AAJ Reflection 1: What I Learned while Writing Zurth

01 Mar, 2026

I have always found creative writing daunting (thus all my circuitous efforts over 2024 to do anything at all except buckle down & write). But for my Antarctic Adventure Jam entry I have consciously tried to enlarge my comfort zone to include adventure/scenario writing. It' an amateur effort for sure, but I've been getting the reps in -- 79,000 words across 501 hexes, including a 450-room megadungeon.

Through trial-and-error, I found a few techniques particularly helpful & have compiled them here to consolidate & reflect on what I've learned.

Caveat emptor: One thing I learned while running AAJ is that creative processes are incredibly diverse. The items below should not be treated as unqualified advice, rather as a self-reflection of what worked for me. They may not have even produced a good setting, but at a minimum they helped me get past some deeply entrenched writers block. Also note that the goal here was to write & publish a big hexcrawl for its own sake -- please don't take this as advice about preparing a typical campaign.

Writing by sub-region

I used natural borders to divide my slice of Antarctica up into smaller sub-regions. Although I prefer 'bottom-up' worldbuilding (see below), it helps to have a few rough concepts to riff on for each locality. And on a practical level, it's easier to work with a handful of smaller documents.

I also found it helpful to stock each region with a different level of density, which I wrote about here.¹

Using (modified) Wolves hexfill prompts

I tried a few different procedures but landed on making my own modified version of Luke Gearing's tables, incorporating some of blark's adjustments & many more of my own. You can try it here.

There's a simplicity to the Wolves tables that I appreciate, particularly the clear partition into "Lair", "Settlement", & "Weird". For an OD&D setting, this is really all you need.

Using a Perchance generator helped. Although I like physical dice, I could not have written so much without the efficiency of having a new prompt ready at the click of a button.

The downside of the Wolves fills is that they are non-specific. "Relation to nearby hex" doesn't offer a lot if you haven't put a lot of detail into the surrounding hexes, and entries like "Strange tutor" are only easy if one has a pre-planned list of spells or fighting techniques (which I didn't at first, but built up over time). The next two points helped address this issue.

Anchor hexes

Any time I had a distinct idea for a hex, I just hand-placed it somewhere that seemed logical. (Probably a no-brainer for some, but there are others in the jam who went hex by hex with no problems). Alternately, sometimes I would "let the map speak to me" & invent a point of interest for a particularly interesting location (e.g. a lone mountain hex, peninsula, or forest clearing).

A corollary: rather than methodically generating prompts hex-by-hex or deciding up-front that I wanted to key, say, 1-in-6 hexes, I found it much easier to densify the map gradually, adding important hexes first & then semi-randomly filling in sparse areas until things felt stocked enough.

Spark words

I took many of the nouns & adjectives from the random tables I had compiled ages ago and mashed them into a spark table with over 1,000 entries. My hexfill generator appends two of these words to each hexfill. They're usually weird & abstruse, but I find it more fun to puzzle on a strange/surprising image than one that is thematically appropriate; the role of the sparks in my process is to push the results outside of what I'm already imagining. Examples:

• Princess Ranseur
• Skull Mercy
• Crystal Meteoric
• Mocking Crystal
• Squalid Appraiser
• Pit Infernal
• Valet Glory
• Library Benedictions
• Eidolon Sculptor

Obviously, some of these are quite hard to work with, but I think that even making the effort to incorporate the associations & imagery they inspired led to better descriptions.

But I have a strict rule (see next point): once the hex is written I erase the spark words, thus freeing me to evaluate the hex on its own terms during subsequent passes (below).

Bottom-up worldbuilding

I find that accounting for broad facts about a setting is stifling when writing specific location descriptions; I have a hard time proceeding from general to specific. It is much easier (for my brain, at least) to place whatever I want into a hex and then use abductive reasoning to develop a broader understanding of the setting after the fact.

In practice, this means that I didn't do much to develop factions & historical eras (see below) until very late in the writing process.

Jared Sinclair's dungeon design post is apropos:

But for the love of God, don't try to make it make sense. I repeat: Do Not Make It Make Sense. The players will either do that on their own, or they won't care enough for it to matter.

Multiple passes; wide then tall

I find it very, very difficult to write a detailed hex from scratch (doing this deepened my admiration for those who can). Often, a random hexfill prompt was useful for writing 1-5 sentences, after which I lost steam. The best remedy was to delete the prompt, forget about the hex for a while, & then return to freely expand & add detail based on whatever was written before.

One of the most fruitful writing periods was as I was migrating my draft from Typst to markdown (necessary for webmap compatibility) -- copy-pasting my work hex by hex helped me identify many areas to expand upon. Reading my own work is often uncomfortable, so big format changes can be a good way to force it.

The difficulty I ran into here was in balancing the effort of adding new hexes against that of deepening existing hexes. In retrospect, I probably should have keyed a smaller area & left myself more time to build on the hexes I had written.

Beginning with prose

Regardless of whether prose, bullet points, or a hybrid is better for the final version of a key, I find it helpful to start by describing what I'm imaging in natural language - just say what's there, then organize it by whatever schema you're using.

Writing one type of feature at a time

I found it helpful to focus each writing session on one specific task, e.g. keying dungeons, filling out weather tables, describing villages, rolling up NPCs, etc. The benefit is twofold: there's an efficiency gain from not having to swap between physical & mental tools (types of die, specific pages with random tables, but also headspace & types of cognitive task), and it can be helpful to take a break from certain kinds of writing to focus on others. As noted above, distance is often needed to get the creativity flowing again.

