Descent into Avernus - From a Dungeon Crawl Perspective

Cover of Descent Into Avernus

With the basic ideas behind dungeon crawl style play covered, I'll be taking a look at the current state of contemporary Dungeons & Dragons adventure design and how it succeeds or fails to deliver a Dungeon Crawl or Classic Play experience.  Specifically I'll be looking at the recently published Wizards of the Coast ("WotC") campaign book "Baldur's Gate: Descent into Avernus" ("Descent"). My goal isn't to attack or denigrate Descent or the play style it supports, but to discuss where and how it follows design principles that support classic play, where it departs from them and to what effect.  I may also be able to offer some ideas that will help others run the adventure in a more classic way emphasizing: dungeon crawling, player choice, and open worldbuilding.

Fairly typical of WotC's contemporary adventures Descent is a 200 plus page series of adventures that make up a campaign that will take characters from 1st to approximately 14th level.  It's designed for many sessions of play and an epic scope. The campaign is the product of a large team of authors, designers and artists including D&D's current creative leads Mike Mearls and Chris Perkins. It's also nice to see that much of the cartography within is the work of Dyson Logos, a blogger and map maker who I consider to be broadly part of the same community as All Dead Generations and whose distinctive cross hatching style is inspired by classic map design.

FIRST THE STORY

Descent is an epic story, and this is it's first goal, and the first way it departs from classic sensibilities.  Organized (as are the vast majority of contemporary WotC adventures) into Chapters Descent is a linear narrative where the players follow and unfolding danger, overcoming challenges and gaining power as they go. It's writers don't countenance players deviating significantly from the chapters, their order and the consequences or events of each.

This is a very different way of playing, worthy of a larger discussion about player agency, world building and illusionism, but for now it's simply the way this adventure is designed. Descent is an "adventure path" - a series of scenes (some including longer adventures) that have predetermined endings and tell a particular story that defines player goals and bends player decisions and actions to it rather then resulting or transforming from player goals and decisions. Descent however is an exciting a pretty interesting story - an epic fantasy that has significant charm and promises adventure on a grand scale.

The entire city of Eltural has been drawn into hell, and with it one of the leaders of the character's city Baldur's Gate.  The resulting power vacuum and the machinations of the same diabolical forces that carried off Etural now threaten the character's home with the same terrible fate.  This is a great situation to start a campaign - the growing influence of nightmare powers disrupts the lives of the characters and the community they live in. The ways that Descent manages and aggressively twist a specific adventure from the situation are less inspired, and depart significantly from the practices and principles of classic play.

There's nothing in the basics of Descent's starting situation, the story that precedes play and sets the base state of the game world, that is unsuitable for classic play, which may be a surprise if one has an absolutist view of the design of older Dungeons & Dragons. The idea that "sandbox" or open world play must be entirely driven by player generated goals, and that the universal character goal in classic play is the acquisition of wealth is to a great degree a lack of foresight and imagination. While many location based adventures or modules from the 1980's start from a static or neutral position, and the scale of older published adventures is usually limited to a specific location meant to be added to a larger campaign, this is not universal, nor is it necessary.

Allowing, encouraging and reacting to player goals as a GM or designer doesn't depend on presenting a world that's completely open or offers moral blankness.  Players need information and knowledge to decide what their characters goals may be - and the initial situation, the setup in narrative terms or perhaps "base state" should offer those. Desperate drifters newly arrived seeking opportunity in a lawless hinterland is a classic and simple base state drawn from the Western Genre. It's found often in classic adventures such as B2 Keep on the Borderlands, but the first published adventures also offer different base states such as the characters acting to protect their threatened polity in G1 Steading of the Hill Giant Chief or seeking to infiltrate and sabotage a rival state in Temple of the Frog.

Beginning with a setting threatened by the forces of hell is entirely within this tradition, and player choice at the highest level of the story can be protected when a designer or GM doesn't assume or force a specific player response to such a threat.  Players might choose to: resist the hellish incursion, flee to safety, profit from the conflict, or even side with the denizens of hell. Any of these possibilities, or a combination of them, offer plenty of opportunity for adventure.

Descent however does force a specific set of player goals, often clumsily, and it's reliant on the players accepting the role of their characters as saviors and protectors of Baldur's Gate.  While this is the role most players will take presented with Descent's base state, there is no need for the adventure to force and compel it through an adventure path. It's unclear if WotC provides campaign length scripped adventure paths as a means of cramming long campaign into a specific page count, because it's the form their designers are most familiar with, or because they believe it best fits the play style of Dungeon & Dragons 5th edition.  It's not a necessity however, other ways of playing are possible and also rewarding to player and GM.

