Why worldbuilding is the asymmetric risk

A novel manuscript exports cleanly to almost anything. The structure is linear: chapters, scenes, paragraphs. A worldbuilding system is shaped differently, it's a graph of relationships, with custom fields, cross-references, timeline entries, maybe a custom calendar with cycles and eras. Every vendor stores that graph in their own way. Export usually means “here's a flat dump of names and descriptions, good luck rebuilding the relationships.”

If the vendor disappears, you don't lose the world, you lose the structure that made the world usable.

The two questions that decide it

Before committing a series to any tool's worldbuilding system, ask:

  • What does each entry look like on disk? A file? A row in someone's database? A blob in an unspecified format? If you can't answer this, you don't own the data.
  • Could you, in principle, write a script that reads your worldbuilding and renders it into a different system? If yes, the format is open. If no, you're a hostage on a long enough timescale.

What “your files” looks like in practice

The most portable worldbuilding setup is also the simplest: every entry is a file in a folder. Characters in characters/, locations in locations/, magic systems in magic/. Each file is structured text, Markdown, YAML front-matter, anything a script can read.

Edda keeps its working data in a single database file on your own disk, and gives you two open ways out. Turn on the plain-text mirror (desktop, Settings > Data) and Edda writes every character, location, article, timeline and chapter as a Markdown file with YAML front matter, in a folder you choose, each time you save: text you can ls, grep, and back up with rsync. And any project exports as a single JSON project file that carries every table, relationships included. The interface is replaceable; the data isn't.

Custom calendars and the “features become bait” problem

Worldbuilding tools love advertising custom calendars, fact graphs, and contradiction detection. These are real, useful features. They're also the lock. The fancier the proprietary feature, the harder the migration if you ever want to leave.

The right test: does the feature operate on standard underlying data, or does it require its own proprietary store? Edda's custom calendar, for example, is just structured data that travels in the project file like everything else: the feature is the rendering, the data is portable. That's the version of a fancy feature you want.

If you're already locked in

If you've spent two years inside a closed worldbuilding tool, the migration is annoying but possible. The pragmatic playbook:

  • Export everything you can, even the lossy flat exports.
  • Take screenshots or PDF prints of any view the export doesn't capture (especially relationship graphs and timelines).
  • Move new entries into the new system first. Backfill the old ones as you actually use them, not in a big-bang migration.
  • Resist the urge to rebuild perfectly. The first goal is “new work goes into a portable system,” not “ten years of legacy data is pristine.”

The principle

If you write fantasy, science fiction, or any series with a real world behind it, treat your worldbuilding the way you treat your manuscript: it's yours, it stays on your disk, and the vendor's role is to give you a good interface on top of files you already own.

Back to Privacy-First Writing Apps: A Working Author's Guide.