The file you can still copy but no longer open

This is less a set of instructions about file format obsolescence than an argument, and it is worth saying so at the start.

Two different failures get confused

Media failure means the physical carrier degrades, which is the problem most people picture when they think of digital loss. Format failure means the data is perfectly intact and no available software can interpret what the bytes are supposed to mean. The second failure is far more common in practice and much less visible, because the file continues to exist and copy normally.

Backups protect entirely against the first and not at all against the second, which is why backup alone is not preservation. A folder of files nobody can open is indistinguishable from a well-maintained archive until somebody tries to read one.

Documentation determines recoverability

An open, published format can be reimplemented by anybody, so software to read it can be written long after the original vendor has gone. An undocumented proprietary format must be reverse engineered, which is laborious, legally uncertain in some jurisdictions and often incomplete.

Formats tied to specific hardware, licensing servers or activation are effectively unreadable once that infrastructure is withdrawn. Encrypted files with lost keys are unrecoverable by design, which is the intended behaviour rather than a defect. Choosing formats on the basis of documentation rather than convenience is the single most consequential preservation decision.

Institutions have workable answers

Archives typically normalise incoming material into a small set of well-documented formats and keep the original alongside it. They then migrate periodically, checking that each generation still opens and recording what was done at every step.

What is easy to forget is that fixity checking, which compares stored checksums over time, detects silent corruption that a casual copy would propagate unnoticed. Emulation offers the alternative approach of preserving the original environment rather than converting the file. Both approaches are established practice, and the choice depends on whether the behaviour of the software matters or only the content.

Households do none of this

Personal collections are stored in whatever the current device produces, spread across services, and never checked until needed. Photographs and video are relatively safe because their common formats are widely supported and enormously popular.

By the end, documents, project files, notes in specific applications and anything produced by a discontinued service are much more exposed. Cloud storage protects against media failure and creates a new dependency on an account remaining active and paid.

Account closure, inactivity policies and provider withdrawal are realistic loss scenarios that most people have not considered.

Practical steps that actually work

Keeping important material in plain, widely supported formats removes most of the risk at almost no cost. Exporting from any application that uses a proprietary format, and storing that export alongside the original, is a cheap insurance. Holding copies in more than one place, including at least one that is not tied to a single provider, addresses the account risk.

Opening a sample of old files occasionally is the only way to discover a problem while it is still solvable. None of this requires technical skill; it requires the habit of assuming that current software will eventually stop existing.

The record this produces

The likely outcome is that this period will be well documented in photographs and poorly documented in everything else. Correspondence in particular has moved into services that individuals cannot easily export and institutions cannot easily collect. Archivists have been describing this gap for a long time, and the tools have improved faster than the habits have.

In truth, the loss will not appear as an absence of data but as an abundance of material nobody can interpret. That is a strange kind of forgetting, and it is the one this period is actually building.

The takeaway

The bits will be fine. The question is whether anything will still know what they were supposed to mean.

The loss is small and cumulative, which is why it goes unremarked.