The introductory material to Descent is even rich enough with factions to encourage and respond to player choice, motivation and action. The gazetteer of the entire city, complete with random encounters, in the first Appendix even more so. Baldur's Gate has several factions with competing goals: a mercenary military, venal merchant rulers, refugees and their disgraced paladin leaders, cults and regular citizens - all of whom the adventure gives basic motivations and goals, and all of whom are forced to act by the threat of hell's dominion. Some of these motivations and plots are twined into the Baldur’s Gate character creation Appendix, but like the gazetteer they are disconnected from the larger adventure.  

Instead of using the world building information it includes, Descent  provides a rather rigid plot, a sometime clever and sometimes clumsy path through the intrigues of Baldur's Gate, using most of its factions as mere background and color but with one set of outcomes and one heroic ending.  This is a wasted opportunity, and the unwillingness to offer multiple hooks and faction alliances is especially befuddling given the wealth of detail in the appendixes, grey v. grey morality of the factions, and inherently suspect nature of Baldur's Gate. The adventure casts the character's as reluctant anti-heros compelled to align with brutal mercenaries, but doesn't allow players to explore options for betrayal, scheming and factional struggle?

Interweaving proper details about the leadership, goals, resources and schemes of various factions into the adventure rather than feeding a specific story of intrigue scene by scene and drip by drip to the players would do a great service to what is ultimately a creative premise, with compelling supporting material.  Specifically, there seems to be no reason why the characters must be impressed by Baldur's Gate's mercenary army as trouble shooters rather than refugees or agents of one of the city's ruling families. The character creation options even include these ideas, but playing the adventure as written working them into the adventure path would be difficult.

The tendency to eschew opportunities for player choice in favor of brute narrative force, despite background and story opportunities that easily offer it, is a repeating theme in Descent, perhaps to the point where it becomes a design principle of its own.  Distrust or disregard of players and the possibility that they may figure out ways of navigating situations beyond being compelled to engage in tactical combat scenes and a distrust of GMs to balance varied NPC and faction interests or respond to player action without a simple script.

There can even be an argument for this kind of design and GMing - a game where the vast majority of play is spent in tactical combat, and everything else needs to be pushed through quickly so that combat game time is maximized.  This is not the goal of classic play however, and it minimizes the big advantages that TTRPGs have over computer RPGs: a thinking referee who can adapt to player decisions and changes in the setting that go beyond prior scripting.  

STRONG ARM OF THE LAW - HOOKS VS. COERCION

Following from a general lack of choice and a design that clings, terrified perhaps, to a long simple script the actual "hook" of the adventure is forced conscription as agents of a brutish Baldur's Gate mercenary captain and root out a kill crazed cult.  While Descent makes it clear that the character's must take the mission provided by Captain Zodge, and even enforce this assignment with assassination squads to keep the party following the storyline, Descent fails to offer any meaningful reason why either Zodge wants amateurs doing his cult hunting or why he cares to waste resources to force unreliable agents to act or punish them when they don't. This is another wasted opportunity, or more likely a decision to remove all complexity and nuance in a scramble to reach a combat encounter and provide a straightforward story. 

The shame is that there's plenty happening in Baldur's Gate, and story of nobles plotting with devils and cults to drive the city to chaos and ultimately into hell has plenty of entry points for the party to get involved, and is one that should be important to multiple factions. Mercenaries, less diabolically inclined nobles, refugees, common townsfolk and even pirate or trade captains would all take action and each offers potential patrons.  There's no reason to use powerful squads of guards and flaming skulls to chase, threaten and force the party into a specific series of interactions with the cult.

A counterpoint to this is the tradition of the forceful introductory hook.  In open world campaigns the first hook is always the most difficult to sink into the players.  Characters are ill formed masses of backstory, or blank slates, and the players don’t know anything about the setting or adventure.  There’s literally too little information for the players to quickly begin adventuring because they don’t know enough to decide where to explore, who to talk to or what to beware of.  Pages or minutes of GM explanation and background are not the answer - they will be ignored and forgotten. Instead the GM should offer a simple and compelling problem that personally threatens the characters and needs immediate solution.  A lot of times this solution looks pretty much like forced narrative - and is: a shipwreck or the party imprisoned are classic campaign beginnings that offer no player control. Players don’t get to say that their PCs refuse to travel by water or picked a different ship.  They don’t get mechanical options for avoiding capture or succeeding at their crime. These sorts of initial hook are perfectly acceptable as the campaign needs to start somewhere, though they become less coherent the greater amount of background players start their characters with.  Descent doesn’t have one of these introductory hooks - it simply begins with a description of Baldur’s Gate, and forces the party to meet with a Mercenary who impresses them into service.

