Digital Preservation: File Formats That Will Still Open in 20 Years
Some file formats age well and some quietly become unreadable. Which formats to choose for long-term records, when to convert, and how to keep the originals honest.
Paper left in a dry box is readable in a century. A digital file left alone for twenty years may not be, not because the bits have decayed but because nothing can interpret them any more. Formats tied to one company's software, one version, or one device are the ones that go first. If you keep records for a long time, the format you save them in is a preservation decision.
What makes a format durable
- Openly documented. The specification is published, so software can be written to read it even if the original vendor disappears.
- Widely used. Formats used by millions of people are supported for decades because too much depends on them.
- Self-contained. Fonts, images and data are embedded in the file rather than linked to something elsewhere.
- Not encrypted or copy-protected in a way that depends on a key or service that may not exist later.
- Simple where possible. The fewer features a format relies on, the fewer things can break.
Practical choices
| Content | Good for the long term | Riskier |
|---|---|---|
| Final documents | PDF/A | Word processor files from old or niche software |
| Editable text | Plain text, open document formats | Proprietary formats tied to one product |
| Tabular data | CSV with a description of the columns | Spreadsheets with macros and external links |
| Photographs | TIFF, high-quality JPEG, PNG | Camera raw formats alone |
| Audio | WAV, FLAC | Formats needing a specific player |
| Standard message files (EML, MBOX) or PDF of key messages | A single mail program's database |
PDF/A deserves a special mention: it is a version of PDF designed for archiving, which embeds fonts and forbids features that depend on outside resources. For finished documents, it is the safest general choice.
Convert at the right moment
Convert records to a preservation format when they become final, not decades later when the software is already gone. For a contract, that is when it is signed; for a report, when it is issued. Converting at the point of closing a matter is a good habit because the work is fresh and someone can check the result.
Keep the original too
Conversion can lose things: formulas in spreadsheets, tracked changes in documents, camera metadata in photos. Where the original has value, keep it alongside the preservation copy, and record which is which. A consistent naming pattern helps; see file naming conventions people actually follow.
Prove nothing changed
For records that may need to be relied on, compute a checksum (a digital fingerprint) when you store the file, and record it. Checking the fingerprint later shows the file is identical to what was stored. Many archives run these checks on a schedule to catch silent corruption.
Check, migrate, repeat
Preservation is not a single act. Every few years, check that your stored formats still open in current software, and plan migrations for any that are becoming obscure. Make sure the storage itself is backed up in more than one place. And make sure your retention schedule says which records need to survive for decades, so you spend preservation effort on the records that need it, not on everything.
