This is a walkthrough of how the pipeline is currently structured.

Data layout

Language data lives here in this website repository, not in the language lab tool itself. The tool runs loads these data sets when it starts up, centralizing my language data here for future features (e.g., a glossary, a fan name generator).

What's in that data? At a high level:

• A manifest declares which languages exist and where their files live.
• Each language has its own profile and lexicon.
• Optional layers include eras, dialects, and ontology overlays. These overlays describe how concepts relate to one another, so the system can reason about meaning instead of relying only on literal word matches.

That structure forces discipline. If the generator produces something incorrect, the fix usually belongs in the data rather than in the generation logic. It also means, theoretically, the tooling gets better the more data I can shove into it. That takes some of the pressure off of me to build our grammatical or lexical rules like I'm Tolkein. I'm really not good at languages, so this is my programming cheat code.

The pipeline

The pipeline is a sequence of constrained choices. As we move through each stage, it reduces the space of valid outputs before the next decision is made. We sort of narrow-in on a valid word form rather than generate it.

1) Language, dialect, era

The first step selects a language, an optional dialect (only when the language declares dialects), and an era. So far I have 1 language (Thalen), 2 dialects (Upper Thalen and Southern Thalen), and 3 eras (Ancient, Old, & Modern).

Those choices determine:

• phonology and phonotactics, the sounds the language uses and the rules that govern how those sounds can legally combine
• morphology templates and constraints, the patterns used to assemble words from smaller pieces, and the limits on which patterns are allowed
• sound change rules applied across an era chain, a list of transformations that gradually alter pronunciation over time
• dialect specific overrides and affixes, local adjustments that reflect regional speech or cultural variation

(No, I didn't know most of these terms before I started this process. This is what search engines are for!)

Era selection matters because forms are evolved forward through declared changes. A modern output is the result of applying those changes in order, not a fresh construction. This is my digital eTolkein methodology.

2) Purpose and name classification

Next comes the purpose of the word: name, place, thing, or abstract concept.

Names are further split into personal and family forms, since those tend to follow different structural conventions. This choice controls which morphology templates are allowed and which roots are filtered out. For example, roots associated with institutions or collectives are excluded from family names unless explicitly requested. In early test all my families ended up being called "Clanguy" and the like. Not great.

This step establishes expectations early so later stages are operating within a compatible structure.

3) Register, contact, orthography

These are modeled as separate layers because they affect different aspects of the output word.

• Register represents social tone, influencing whether more formal or everyday vocabulary is preferred
• Contact models influence from other languages, allowing borrowing rules to activate when cultures interact. I expect this will have a greater effect as I add new languages to the models, but I need to research more on how these influences work: root borrowing, sound shifts, etc.
• Orthography controls how a word is written, converting an internal sound representation into a visible spelling

Keeping these layers separate allows the same underlying word to appear differently in different contexts without changing its core structure.

4) Intent and semantic grounding

Intent acts as the semantic input, essentially a short description of what the word is meant to convey.

The lab breaks that intent into tags, then expands those tags through an ontology. The ontology is a network of related ideas that allows the system to connect concepts that are close in meaning even if they are not identical.

This has been the biggest time sink in the project, and seems to have the greatest degree of impact on the quality of the generated words. I don't think my work here will be done for some time to come.

Roots are then scored based on how well they match those expanded tags. Matches can occur through:

• root identifiers, the internal labels used to track meaning
• tags, short semantic labels attached to roots
• gloss text, which is a plain-language description of what a root means, similar to a dictionary definition

Using gloss text allows the system to catch softer matches where terminology differs but meaning overlaps. For example, “healing” can connect to “medicine,” and “footprints” can connect to “trail,” even if those exact words are not present in the prompt. I've been scanning thesauruses to help out here.

A secondary bias layer exists for culture specific themes, but it only activates when literal intent fails to produce matches. This prevents local flavor from overwhelming explicit prompts (an early failure mode).

5) Root selection with multi tag coverage

When an intent contains multiple strong tags, selection favors coverage rather than repetition. This sounds obvious, but the code didn't think so at first. "Fast-fast-fast-mcgee" is not a good pattern for a name, it turns out.

If three or more tags score highly, the generator prefers templates that support multi-root compounds. Roots are chosen to represent different aspects of the intent before reinforcing a single dominant meaning.

This allows complex prompts to resolve into structured compounds instead of collapsing into a repeated theme.

That sounds very fancy, but it just means I have a bit more variety in the outputs that come through. If I give a long phrase for the meaning of a word then some of the options that generate will probably be longer to try and encompass all those ideas. I'm building it with a range of outputs though so even when I get long versions some of the options will focus on the strongest root and lean into that alone, keeping things short.