So far this is fine, the party was unlucky enough to be swept up in a situation of civil unrest and forced to work for a faction.  Presumably the Mercenaries are drafting other able bodied, or armed strangers as well. This is the initial problem - that the characters free choice and movement have been constrained by induction into a despotic police organization.  They have the option of relishing in their new found power over civilians, becoming agents of the mercenary junta or getting out. What if they simply throw away the badges they’ve been given? What if they actively seek out revolutionaries? Why are they given spy duties rather then forced into crowd control under the stern watch of a mercenary veteran or three.  The hook exerts control of player action for too long and both elides the actual problem that the players face (low status membership in the mercenaries) and the realities of that problem (newly drafted thugs don’t get to be independent secret agents and fixers). In other words, while a coercive hook might be reasonable and even helpful to start a campaign, it needs to be both simple and offer a direct problem.  The recruitment hook here doesn’t do that. Better to simply drop the party in Baldur’s Gate with a clear problem - destitution and hunger. Present a clear option (joining the Mercenaries) and a few others (working for other factions or becoming public heroes after stopping a cult massacre) that don’t force the players to deal with a problem (the kill cult) secondary to their personal one (being drafted into the mercenaries).    

While I don't propose turning Baldur's Gate into a full fledged urban setting, clear details on the factions, and a rumor table would allow a far more organic revelation of the noble/devil/cult plot.  Even a series of random daily happenings: pirate ship docks with disturbing news, refugee riot, paladin terrorism, mercenary crackdown, cult murder spree, or nobles dueling in the street would be a simple way to pull the character's into the orbit of the city's factions, earn their trust or dislike and lead deeper into the adventure.

It is possible to mechanize factional relationships in this sort of scenario with a simple point system. For example, the mercenaries will offer the party the mission (with a substantial reward) to destroy the cult once the party has gained 3 points of reputation with them. Points in turn can be gained by player action: turning in refugee leaders, stopping cult atrocities, or bribing mercenary leaders.  Each faction can offer this sort of point system, as long as many of the actions involved are contradictory - the mercenaries approve of turning in refugee paladins, the refugees will turn against the party for it. In a longer campaign such faction associations can also offer access to resources like special equipment, henchmen, new spells, information, or even troops - as long as the party stays in favor.

Beyond the uncharitable implications that WotC's designers simply dislike or are incapable of designing adventures with player choice, there may be a mechanical argument for Descent's forced narrative - level based gatekeeping/pacing of content.  Like most recent WotC products, Descent's story is firmly wedded to milestone leveling. Finish the first (tavern) encounter, go to level two, finish the second (cult sanctuary) and advance to level three.

This points to a concern about the 5E system that I've found in my own games - character power level increases rapidly and mirrors a rapid increase in the combat potency of monsters.  Encounters and areas designed to challenge 3rd level characters may quickly annihilate 1st level ones… but this concern isn't new, the same can be said for older editions. Combat in early editions tends to be deadly, yet combat balance was never systematized, as it is In 5th edition. In past posts All Dead Generations has covered the idea of level vs. encounter based design, and Descent provides an example of a downside of the first: providing a lack of opportunities to outwit, outmaneuver or otherwise overcome foes outside of direct combat means that the designer has to make sure that every foe can be beaten in direct combat while using complex encounter design to keep this combat from becoming a boring slog.

A cycle appears where the designer needs to assert more control over each encounter to avoid risking it being impossible or boring, which in turn means that to be interesting the encounter needs to appear or be riskier (risk that is appearance only, because it would be unfair to force players into combat that was unwinnable). This kind of encounter based design aims or ultimately succumbs to the urge to make every encounter a combat that taxes the characters' resources fully. Combat encounters that utterly drain the party also require recovery afterward and so each is a discrete scene in the adventure rather than an element of a larger level.  This is how one designs tactical combat games, but It's detrimental to exploration play and player choice because it encourages or requires a rigid structure.

Descent has fully internalized this sort of tactical combat centered design, and so to the classic play designer or GM it appears to waste opportunities for non-combat play and perversely cling to a forced narrative under the guise of play balance. Difficulty can be gauged and overcome in multiple ways. I'll discuss some of the more encounter specific mechanisms in the last part of this post, but choice and balance can be built in even at the distant level of overall adventure structure.

For the first section of Descent the basic structure is a trio of locations (cult headquarters, a ship converted into a bar/thieves' den, and a novel's mansion with attached dungeon). An introductory scene (and of course a combat encounter) in another bar provides a clue to the first of these locations. There is no reason that these locations need to be overcome in a specific order and even the text of Descent suggests this, allowing that a party may rush to attack the mansion after exploring the cult catacombs. Leading the players through a specific narrative and order of these encounters becomes even more obviously unnecessary when one recognizes that they are all headquarters for aspects of the same organization.  This fact also suggests way to assuage any fear that the difficulty of a specific location won't be properly gauged to character level progression.  

