Dragna's Blog: Tsolenka Pass, Travel Encounters, and the Reference Frames of Chult

As you may (or may not) already know, I’m in the process of planning a Tomb of Annihilation-inspired campaign for my players. The players of my first-ever Curse of Strahd campaign gave me the module as a gift when we finished it, and I’ve been in love with Chult as a setting ever since. There’s just one problem with it.

I hate the Tomb, and I hate the Annihilation.

Tomb of Annihilation

For those who aren’t familiar with it, Tomb of Annihilation is a level 1-11 adventure campaign divided into four basic parts:

• In the first “act,” the players arrive in the continent of Chult to investigate a “death curse,” which has caused resurrected people to suddenly start wasting away and dying. Upon their arrival in Port Nyanzarou, the players have to get their bearings, secure a guide, and obtain survival gear in preparation for their efforts to find the cause of the death curse.

• In the second “act,” the players traverse a massive hexcrawl in order to find the source of the death curse - which, they eventually learn, lies in the lost city of Omu. Although the players can obtain a map of the borders of Chult, the interior of the continent is uncharted, requiring them to trawl through (up to) eight thousand square miles’ worth of jungle in order to find their ultimate destination.

• In the third “act,” the players arrive in Omu and must deal with a yuan-ti cult at the Fane of the Night Serpent, while also obtaining relics of nine trickster gods in order to progress to the next act.

• In the fourth “act,” the players descend into the eponymous tomb: a multi-level megadungeon created by the lich Acererak. Upon reaching the base of the tomb, the players must defeat Acererak and the atropal he’s feeding there in order to lift the death curse and free the souls that Acererak has captured.

The vast majority of the adventure is consumed by acts two and four—the hexcrawl and the megadungeon. The hexcrawl, in particular, is among the purest definitions of a “sandbox”: it’s an assortment of a few dozen points of interest (each one with no more than four to eight “rooms” within it) interspersed with regular random encounters, weather effects, and other wilderness hazards.

I have numerous problems with this setup.

Let’s start with the second half of the title: the Annihilation. The stakes of “there are approximately sixty people you don’t know or care about and they’re all dying of a curse that statistically only affects powerful adventurers” are, themselves, not very compelling. There’s no core stakes here—it’s a picture-perfect example of an excuse plot.

Furthermore, as with the Mysterious Visitors hook in Curse of Strahd, it makes absolutely zero sense for the Plot to ask the players—a group of random first-level PCs—to go on this big, dangerous, important, continent-spanning Adventure. It breaks all sense of verisimilitude to believe that this party will just so happen to succeed where all others have failed (and have I mentioned that the adventure makes basically no mention of the other adventuring parties it insists have already attempted the quest?).

There’s also a massive gulf between the initial hook and the eventual climax. Unlike Curse of Strahd, where the players are constantly exposed to the harms that Strahd has caused and the importance of defeating him, Tomb of Annihilation completely ignores its hook for all fifty-odd sessions’ worth of content between the first and last session. I would not be surprised if a majority of players forget why they’re even in Chult by the time they learn where Omu is.

What about the first half of the title—the Tomb? The Tomb is ridiculous; it’s a complete hard left turn from the entire rest of the module. The players have spent six or seven levels happily romping around a vast, wide-open continent, building relationships with each other or the NPCs they encounter, and now they’re stuck underground in a single location for the next four levels.

Did I mention, by the way, that the Tomb is an unfair meatgrinder? It’s honestly kind of laughable that the module presents an optional “meatgrinder” variant rule; there’s genuinely no need for it with a dungeon as deadly as this. I would bet money that most parties don’t wind up with even half of the same members they started with by the time they reach the end of the Tomb—and if that’s the case, then why even bother telling a story about these characters at all?

Oh, right—because it’s an Excuse Plot, and we don’t actually care about these characters or the story that they’re telling. We just want Cool Megadungeon Traps.

That might be someone’s cup of tea. But it’s definitely not mine.

The Lost Treasure of Omu

With both the Tomb and the Annihilation removed from Tomb of Annihilation, what’s left to run an adventure with? The setting of Chult itself: the entire continent (and the accompanying continent-spanning exploration narrative) that fascinated me in the first place.

But what story to tell? After sorting through a number of possible concepts, I settled on an original concept I decided to call The Lost Treasure of Omu, in which the players (the apprentices of a retired archaeologist) are commissioned by their mentor to retrieve a jeweled dinosaur egg when it’s stolen by the Zhentarim and brought to the continent of Chult. The mentor, who stole the egg from its parent when she was younger, more reckless, and more selfish, can share that the egg is no ordinary egg, but the offspring of the near-immortal King of Feathers, a magical tyrannosaurus rex and the “king of the dinosaurs.” 

Once the egg is obtained, the players must return it to its parent in the long-lost city of Omu, where it must be placed atop the Ziggurat of the Sun at the hour of Ubtao’s Shadow—a total solar eclipse above Chult that occurs once a millennium. If they fail to do so, the egg will fail to hatch—and the line of the King of Feathers will be broken.

