The phrase comes from Steph Ango, and the argument is short enough to state in one line: the file format you write in will outlive the app you write it in, so choose the format deliberately.
It sounds obvious and almost nobody acts on it. People choose a notes app by feature list, migrate three thousand notes into it, and discover the ownership question only when the app is acquired, discontinued, or rewrites its sync layer.
The uncomfortable arithmetic
Consider how long the things in your notes are supposed to last. A journal entry from your twenties. A reading note for a book you will teach from. The reasoning behind a decision your company still lives with. Ten, twenty, forty years.
Now consider how long notes apps last. Some of the most-recommended apps of 2010 are gone, frozen, or barely maintained. The ones that survive have often changed hands, changed pricing model, or changed what they are for.
The mismatch is the whole problem. You are storing decade-scale material in year-scale containers.
What "format" means here
Format is not the app's export button. It is what your notes are, at rest, on disk.
| Storage shape | What you own | Risk on app death |
|---|---|---|
| Plain text or Markdown files in folders | The notes, fully | Near zero. Open them in anything |
| Proprietary local database | An opaque file | Depends entirely on an export tool still working |
| Cloud-only, no local copy | An account | Total. Access ends when the service does |
| Cloud with open export | A download you have to remember to take | Moderate. Exports lose links, formatting, structure |
The important row is the last one, because it is where most people actually are and it feels safe. An export button is a promise, not a property. It usually produces a zip of HTML with the internal links broken and the metadata gone, which is a backup of the words but not of the structure.
The principle in practice
Prefer formats that predate the app and will postdate it. Markdown, plain text, and standard image formats can be read by hundreds of programs and by a human with a text editor. See plain text notes and Markdown for notes.
Prefer files on a filesystem to rows in someone else's database. A folder of files can be backed up, versioned, searched, and moved by tools you already have. This is why how to back up your notes is trivial for some setups and impossible for others.
Judge lock-in by what leaving costs, not by what the export button claims. The test: if this company shut down tomorrow with no notice, what would you still have? If the honest answer is "a zip file I would have to clean up by hand", that is the real cost of the app.
Accept that some things are worth the lock-in. This is where the principle gets misused. Databases, live collaboration, backlinks resolved at query time, AI over your corpus: these are genuinely hard to represent in plain files. If a feature is the reason the app is useful, paying for it in portability may be right. Just do it knowingly.
Where the principle is weakest
File-over-app is a strong argument and it is not free. The honest costs:
- Plain files do not sync themselves. You inherit that problem, and sync conflicts in a folder of Markdown are their own genre of pain. See syncing notes across devices.
- Structure lives in convention, not enforcement. Nothing stops you naming files inconsistently, which is why how to name notes matters more in a file-based setup.
- Some formats are only nominally open. Markdown has dialects. Wikilinks, callouts, embeds, and frontmatter differ between apps, so a Markdown vault can still be app-shaped in practice.
- Mobile filesystems are hostile. On phones, "a folder of files" is a worse experience than it sounds.
The version of the principle that survives contact with reality is not "only ever use plain text". It is: know which of your data is portable, and make sure the irreplaceable part is in the portable half.
A five-minute audit
Ask these about whatever you use now:
- Where do my notes physically live? Name the folder or the service.
- If I export everything today, what breaks? Links, attachments, tags, dates?
- Could I read the export in ten years with no special software?
- Which of my notes would I actually miss? Are those the ones in the fragile format?
Most people find the answer is a split: the fast, disposable capture lives in whatever app is convenient, and the material that matters should be somewhere duller. Moving only that second category is a small job, and it is the whole benefit.
The connection to everything else
This is why the personal knowledge management conversation keeps circling back to files. A second brain that cannot be moved is a rental, and the whole premise of external memory is that it lasts longer than the thing it is standing in for.
Related: encrypted notes for the privacy half of ownership, note versioning for history, and how to switch notes apps for when the audit tells you what you were afraid of.