With the addition to the random encounter table (tables exist in the Appendix, with no guild to using them and largely without reference to the demonic plot), faction information/reputation rules (information is already in the appendix) it would be easy, efficient and helpful to add a timeline of how and when the noble/devil plot progresses.  Descent fails to do this - beholden to its forced march of a narrative it doesn't contemplate any other variety of player failure beyond the death of the party in combat. A classic adventure, at least a good one, has a different sensibility - one where the players actions shape the world but don't define it, and thus the possibility of failure (or player choice to aide the more sinister faction) doesn't end the adventure.  The game isn't lost if Baldur's Gate sinks into Avernus - the GM already has material for a campaign about a terrestrial city trapped in hell. Yes, the story changes, especially where the players abetted such a disaster - but as long as the GM has or can build enough content to decently respond to player defeat or unexpected events the campaign goes on.

In Descent, a timeline of the plot (including how setbacks like the destruction of the murder cult effect it) would be beneficial and allow for locations to be less static. If combat difficulty was a grave concern, the plot's progression could increase opposition within the locations that the plotters control, with cult recruits, mercenaries, or summoned devils appearing either in response to plot setbacks or simply time.

Descent doesn't want this sort of complexity, instead choosing to propel the plot along a single vector with only small variations with a series of sub-chapter scenes and small dungeons.  Below are examinations of the first two of these subchapters, again from the perspective of how they differ from or might be transformed into classic play. While some of these observations are critical, they aren't aimed at denigrating Descent or the play style it serves, only emphasizing how it differs from the classic play style and perhaps how the same content can be turned towards classic play.

DEN OF SCUM AND VILLAINY - SCENES AND COMBAT AS NARRATION

The opening subchapter of Descent into Avernus has the party newly inducted as agents of the

Descent Into Avernus Interior Art
A cult attack

Baldur's Gate's mercenary army.  Tasked with making contact and reviving a report from an intelligence source at the Elfsong Tavern. Instead the party will be drawn into a conflict between the source and the pirate crew she deserted from.  A classic bar fight caps the subchapter and at the end the source (or her spirit via necromancy) gives up the location of the cult catacombs.

That's how Descent wants this subchapter to work, and It's really the most likely outcome. The subchapter is a spare five pages, which may seem extensive for a single set piece combat and NPC interaction, but includes a full keyed description of the twelve location tavern.  The tavern and its denizens, even the stereotypically villainous pirates, have enough detail to be interesting and evocative - the ghostly elvish lament that manifests occasionally in the taverns for example. While the Forgotten Realms tendency to have social and material culture at wild variation is present - the tavern being the standard 18th century English variety inexplicably used as a poor gambling den in a pirate freeport - the setting material is sound.

Yet, the subchapter is a bit of a mess when it comes to design.  Mess specifically related to how Descent treated keyed locations. Simply, there is no reason to make the Elfsong Tavern a fully keyed location.  Meeting with the mercenary spy and dealing with her pirate foes is a scene, and it doesn't require a detailed description of the taverns rooms. Yet this is what Descent provides, and in doing so it makes this scene unnecessarily long and confusing.  

Classic play focuses on dungeon crawling, but this doesn't make it antithetical to scene based design. Traditionally wilderness encounters, abstracted from the dungeon's constraints and tactical challenges of halls, room and walls have existed as scene - either taking place only in the theater of the mind (“The road is blocked by fallen logs, a cliff to the left, and dense forest to the right - before you stands a man in dirty mail and a peaked green cap. He shouts “”Stand and Deliver, my lads have you dead to rights””) or a quickly sketched battlemap.  The encounter in the Elfsong is the same, a single combat/role-playing obstacle. Excessive detail is unnecessary, though a map of the area can be useful.  

The map of the tavern alone - perhaps with notations about room use: "main bar", "gambling den" or "private dining" is sufficient spatial information to handle tactics (window escapes, donor's barricaded, missed attacks smashing patrons and beer barrels). The goal here is to run a tense negotiation and/or a rollicking bar brawl. How does one run a rollicking bar brawl?  As scenes go it's both easily comprehensible, classic maybe and a way of introducing both a key faction (The diabolic plotters) and a minor faction (a shipload of pirates). The departure from classic gaming, or at least a more open sandbox Baldur's Gate is largely one of framing. Where does the bar brawl scene start? What are the options for the players? What ways can they reach their goals?

As written the party explores the bar (though this begs the question of why? They need only ask after the agent), meets the informer, is cajoled into protecting her from a band of pirate thugs and then gets their information if they succeed, or also if they fail and bring the informers body to their Mercenary patrons.  Not much of this sounds like a good bar brawl? There's a fight with pirate thugs, who have tough guy manners and halitosis, but it's just a set piece D&D encounter… the one decent addition being the ability of both pirates and players to hire other patrons to join the fight as it progresses. Even if the goal here is to recreate a bar brawl scene, a mainstay of the Western genre for generations, a skirmish action with drawn swords and deadly magic will not do this. If the designer is less fixated on a singular vision for the scene, more open to clever solutions then the frame of the scene needs to include more than just the fight.