Meanwhile? The Zhentarim are scrambling to get the egg back, the Red Wizards of Thay are trying to steal it to obtain its magic for themselves, and every faction on the surface of Chult is interested in breaking the egg to use its legendary wish-granting power for their own purposes. That’s even without mentioning the yuan-ti of the Fane of the Night Serpent, which have captured the King of Feathers and seek to use the egg in a dark ritual to release Dendar the Night Serpent from its prison beneath the Peaks of Flame. Can the players retrieve the egg and escort it safely to Omu where it can hatch—or will the Crown Prince of the Dinosaurs fail to see the light of day?

We’ve got a real set of stakes here—the restoration of a family, the redemption of the mentor, the thwarting of the villains’ plans—and a proper deadline with the coming of the solar eclipse. Who wouldn’t want to go on a mission to save and hatch a cute magical baby t-rex?

So now what? Well, we’re starting with a basic two-act structure:

• In act one, the players need to steal back the egg from the Zhentarim.

• In act two, the players need to get the egg to Omu.

(We can expand this further—in my specific case, act two is itself broken down into three smaller acts—but this is the basic skeleton of what we’re looking at.)

Act One is pretty easy to map out. The players get to Port Nyanzarou in Chult (the only major settlement), use politics/guile/investigation to track down the Zhentarim, and figure out a way to get the egg back. We can run this as a pretty standard urban fantasy mystery/heist combo. 

What about Act Two?

Hexcrawls Aren’t Fun

How do we handle the exploration portion of Act Two? When I started developing this idea, I had two options:

• Use the RAW hexcrawl

• Use a linear narrative structure (i.e., the players go to Location X, where an NPC tells them to go to Location Y, where an NPC tells them to go to Location Z . . . )

I immediately rejected Option #1 for the simple reason that hexcrawls rely on content for the sake of content. Between Random Landmark #43 and Random Encounter #198, the entire campaign just becomes this giant mess of unimportant bullshit that has absolutely nothing to do with the narrative whatsoever. (I recognize that this is a controversial statement and will probably make me a significant number of enemies. Nonetheless, it is a hill that I will proudly die on.)

That left me with Option #2: the linear narrativecrawl. I experimented with this for a while—even got as far as the outlining stage—until I grew dissatisfied with it. It wasn’t quite as bad as Descent Into Avernus’s “infinite fetch-quest,” but the Zeno’s Paradox effect of “repeatedly finishing a mini-quest, then being told you can’t finish the Real Quest until you do another mini-quest first” felt like it’d get a bit grating.

On top of that, it just felt like a waste. Here, I had this big, beautiful expanse of uncharted wilderness—and I was planning on walking my players through it on a leash? After writing about method dungeons and the importance of emergent narratives, I felt like they deserved better.

As I was exploring these ideas, I was simultaneously developing my conceptualization of verbs—the categorization and application of the actions that players can actually take in the game (i.e., the source of gameplay). I iterated through several possible frameworks and possible categorizations (e.g., “communication verbs” vs. “interaction verbs”), none of which seemed to have any real utility, until, one day, while rambling aloud to myself in the car, I said the following simple statement:

“There are only two kinds of verbs: verbs that change your reference frame, and verbs that let you act in your current reference frame”

Something about this rang true—so I decided to interrogate it.

Every Frame a Reference

What is a reference frame? It’s your frame of reference (very helpful, I know). It’s your location relative to something else. I’m outside the tavern. I’m sitting at the bar. I’m standing by the window. I’m on my bed. Everywhere is (or can be, or should be) a reference frame.

If you’ve ever played a text-based or point-and-click adventure game, you’re already familiar with the concept of reference frames: You’re standing in a particular part of a particular room, and there are a whole bunch of things that you can interact with. Some of them (e.g., doors) move you to a different reference frame; some of them (e.g., a bookshelf) move you to a different reference frame within the current reference frame; others (e.g., a letter lying on the floor) simply allow you to interact with them directly.

In D&D terms, what does it mean to change a reference frame? It might sound like:

• “I exit the tavern.”

• “I want to get a closer look at the bookshelf.”

• “I want to peek out of the window.”

Now, reference frames exist in every kind of game. Text-based and point-and-click adventure games have what I’ll call “discrete” reference frames: You can be by the bookshelf or by the window, but not in between. By contrast, free-roam games like Mario, Skyrim, or Baldur’s Gate 3 allow continuous reference frames: You might be three feet and six inches away from the bookshelf, and at a seventy-nine degree angle from the window, without necessarily having either within your reach.

One of the most powerful properties of free-roam games is to allow for the existence of any arbitrary continuous reference frame. If you’re wandering the woods in Skyrim, there might be dozens or even hundreds of trees around you. A text-based adventure game can’t possibly allow for a meaningful reference frame relative to each one of those hundreds of trees. However, by leveraging the player’s visual abilities and spatial awareness, a free-roam game like Skyrim makes the individual trees meaningful as distinct reference points for the player’s position, both relative to the trees themselves and relative to the forest as a whole.