Reading other people's hexes

I tried to incorporate monsters, items, spells, & factions from the other jam participants (as behooves the jam organizer). These often provided a needed shot of inspiration and fodder for several new hexes (See Weaving in "threads" in one go below).

Naming every settlement

Inspired partly by Sam Sorenson's Legwork post, I decided to put a named settlement on every hex with 'farmland' terrain, even if I couldn't come up with anything else interesting to put there.² I used the Sword & Sorcery syllable-masher I wrote for this generator to produce villages & towns with ridiculous names like "Xoth-ralath", "Bar Udartophir", "Tov Pylos", or "Shophalbelath".

As goofy as it is, I found that doing this helped the region come to life. Almost every inhabited place on earth has been given a name by the people indigenous to it; while this can only be superficially replicated for an adventure game setting (no way was I making a conlang for this jam), it helps create the illusion that the map depicts an independent reality. At a minimum, it creates a larger menu of places to reference when writing other hex or NPC entries.

Doodles

For better or worse, I've incorporated over 160 of my doodles into Zurth. Often, when I'm feeling burnt out on hexfill prompts, working off of a picture is enough to inject some new creativity.

Oddly, a lack of artistic proficiency is helpful here. I couldn't draw half the things in my hexes if I wanted to, so the things I did draw were limited by my skill & what public domain reference images I had access to. And moreover, the things I drew never ended up quite like I intended. This influx of accidental novelty often inspired me to invent additional details for the written hex description.

Tracking history in eras, not years

I made a list of 17 "aeons" of history, each with a ridiculous name (a horrid fusion of Dying Earth & Infinite Jest) & liberally referred to them any time a hex included something old. I find that using exact dates to keep track of history is too constraining & fussy, but using vague eras is a good way to group different motifs & create a flexible, extensible sense of past.

As far as the length of an aeon goes, I like to remind myself that the ancient Babylonians imagined that their founding kings has reigns lasting tens of thousands of years.

Not anticipating play

Sometimes there's a little voice in my head that says things like "how are players supposed to interact with this?". This voice exists because as excessive as the whole project is, I do want the Antarctic hexcrawl to be something that gets played.

But there's a point where 'writing to be played' passes into 'writing to anticipate play'. In sandbox games, fun is emergent & hard to anticipate. A scenario is not a curated experience but more like a toy, and the goal is to increase the 'surface area' where it can make contact with players. Compressing an imagined country the size of Germany into a hundred pages requires elision, but basing that elision on a pre-conceived idea of who the players are & what they want impoverishes the end result.

Thus, I found it helpful step back & imagine I was not writing for a game, but rather writing a neutral description of a place -- what are the details that come to mind when I'm not imagining the world through the eyes of four or five murderhobos?

It's an approach & a skill I want to develop more. I'm going to quote Scribble's eloquent articulation of a similar sentiment:

I’m beginning to question whether it’s important for my dungeon maps to be game-friendly at all. I’ve decided to begin focusing on things which I find to be beautiful and meaningful, and trusting the people I play games with to find them beautiful as well. This has been freeing.

Weaving in "threads" in one go

I wanted my region to be interconnected & complex, but I found it very difficult write new hexes when there's a checklist of loose ends I have to consider.

So after placing most of my locations with the bare minimum description (see points above about naming every settlement, multiple passes, & anchor hexes), I would deliberately add a single "thread" to multiple hexes in one go. (By "thread" I mean a concept, monster, historical event, institution, or other imaginary entity that links multiple locations).

For example, during a session of Muridcog's Manse, Jay_zer0 asked if it would be possible to open a bank account, so the following day I added a bank to my main trading city along with one-sentence notes about the presence of a bank branch in several dozen other hexes. When writing hexes one at a time it burdensome to consider whether each new settlement had a bank branch, but once I had written most of my hexes it was very easy to expand the setting to include a rudimentary financial system.

This approach also worked well for history (see aeons, not years above), tie-ins to other regions, new monsters, and factions. I guess the lesson is to get the groundwork in first, then work on "installing" things that bridge multiple hexes.

Having written a lot already

Writing hexes is easier when you've already written hexes. The last few weeks of the jam have been some of the most productive because for new hex I had 400+ hexes to riff on.

Running games

A no-brainer, but running even a handful of sessions generated a wealth of new ideas & motivation. Excited for more of this to follow.

Get other people to look at it

Another no-brainer. Special thanks to Forlorn Skulk, Jay_zer0, & EvilTables who provided draft feedback during some part of the year-long writing process.

Condensed Maxims

By me, for me:

• If you use a random generator, do not record the prompt. Write as much as you can, then let it go. Return to what you've written later with fresh eyes.
• Just write. Every hex you put down is something new to riff on in your next hex. Don't be afraid of writing something stupid, a stupid hex that exists is more useful than a brilliant hex that doesn't. Stupid hexes can be redeemed through re-writes or new context.
• Write in a way that bee-lines toward a "minimum viable product". Start with the most important, gameable hexes, then backfill until you've hit the desired density.
• Pick a few places to put in exhaustive legwork, and name or detail every instance whether or not it's "interesting". But don't do this for everything, be at peace with the things you let go.
• Try not to anticipate players. Write the things you enjoy imagining.

#AAJ #musings