The page space used to provide detail about the tavern could be better used to offer options for the Elfsong Tavern incident. Perhaps the informer will suggest drinking with her pirate hunters, and a subsystem or mini-game can be presented that risks/offers drunkenness and reduced combat effectiveness for both pirates and party based on how long they can keep the informer and pirates negotiating.  Likewise a system for a non-lethal brawl with unwritten rules that gain or lose the support of other patrons (generally pulling a weapon means someone will try to hit the offender with a chair to keep the outright murder to a minimum). Suggestions or simple subsystems for sneaking out of the tavern (though the map is useful here - it just doesn't need keys).

Descent makes some efforts in this direction, giving personality sketches for NPCs in the Elfsong, but ultimately these details remain shadow suggestions or unsupported possibilities because Descent encourages a simple set-piece combat solution.  Classic play usually allows for direct combat to overcome obstacles, but the mechanics of older editions and an ethos that elevates clever player solutions and puzzle solving tends to disfavor them. In the Elfsong Tavern There's little space for negotiation and trickery, and because combat is designed as a set piece, only a single option (hiring patrons) presented for using strategic thinking to bend the combat to the players' advantage.

At the end of the scene in the Elfsong Tavern the players will have the location of the murder cult they have been tasked to elminate. The only potential flaw is that this information isn’t something that needs to have come from the informer (or the informer’s corpse - the possibility of using speak with dead to get the information is a great inclusion by WotC and indicative of the sort of attention and quality that Descent generally shows), background suggests that there might be other ways of gaining the cult’s location - following its members back to their lair after an outrage comes immediately to mind, but other informants, stakeouts, review of records regarding food and evil incense deliveries or whatever other amateur sleuth schemes the players concoct would also work in a more openly designed adventure.  An additional inclusion in Descent, and one that also shows a nuanced and mature set of ideas about adventure design, is the possibility that the players may decide to find and seize the pirate’s ship.     

I don’t want to say that a good GM can't or won’t improvise a variety of solutions to the Elfsong scene, but a good product shouldn't just depend on the possibility of good GMs it should offer aides, ideas and support. When offering complex scenes more designers and products should supply worksheets, charts and checklists that help track events. The suggestions above about running a faction and rumor rich city would also benefit from aides aimed directly at the GM.  This becomes even more important in an introductory scenario where part of the product’s goal is to teach new players and GMs that they have more options then direct combat - charts and checklists can not only suggest varied solutions to problems but help GMs run them, including combat set-pieces. I remember a mid TSR era module - 1986's “Treasure Hunt” - that provided GM handouts with checkboxes for every enemy's hit points. This made running it's scene-based battles (PCs v. Pirate group 1 v. Pirate group 2) easier and emphasized the number of combatants, while also highlighting their scavengable equipment. GM aides of this sort are precisely the kind of added value that a published adventure can provide, helping new or overburdened GMs track event clocks, faction reputations, timekeeping, random encounters, NPC relationships and complex combats. WotC would do well to include them, especially if it's focused on publishing scene based paths.

A NOT DUNGEON - SCENE MASQUERADING AS SPACE

It's a serviceable map - wonderfully drawn

The purpose of this blog is to discuss how to play and run classic table top games, or perhaps how to emulate that experience in more contemporary systems - this means an intense focus on how to design and run dungeon crawls.  Descent includes three or four subchapters that can be described as loosely as “Dungeons”, two of which are in the first chapter. Largely the adventure focuses on separate encounters - a design tendency in 5th edition, little climactic combats in locations that are at best Lairs (locations keyed with 1 - 10 rooms and without significant interactivity beyond a main encounter).  These sorts of scenes are much like the Elfsong Tavern rumble, though of course inventively varied.

When it comes to Dungeons (which All Dead Generations loosely categorizes both as a general kind of location based exploration adventure and a location with  10 - 80 keyed rooms) Descent seems fairly confused, but also makes a lot of moves in the direction of classic dungeon design. The first Dungeon, a 33 key location on two levels is a partially flooded cult lair beneath a fancy bathhouse, provides examples of how Descent handles dungeon design.