D&D, however—like a text-based adventure game and unlike Skyrim—generally doesn’t have discrete reference frames. You can’t use spatial awareness to explore a forest because there is no spatial concept of a forest—only a qualitative one. In Skyrim, there’s a reference frame for every possible place you can stand in the forest; in D&D, a player can only possess a discrete position relative to some specific object that the DM has chosen to highlight. (The only real exception is battlemap combat, where the players’ positions are all relative to their enemies, allowing for meaningful navigation even in a white-room environment.)

This is bad news for wilderness exploration in D&D, because: (1) reference frames form the foundation of D&D gameplay, but (2) “a big, open forest” has no reference frames! At most, we can allow our players to make Survival checks to try and navigate the forest in a skill challenge-esque fashion. That, however, is an exception that proves the rule: in order to create meaningful gameplay, we’ve abandoned the concrete reality of the forest for an abstract, symbolic space where the “forest” is less a physical entity than a property of the game. We’re no longer exploring the forest; we’re playing the dice-rolling minigame “Explore the Forest™.”

Hexcrawls sort of resolve this problem, in that they provide some kind of reference frame to relate our exploration back to: the players might be in Hex A6, or in Hex K14, but they’re definitely not in Hex C7. This works to an extent, but ultimately falters because the hexes, both from the player’s perspective and (generally) the DM’s, are fungible: unless the DM has specific notes for either, Forest Hex A6 is basically exactly the same as Forest Hex A7, which is exactly the same as Forest Hexes B2 through B13. And when reference frames become fungible (i.e., non-unique), they fail to serve as meaningful reference frames anymore.

Dungeons of Dungeons

So: If we want to have meaningful gameplay, we need meaningful reference frames. And if we want meaningful reference frames, we need unique, non-fungible, non-arbitrary locations that the players can act in and around. Let’s call them “rooms.” In Chult, this might be a shrine, a village, a fort, a waterfall, or a bay.

Now, remember: There are two kinds of verbs—those that let you change reference frames, and those that allow you to interact within your current reference frame. Let’s call the first category “movement verbs,” and call the second “action verbs.”

Importantly, movement verbs are discrete—if I say I move over toward the window, then, barring interference, the game engine of D&D functionally “teleports” me over toward the window; although the Dungeon Master might narrate the process of movement, I functionally don’t have any opportunity to interact with the in-between places. So when our players move from one reference frame to a connected reference frame (e.g., from a tavern’s entrance to the tavern’s bar), they’re effectively stepping between them instantaneously; the space between is not continuous, but discrete. Let’s call these connections between reference frames “doors.”

So now we’ve got a bunch of “rooms,” connected to specific other “rooms” through “doors,” with each room having a unique layout/environment/etc. What does this sound like? If you guessed “a dungeon,” you’d be correct!

Put simply: For exploration to work in D&D, we must model each potential point of interest or landmark as a dungeon “room,” and each distance between two points of interest as a dungeon “door”—to be instantly traversed (via monologue narration) unless we add an additional dungeon “room” in between them.

To implement this, I took a map of Chult from Tomb of Annihilation and marked every landmark (i.e., “room”) with a yellow box. I then marked every expected traversable path between two landmarks (i.e., “door”) with a blue box, so that I could see the connections between them at a glance. This is what I ended up with:

Then, to make it even easier to visualize how all of these landmarks connected together, I used DungeonDraft to build a “dungeon” version of Chult, with each landmark literally represented as a room:

Now, you can clearly see how everything connects together, as well as how the players’ journey might ultimately lead them to their destinations (the black- and green-colored rooms).

On Mount Ghakis

Interestingly, even though it’s not a true “exploration” sequence (since it’s largely a linear sequence of encounters), we can still apply this concept of “reference frames” to the players’ journey through Tsolenka Pass in Curse of Strahd: Reloaded.

Those who have read the Working Document for Arc R: Trials of the Mountain might notice it already: Every single scene, whether it takes place in civilization or the wilderness, has a sense of place. That includes:

• The players’ recruitment of Kasimir (at his cottage)

• Vargas’s challenge and Strahd’s first trial (at a hunting path crossroad in the Svalich Woods)

• Strahd’s second trial (at any resting place the players might choose)

• The druids’ attack (at the Raven River Crossroads)

• The snow maidens’ attack (by a monument to the roc of Mt. Ghakis)

• The winter wolves’ attack (by the hidden crevasse)

Nowhere is this sense of place more important than the Tyrant’s Second Trial, which can occur anywhere the players choose to rest. In outlining that scene, I had to design seven different strategies that might take place across fifteen different possible resting places, including a possible resting place (Van Richten’s Tower) that, due to its unique properties, was only compatible with a single possible strategy (“The Horde”).

Put simply: Reference frames matter. They dictate what we can do, what we can interact with, and the stakes for both. Perhaps, then, it isn’t so surprising that, in a game titled Dungeons & Dragons, the whole world can be best modeled as a dungeon itself.