is very similar, though it uses a paragraph style instead of a room matrix:

Room 1 [of the lowest temple level]: The Chief of the Keepers of the Frogs lives here. He is a 9th-level magic-user and pos-sesses a Ring of Animal Control. He has six Thaumaturgist assistants (each of which have rings to control the killer frogs). The Keeper takes 24 hit points and his servants take 13, 10, 10, 9, 17, and 15 points respectively. The room contains seven beds, three desks, eight chests with garments and, in the center, a giant stone frog on a pedestal. There is a secret compartment where the Keeper’s treasure is hidden and where three persons may fit. There is a cook-fire in a niche on the eastern wall.Treasure: 16,000 gold, thirteen gems of 1,000 gold piece value each, and 14 pieces of 300–1,800 gold piece value jewelry.

Art from a later edition of Temple of the Frog
This is one of the longer descriptions in Temple of the Frog, and while it offers more depth and detail then Place of the Vampire Queen - including details such as room contents, is pretty sparse. The detail provided is also rather predictable (even that “giant stone frog on a pedestal”). The difference of course is that the paragraph style Arneson uses offers room for expansion. One can’t really build much more detail into Vampire Queen’s room matrices, but a few extra sentences and Room 1 from Temple of the Frog could offer a more interesting description.

This is of course what happened. Over time through AD&D and into onward into later editions maximalism and detail begin to predominate in room keys. In many ways this is worse then ultra-minimalism, because it makes the key nearly unusable, hiding information inside dense lengthy passages. A frustrating example of this kind of maximalism is found in the “Ruins of Undermountain”, a 1991 boxed set by Ed Greenwood (of Forgotten Realms fame) that explored the mega dungeon under Waterdeep. The adventure has been entirely redesigned recently as “Dungeon of the Mad Mage”, but in its original form its a monument to obsessive, useless detail. Consider this key on the first level for an alcove containing a skeleton, a hidden stirge and an ivory scroll tube.

Boxed Set Cover - Ruins of Undermountain
Room #3: Nimraith’s Rest
In this alcove lies a human skeleton,huddled under a riven brass shield. It is clad in the crumbling remnants of leather armor. A metal, visored helm sits upright on the flagstone floor nearby.

The skeleton is human, and although it is intact, even to joint-sinews, it is not un-dead. The shield, split in two and pitted here and there with weak, crumbling ar-eas (from long-ago contact with acid), is effectively worthless. The word ‘Nimraith’ is etched across it, in Thorass.The leather armor has also perished; it will crumble away into dust and mere shavings when touched, yielding only rusty buckles to scavengers. Amid its folds, entirely under the body, are the rotting remains of a knapsack, of the same material. This knapsack contains a smooth, polished ivory tube (worth 2 sp), and a mouldering mass of cloth scraps (once a piece of cheese, wrapped in cloth). Inside the tube are two scraps of parchment.One is Nimraith's Map (see Maps & Charts of Undermountain in the Undermountain Adventures book) and the other is a scroll on which are written two cure light wounds spells.The helm is intact, but eaten from within by rust; a being of average strength can drive his fingers through it while handling it.The helm contains nothing of value, though it contains something.

The helm is also the sleeping-place of a thirsty stirge, who will burst up at any being disturbing the helm, flying to attack the face or any other exposed flesh. If no such easy target exists, the stirge will fly rapidly about searching for one, swooping and darting to avoid missile fire and spells, with which it is familiar. The stirge must be killed to remove it from a struck victim; attacks against it while it is attached may hit its victim roll a second at-tack roll against the victim’s AC to determine.

Stirge (1): Int Animal; AL Neutral; AC 8;MV 3, FL 18 (C); HD 1 + 1; hp 9; THAC017; #AT 1; Dmg 1-3 plus blood drain of 1-4 per round thereafter (to a 12 hp maximum); ML 8; XP 175.

I won’t go into the number of ways this key is poorly written, and it certainly gives the reader a lot of information to work with: the acid rotten shield, the darting stirge, and the foil thin helmet - but all of these details offer very little to play, can be edited down substantially or are the sort of thing a competent GM should improvise. Imagine trying to use this description in play? The one sentence boxed text is easy enough to read, but the stirge ambush is concealed after the lovingly detailed ruin of the skeleton and its equipment. A GM struggling to use these details is likely to miss the stirge entirely, unless they’ve prepped the room extensively.

If they have prepped the room it’s likely to write something like : “STIRGE under helmet! Skeleton heal scroll < pack” on the map or a piece of note paper. While the GM may remember some of the specific types of decay suffered by the dead adventurer’s equipment, they are unlikely to remember it all, and if they have to refer back because a player is insistent about examining some of it, it’ll be hard to find Room #3 (with it’s confusing name, lost amid even longer blocks of text), and harder to pull those details out of the dense description. There’s enjoyment in these sorts of maximalist room descriptions - a joy in knowing all the details, something akin to a forensic or archeological reconstruction of the space and its inhabitants. There’s also a joy in minimalism and formalism - ordered and catalogued space that quickly fills the map.

However much fun it is to read the exhausting detail of Undermountain or to take in the huge scope of Palace of Vampire Queen in a glance, these design extremes don’t improve playability. At the heart of adventure design a key still needs to be: Brief, Clear and Evocative and for all its brevity minimalism is never evocative, and maximalism quickly loses brevity and clarity. Dungeon of the Dead Three may not be optimal, and it’s easy enough to find problems with its keying, but it does find a useful line between baroque excess and sparse formalism that is worth emulating.

1 comment:

  1. Those posts about Descent to Avernus (and actually so much more) are fantastic. Thanks!


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...