Like much of Descent into Avernus, on first appearance the bathhouse, or Dungeon of the Dead Three, is well designed.  The overall concept of an expansive underground lair beneath a mundane building is excellent for an urban cult lair, a hidden sanctuary. Likewise the guard rotation on the upper level (cultist enemies only at night) makes sense.  Many other decisions: overall dungeon map design, the use of empty rooms, the inclusion of environmental effects (support beams and collapsable ceilings) and fidelity to theme are also admirable … but the Dungeon of the Dead Three is simply not a dungeon in the classic sense and avoids many of the elements that make for dungeon crawl or exploration play, obviating most of these positive points.  Empty rooms do little without timekeeping, supply, and random encounter considerations and map design means a lot less when all encounters within the dungeon are set-pieces/lair encounters. Descent offers a series of effectively linear combat scenes rather than an actual level based dungeon for players to explore. From a design principle perspective Dungeon of the Dead Three fails to include Timekeeping, Supply and Threat in classic ways. It simply doesn’t engage with the first two and it approaches Threat from an encounter design model that is nonsensical in a dungeon, let alone one occupied by an intelligent faction.

Threat:  There are no random encounters in the Dungeon of the Dead Three, it is a “monster zoo” of the enemies waiting in their lairs for the players to barge in and kill them individually.  Illusionism and some creativity are used to make these individual encounters interesting and dynamic, including a “guard” scenario where a lone guard summons four others from a nearby room, but largely the cultists fail entirely to act like an organized foe, meaning tactics beyond frontal assault are unavailable or at least unsupported.  The lack of random encounters and static dungeon structure are again a standard for 5th edition because of its emphasis on character rest between combats, multi-hour “short rest” are a part of an encounter and combat system that seeks to make almost every encounter an epic struggle with victory dependent on heavy use of character resources.  Obviously a random encounter system (or a more complex patrol system - something entirely appropriate here) would disrupt this rest and encounter structure, so it has been discarded. This is not necessary, and random encounters are a primary method of putting pressure on the party to explore quickly, cautiously and with a specific goal. They give meaning (as they make the time spent in the dungeon and distance from exists important) to the map of the dungeon and its empty rooms, because the dungeon’s spaces not only determine how risky escape from the dungeon is but offer opportunity for tactical decisions.  Limiting a dungeon to set or lair encounters makes its spaces meaningless, not only allowing rests that make each encounter discrete, but effectively making any exploration between encounters a waste of play time, with no effect on the game.

Descent’s individual encounters are also highly staged, and based entirely on narrative and difficulty rather than internal consistency within the dungeon or some kind of dungeon ecology.  Monsters stay in their areas, and when encountered vignettes occur. Almost as if each lair has a little “cut-scene” waiting for the PCs to appear as a way of showing that the Dungeon of the Dead Three is not simply a series of combat challenges strung together.  It is however, these vignettes offer almost no interactivity for the GM or players. For example, a likely first encounter within the dungeon proper (there are straightforward guards above in the bathhouse - but only at night) is with a group of cult acolytes where “any sign of the characters approaching lets them set up an ambush of sorts.”  The acolytes play dead and attack anyone entering their chamber. This is a perfectly fine ruse, and it’s good to see some mention of enemy tactics, but it begs the question of what actions will alert them? Without further indication, and an idea of what the Acolytes will do if they don’t detect the party’s approach, this ambush becomes more or less automatic.  The issue is compounded because there is no mention of the light conditions within the Dungeon of the Dead Three - though many of its residents can see in the dark (as can the majority of 5th edition characters). Light and noise are the obvious alarms, but it’s entirely unclear what the likelihood of the cultists detecting the party noise or light, and it’s unfair for a GM suddenly punish players for failing to pay attention to a risk that has previously been excluded from the game.  Much like a GM immediately deciding that the characters are all starving because players failed to mention that they were eating during down time between session, even the sudden adoption of logical new rules and consequences can be a form of GM antagonism when they aren’t made clear before hand. In this case the wholesale use of surprise rules, light rules, encumbrance and the framework of classic dungeoneering, including and reaction rolls (how do the cultists know they should ambush anyone they hear coming towards their mediation room) can help.

Of course one can’t simply dismiss odd encounters like the seemingly automatic cultist ‘ambush’ as purely poor design - too much good design exists in Descent for that.  These encounters seek to do something, they aim for a particular style of play, the creation of what I think of as “moments”. If a scene is a more general approach to design focused on stringing the adventure together as a story rather than a location and creating narrative coherence rather then spatial, the moment is a miniature sort of scene, a brief experience of something the designer thinks is impressive, exciting or evocative.  Corpses suddenly revealed to be flail wielding death cultists, leaping up to attack the adventurers, the party interrupting a scene of torture while the victim can still be rescued, or chancing upon a falling out among the diabolical plotters - all are moments. Moments aren’t a bad thing necessarily, but they exist dependent on the party stumbling on them. The cultists wait eternally playing dead, the torture is endlessly poised before its lethal climax, and the backstabbing breakdown of trust among enemies is paused at a specific point until the party bursts in.  All of these incidents become predetermined bits of imagery for the players to passively consume, rather then part of a coherent spatial and thematic whole - the dungeon level.