6) Morphology and sound shaping

Once roots are selected, the system proceeds through a fixed sequence:

1. apply compounding rules and affixes defined by the language
2. run the result through the era chain sound changes
3. validate the output against phonotactic constraints
4. repair violations using minimal, constraint-aware adjustments

Repair operations favor small local fixes, usually vowel insertion guided by syllable templates, rather than broad substitutions. This keeps adjustments aligned with the language’s internal logic. For instance, a root ending in a hard consonant butting up against the next root beginning with another hard consonant would make for an impossible pronunciation, so I squish in a vowel between them. Even these fixes I've managed to export into the language data itself, so those patterns can be regional or belong to the dialect rather than the whole planet.

7) Ranking and output

The generator produces a batch of candidates, then filters taboo roots and structurally broken forms. Results are scored, deduplicated, and returned as five options.

Earlier entries emphasize broader semantic coverage and longer compound words, while later entries lean toward the dominant intent and shorter words. This ordering gives me a range to choose from when selecting my match.

Validation and debugging

A validator runs across all language data to catch schema errors, missing references, and unreachable forms. One enforced rule checks that proto roots can reach their modern forms through the declared era chain. This has turned out to be a huge help since I'm not very consistent in what I shove in the system. Once it starts running it actually corrects even my source data to a degree.

At this point the lab produces results that are explainable at every step. I can validate each one, see where things are going haywire, and course correct. Once I begin generating content I feel meets a quality threshold, I've built in tooling to save those words back into the master data to further train future generation.

I'm looking at some complex rules for ritual magic here which will not make their way into this book. I have over 15,000 words written about a war with Qartago (the second one, actually) which is being removed. Qartago, by the way, was the old working name for Qadashur. The Hierophant was exploring a different genre at the time.

And these words aren't all for naught. There are character sketches in here who will transpose into this new reality and find life in a slightly different shape. I have set-pieces and theatrical moments which can be borrowed. In the war files I worked out many types of ship that traverse the Ioma, all of which can journey onward.

The difficulty of managing it all is still there, though. So I set out to try something new. I've created the Constellation, a living wiki built of the interconnected ideas across this site: in the map, in the world-building, in these blog posts, and even in the calendar. The natural ways these things link together is graphed in 3D and the nodes are clickable to explore. The format is a little strange and it may seem less intuitive than many wiki's you're used to but this has two big advantages for me.

1. I don't need to do anything to update it. The node graph builds automatically based on the site content I add. It will grow and evolve as my writing does. That's far more maintainable than using something like wikipedia's MediaWiki software.

2. And it allows me to visually see segments that aren't connecting. Are they old fragments of language or novelty? Are they ideas which I haven't developed enough to find connection? It's a framework for me to constantly improve the world.

I'll continue to evolve the interface for this as well as time permits. Maybe there's a way to make it easier to browse, or get those labels to feel more natural. In the meantime, play around with it if you feel like exploring new places. The way the map portion came together has been great.

I'm nearly done with my present focus area: renaming the world's locations. I have all the names and rough profiles in place. I'll be moving over to the graphical portion of the map next in order to sync it with this new vision. Slowly but surely the sources are becoming more trustworthy. That will give me so much to lean on as I move forward.

This initial version leans on a straightforward pipeline: take the post content, hash it, send it to a text-to-speech provider, and stash the resulting MP3 in "Netlify Blobs". My blog engine reads a simple map to know when a post has audio, and a special back-end function streams it to the page. If the words haven't changed, the hashes stay put and nothing regenerates. If they do, a quick sync updates the blob and the map, and Bob's your uncle.

Right now the voice is a stock default model because it let me move fast and get something working. The next step is training a local model on my own voice so the Chronicle sounds like me instead of a robot queen. And for the stories that matter most (short fiction, readings of lore, or anything that wants a human performance) I'm keeping the pipeline flexible enough to drop in manually recorded audio whenever that feels right.

It's a small thing, but hearing the world speak makes it feel more alive, and it keeps the door open for everyone who prefers to listen in.

Did you listen to this post? What do you think?

Going beyond the surface

Those early drafts lived in sketchbooks long before The Hierophant’s plot found them. I’d trace constellations, shade in continents with colored pencils, and then set loose whatever campaign or speculative idea I was toying with. The setting cycled through fantasy bestiaries, prehistoric survival epics, even science projects. It waited, patient and unused, until the book’s central theme clicked: What happens after death?

Once I started comparing afterlife calendars, the map came alive. Forty-nine days in the Tibetan Book of the Dead, forty for Eastern Orthodox judgement, seventy for an Egyptian embalming, thirty for Jewish sheloshim, eleven for the śrāddha rites in parts of India. Every tradition marked time differently, but each assumed similar tiers of nearness to the sacred. My notes circled a common intuition:

The layered afterlife and cosmological systems arise from the suspicion that reality is stacked in ascending degrees of closeness to the sacred, Mircea Eliade’s ladder between the sacred and the profane, and that the soul’s journey is an upward movement through each chamber of truth.