These petrified moments are antithetical to the idea of a living, responsive dungeon and the idea of threat, factional conflict and the monsters that adapt and react to the party.  While it’s always neat to have one’s encounters doing something if the party comes across them unaware, when the mis en scene of the encounter overtakes the monster as something with its own goals, plans and ability to respond or adapt a great deal of flexibility is lost and every encounter becomes more of a discrete static moment.  If the party flees from the cultist’s playing dead what will they do? If they return later will the cultists try the same ruse? The over reliance on creating a dramatic moment pushes the players into interacting with it in one way only, and leaves the GM with fewer tools to run the location.

The Dungeon of the Dead Three is not a simple ruin, overrun with monstrous vermin or the unquiet dead - it’s the headquarters of a kill cult, filled with fanatical diabolists.  However insane they may be, the cultists are an organized foe, and because of its over reliance on the drama of moments Descent doesn’t treat them as such. In the context of classic location design, The Dungeon of the Dead Three is not an exploration dungeon like “B1 - Search of the Unknown”, or a trap dungeon like “S1 - Tomb of Horrors”, it’s a siege dungeon like “G1 - Steading of the Hill Giant Chief”.  The party is infiltrating or storming the home of an organized faction of monsters how should logically defend each other in an organized fashion. The cultists in Descent do not, despite the implication that there’s an organized structure, and even a moment where a single cult guard bangs on a door to draw his follows out in defense, but there’s no indication about how the cult will know that they have been attacked in their home, how they will respond or what resources they can bring to bear.

With siege dungeons there’s usually two phases of the adventure - first an infiltration, the party trying to sneak past defenses, and a tense exploration of the location, often with efforts at silent assassination.  When this inevitably fails and the monsters raise the alarm the scenario shifts as the players gauge their ability to win a direct confrontation and the monsters follow a defense plan, usually culminating in an effort to hunt down and kill, capture or expel the intruders.  For this to work there’s a set of specific GM tools that help:

• An order of battle - detailing the total number of monsters available to the defenders, their leadership and how they will respond to setbacks and loses. 
• Tactics - one or more sets of ideas about how exactly the monsters protect their lair (frequency of patrols, warning systems, responses to invasion, special traps or plans they set into motion). 
• Notes - on the goals, psychology, resources and potential reinforcements the defenders have. Changes they might make to the dungeon after an initial attack or what they will do with captured enemies.
  

All of these resources aim at creating a location and faction within it that the players can approach in multiple ways, direct attack, infiltration, guerrilla warfare, negotiation or trickery.  They also allow for degrees of success. The party may succeed in clearing the dungeon, die and be replaced by a new set of characters, flee and return multiple times, be captured and escape, allow some of the cultists to escape or even change sides and decide to work with the cult.   All of this rich possibility is abandoned in service of providing a few cool moments before or during more complex, varied, direct combat encounters.

The unfortunate part is that this isn’t even necessary to create compelling narrative.  Noting that the cult acolytes use the tactic of playing dead (presumably they are good at it in some special way) is likely to create the same effect as forcing the tactic into a specific encounter - at some point one the GM will get to describe a group of robed “corpses” suddenly looming from the flooded floor, fling open their rags to reveal skull headed flails and attack screaming foul prayers. Even the largest of moments provided in the Dungeon of the Dead Three is easily converted into an open ended note available for a smart GM to apply when appropriate. 

In the dungeon’s climactic encounter, scene or moment the party comes upon an ongoing battle between one of the noble conspirators and a cult assassin.  Lesser dead cultists are scattered about and the assassin flees allowing the party to begin negotiations with the plotter who has decided to give up the cause and flee town.  This moment is presented both as a cinematic detail, and as a way to give detailed instructions to the party about the nature of the plot, the next locations to explore and the goals of the plotters.  However, much of the same could be accomplished, without relying on a specific set encounter, simply by including the plotter and assassin in the order of battle as cult leaders, noting that the assassin distrusts the plotter and wishes to kill him, especially if he feels the cult has been betrayed, and that the plotter will give up the plot and flee town if given the opportunity.  The same situation is entirely likely to occur, but it’s not tied to any one key, allowing a more organic situation to unfold based on player choices within the Dungeon of the Dead Three.

This sort of approach to dungeon encounters of course depends on the designer trusting the GM to apply open ended situations that allow a story to emerge from play rather than providing a specific story that the GM replicates and every table.  It also requires trusting the players to solve open ended problems without being carefully led through them, and it asks that designers give up control of their creations to each table of players to experience differently. The possibility of failure, or sub-optimal choices with unexpected complications and consequences is introduced by redesigning the Dungeon of the Dead Three with classic play principles, and that’s just a difference between them and the path style adventures that WotC currently offers for 5th Edition.  Knowing the differences, spotting them and making a conscious decision about what sort of game and what sort of playstyle one wants to run however is a useful choice - and one that need not be total - running the Dungeon of the Dead Three in a classic manner, while retaining the path structure of other parts of Descent is entirely possible.