So what does that first chamber look like? Religions have been sketching it for millennia.

And the list keeps going. I’m drawn to how broad the consensus is: no matter the language, people keep pointing to a lowered sky that doubles as a waiting room. That common thread is the compass I keep following.

Mapping accuracy

Trading colored pencils for a digital atlas created its own headache. The map begged for a “show me the stars” toggle, but my ancient coordinates were… guess work. Teenage me converted right ascension and declination into latitude and longitude on a TI-83, which was fine for doodles and terrible for precision.

So I tore it apart.

I rebuilt the math from scratch, piped in modern star catalogs, and wrote a converter that projects celestial coordinates onto the Mercator grid that Leaflet expects. Then I nudged continents across 190 degrees of longitude, reshaped coastlines, and started reconciling the geography with lore I’ve added over the last few years. It’s messy right now, but useful mess often is.

If you poke the map today you’ll catch it mid renovation: shorelines half-drawn, city labels migrating, mountains waiting for their ridgelines. Hit the “S” key anyway. Star Mode still works, and seeing the constellations flood the world never gets old.

• Abawok - Coptic: Corvus - Crow - Abwk
  • Abowk - Algorab - Crow - Abowk
  • Brusoot - Kraz - Chariot - Brusoot
  • Kepeh - Alchiba - Tent - Covered Space - Kepeh
• Nahrina - Minkar - Nostril - Face - Nahrina
• Tinh - Gienah - Wing - Tinh

Picking a language

Originally I'd leaned into Coptic as the linguistic root for my working names in this region. There wasn't any particular reason for that. Languages weren't high on my list of focus areas for a long time, and so I just distributed them around the globe and thought it good enough for the time.

I knew I wanted some historical depth to my place names. They would grow and change with the history, but what that history was still needed to be worked out. How can you work out a place's history until you know its environment, ecology, staples, relationships with neighbors? Like everything else in The Hierophant, I was trapped by the systems thinking. I needed to understand it all in order to understand the parts.

The language had to follow the same arc. Using Coptic phonetics as filler never matched the society I was sketching, so I built a Lokhan sound set that serves the culture: aspirated kh, tight s/z/l clusters, clipped vowels, final-syllable stress. Those rules grew out of the civic rituals, so the sounds and the institutions reinforce each other. Once the morphemes started earning their keep (lokh + -var for the nation, val for harbors, -rel/-vyr for edges and glides), the names started coming together.

Giving it life

Lokhavar fits. The new write-up leans into the feather lexeme that runs through the Lokhan language, so the Feathers Circle, the Kelresh Belt, and the Lokhfire Nights all feel like they belong to the same place. I finally wrote down the Mask Houses, Sazrel, Lazkhor, Velrun, and let them anchor the politics instead of leaving the archipelago implied.

Oh! I should note here that this is an archipelago. This is one area where my graphical map is lagging behind the text. The block of land appears quite solid, but this whole region is a network of thousands of islands. That's going to take some work.

Getting the name right also meant giving the cities their due. Valokh now steps down basalt terraces into the parliamentary harbor, Zavrel gets its equinox chariot launches, Kelresh has those geothermal arcades drying reed-wool, Nasrel holds court in its acoustically tuned basins, and Talvyr keeps watch from the glider ramps. Each city got its own page so the map labels aren’t pointing at ghosts anymore.

Will the story ever travel here? It's not likely. My current plotting keeps most of the story along the Ioma. So why bother?

Well, it's a part of the system. People are from there. Trade is from there. I may encounter characters is my book who spent time there. And even if it remains a distant speck, now it's a living speck.

Cleanup

Now comes the hard part. After I've done the renaming I need to go back through all of my other notes and find references to the old names, systems, cities, rituals, etc. I need to update the balance of the system to take in those changes. This is the one space I have allowed myself to dip into that dreaded evil, AI. I don't want it to touch the generation of my world in any way. This is my brain's work. This is my passion, my dreams. But double-checking thousands of notes across multiple projects that might reference things in passing? Yeah, cool, go ahead and scan away little robot.

In this case it was really helpful in identifying some issues with the calendar system I'd started for this area. Two of them, actually. I was able to cut some threads, remove old ideas, and just smooth over the rough spots.

It's never perfect, though. I can't blame the LLM for this, but this book has gone through so many iterative ideas, settings, and growths, that my notes sometimes have some true nonsense in them. Did you know at one point I had an evolutionary extension of neanderthals in here? Or that the story went steampunk for a good two weeks? I think I still have an airship reference in the atlas somewhere I'll need to clean up eventually.

Once I have all of my nations and cities properly named I can do a clean run-through of everything with a red pen. It shall be glorious.

Just a few more to go.

From stars to borders

I began by pulling in the official constellation boundary tables from the International Astronomical Union. Those are the straight lines that divide the sky by right ascension and declination. My mapping library, Leaflet, can draw polygons, so the first plan was simple: convert those celestial coordinates into Mortum Caelum’s world grid and use them as national borders. Easy, right?

Early trials used affine transforms and polynomial fits. (Yes, I had to look those up and ask ChatGPT to explain them.) On paper they worked, but the map looked wrong. Some nations stretched like pulled dough while others shrank to crumbs, and the edges refused to sit comfortably against the terrain. It was the classic case of a model that is technically correct but visually false. I tried to use offsets to force it, but no matter how I poked things didn't line up.

Grounding the sky

So I reversed course. Instead of forcing the sky onto the land, I let the land speak first! Each nation and city already had latitude and longitude anchors from worldbuilding notes I wrote decades ago, so I built polygons that wrapped around those points. The first pass used convex hulls (the mathematical version of pulling a tight rubber band around a set of pins) but overlaps and gaps quickly appeared where borders met like puzzle pieces that didn't quite fit.

The breakthrough came with a Voronoi approach. Imagine every city sending out a signal and every spot on the map pledging allegiance to whichever signal is nearest. By computing that pattern and merging cells belonging to the same nation, I ended up with a complete tiling of the map. No overlaps, no empty space, and every settlement resting safely inside its own borders. Huzzah!

Cleaning and trimming those cells into GeoJSON gave me a working political layer that Leaflet renders in real time. Watching the shapes lock together like stained glass was its own quiet reward.

Refinement and rhythm

This round of work built on last session’s improvements: faster load times, better label avoidance (so they don't overlap when cities are close to one another), smoother panel animations. There is still polishing to do, especially along the coasts. For now the sea boundaries spill to a broad bounding box so that every nation touches water. In a later pass I’ll draw the maritime lines by hand, guided by the printed sketches in ancient notebooks. Oh, and I should probably do something about the poles. Things get goofy at the extremes.

Why this matters

None of this labor is about perfection in code. It’s about creating a world that feels inhabited. Readers will never check if a Voronoi tessellation defined a border or if I adjusted an outline with polynomial regression, but they will sense whether the geography holds weight. The mix of astronomy, geometry, and storytelling is what makes Mortum Caelum breathe, and the map should honor that.

Now a single command can regenerate every border. Add a city, run the script, and the land rearranges itself around the newcomer, a quiet, almost magical act of creation. It may seem an indulgent effort, but each small calculation ensures that when someone zooms in to explore, they meet a landscape tended with patience, curiosity, and a trace of wonder.

From the start I said the process should be as enjoyable as the book itself, and that promise finally has a home. The blog is alive, wired through Eleventy so I can log the strange detours, lore fragments, and production notes in real time. Running the pipeline feels as ordinary as making tea, and that ritual is what keeps the documentation flowing instead of stalling behind “someday I’ll write about this.”

I have always insisted I don’t need a publisher because I am a marketer by trade, so this stretch of work doubled as a reminder: I can build the entire runway myself. Cleaning the tooling, tuning the watchers, tightening the Netlify deploys, these are the quiet systems that let me ship on my own terms. The layout refinements, from the Ioma-stars image to the non-sticky header and palette-friendly cards, make the whole space feel authored, not borrowed.

The CRM sign-up flow had to match that independence. I designed the form, double opt-in messaging, and stitched together a private broadcast system so every message stays personal and under my control. Testing the send and seeing it land felt like setting up a press of my own: no third-party gatekeepers, just a direct line to the readers who want to follow along.

None of this is flashy work, but it clears the path for curiosity to keep moving. Now the site can keep pace with whatever notes, sketches, or lore pages come next, and I can keep proving that building a world, and sharing how it’s built, belongs as much to the maker as the finished book.

The next content project is with the Atlas. I've completed the naming of five of the twenty regions in my world. That work has involved crafting unique cultural history, ecology, and mythological influences to stub out a proto-language for each. Then I apply that language to the source-meanings of the nations and then their major cities. It's slow work moving from tentative placeholder names I've been using for years, but it also feels like a puzzle piece being slotted perfectly.

Once I've completed the exercise for all the places in the atlas, I need to go back and work on their names for one another. Languages! Yeesh. These aren't my strong-suit, so it's taking a while. I'm determined not to accept boring fantasy names, though.

Keep an eye on that atlas content. I'll send out a Dispatch email when we hit a completion milestone. Until then, back to the forge.

This is a space for process. I want to show every corner of the work, from sketches and prototypes to lore and logic. Some days that means writing about theology or ecology. Other days it means building a calculator that translates a fictional calendar or a map that breathes like a living continent. It is all part of the same story.

I am building a world piece by piece, because that is the kind of art I love most. It does not need to serve anything but wonder. If you are here, welcome. Stay as long as you like. Explore. The doors are open.