Timekeeping & Supply: Beyond the issues of threat, The Dungeon of the Dead Three makes some choices that eliminate the other two elements of the classic dungeon crawl, timekeeping and supply.  While a GM might still track light sources, time, encumbrance and equipment there’s no mechanics tied to these things that makes them interesting or threatens them in The Dungeon of the Dead Three.  As noted here in the past, the removal of character resources other then hit points makes every problem and every risk one that must deplete hit points to have meaning. When risk advances on multiple axis and through multiple metrics it’s far more interesting for the players to balance them, conserving what’s depleted and spending what’s plentiful - which means approaching obstacles in different ways.  Dungeon of the Dead Three even presents obstacles beyond lair encounter: several traps (yellow mold, an animated set of gauntlets, an empty trapped sarcophagus, and an altar that raises skeletons) and a pair of secret doors, though it doesn’t do so especially creatively.

The traps all require intentional player interaction (in the case of the altar reciting an ominous phrase written on it), aren’t obstacles and offer no reward for circumventing them. For example the sarcophagus is empty - meaning the best way to respond to triggering it is simply to retreat and close the door.  In each case the lesson these traps teach is to ignore dungeon dressing and proceed quickly to the set-piece combats. Nor are any especially tempting, asking players to overcome caution for obvious rewards (mechanical rewards in 5th edition are limited to either XP from combat or reaching pre-scripted milestones, so this may be of limited effect even if traps protect treasure), and none offer a faster or better way to better navigate the dungeon (also of dubious utility where time taken to explore has no meaning).  Still the traps themselves are enjoyable enough - the spectral axe over the tomb of a screaming barbarian is especially fun and they don’t detract from the adventure, only fail to offer variety from the combat challenges.

Two secret doors provide actual obstacles, but do so in a way that is widely disliked in classic play circles (though common enough in old adventures).  The secret doors in The Dungeon of the Dead Three act as obstacles to progress - they must be discovered to reach important dungeon areas, and indeed are the only way to access those areas.  This means that should players have difficulty finding them they can’t venture deeper into the location - a problem compounded by the linear, sequential nature of Descent’s story. The way Descent approaches secret doors is also mechanical - few details about the door’s construction and trigger methods are included, making searching for them a pure question of successful die rolling - something that increases the risk they will never be discovered.  Secret doors work best when they conceal treasures or shortcuts, not as hard obstacles which must be overcome to complete an adventure, and they fell best in play when they can be discovered with either a mechanical search roll, or clever player interrogation of the environment. To do this well the designer has to present more detail about the doors: triggers, how well detection methods like tapping or blowing flour into cracks work, and characteristics of the door more generally - Descent doesn’t.

As an adventure The Dungeon of the Dead Three may be a fun play experience, but as a dungeon - a place for exploration and risk v. reward based play it fails.  It’s simply a series of evocative combat encounters written onto a map with little concern for how the players will navigate the space and none for how the denizens of the dungeon relate to each other, the characters or the larger adventure.  One could transplant it into a flowchart with ease, or perhaps even a linear “wave” style prolonged combat challenge, and this represents a loss of what makes classic dungeon crawl play fun and exciting.

CONCLUSION

There are some harsh words for Descent into Avernus in this essay, but that isn't to say it's a bad adventure. Mistakes and shortcuts are inevitable in a 200 plus page adventure of epic scope written by a committee, and despite a few, as well as an unnecessarily fearful approach to player choice and GM ability, Descent has evocative content and scope.  There are also many great design decisions - even in the first few subchapters I’ve looked closely at that are counter to classic play. Yet they are design decisions, conscious design decisions to eschew exploration, center cinematic narrative and evocative moments over player choice, and efforts to assure a consistent plot across every table that plays Descent.  As much as I disagree with these choices, they seem to have been made intentionally, and making them out to be the sign of a bad adventure is simply elevating my particular play style above others.


Descent into Avernus is first of all not a bad adventure, it’s basic plot and the ideas behind most of its locations are compelling, interesting and made me want to run it.  Many of the details and characters within are likewise intriguing and far more nuanced then many WotC adventures. The stakes feel epic enough for the scope of the adventure and the art and map making within are extremely good.  It’s not a classic adventure however, and it’s a shame that it doesn’t take more from that well of 1970’s and 1980’s adventure knowledge to move beyond the constraints of the adventure path and match its epic scope with actual opportunities for player decisions that matter.