A Fragile Kind of Permanence
Digital data is often described as perfect and permanent, since a copy is identical to the original and can be duplicated endlessly. That is true while the data is being actively copied and checked. Left alone, however, digital information is surprisingly vulnerable. Researchers and archivists use the term digital decay for the gradual ways stored data becomes damaged, inaccessible or meaningless.
It helps to think of three separate layers that can fail: the physical medium, the format the data was written in, and the software and hardware needed to make sense of it. A file can be lost at any one of these levels even when the other two are fine.
Layer One: The Medium Wears Out
Every storage technology has physical weaknesses.
Magnetic media. Information on magnetic tape and floppy disks is stored as tiny magnetized patches on a coated surface. Over time, the coating can shed or become sticky as the binder holding it together breaks down, and the magnetic signal can weaken. Heat, humidity and mold speed up the process. Tapes that have been stored badly sometimes need special treatment before they can be played at all.
Hard drives. Hard disk drives are mechanical devices with spinning platters and moving heads. Bearings can seize after long periods without use, and the electronics eventually fail. Even if the platters are intact, the drive that reads them may not be.
Optical discs. Pressed discs and recordable ones differ. Recordable discs rely on dyes or other layers that can degrade, and the quality of manufacture varied widely. Scratches, delamination and corrosion of the reflective layer can all occur. The commonly quoted lifespans for optical discs are estimates, not guarantees, and real results vary with brand, storage and handling.
Flash memory. Solid-state storage holds data as electrical charge. That charge can leak away gradually, and the rate depends on the device, its age and the temperature. Devices left unpowered for long periods may lose data, though exactly how long is not something to promise.
Older media. Punched cards and paper tape can last well when kept dry, which is one reason some early data survives while later material does not. Paper has its own enemies, including damp, insects and fire.
Small errors can also build up silently. Often called bit rot, this describes individual bits flipping because of material defects or stray radiation, without any obvious sign. Systems that store checksums, which are short values calculated from a file’s contents, can detect such changes, but many personal collections have no such protection.
Layer Two: The Format Becomes Unreadable
Suppose the medium survives. The next question is whether anyone can interpret what is on it. A file is a sequence of bytes, and its meaning depends on a convention agreed by the program that wrote it. If that convention is proprietary or poorly documented and the program vanishes, the file may become a mystery.
Text is a good example. Different systems used different codes for letters, accents and symbols, so a file that looks fine on one machine can appear garbled on another. The history of character encoding is full of such mismatches. Word processor documents, spreadsheets, databases and graphics each had their own formats, many of which changed from one version of a program to the next.
Some formats are easier to preserve than others. Plain text, widely documented image formats and open specifications tend to remain readable, because many independent programs can handle them. Closed formats tied to one program are riskier, although in practice enthusiasts have often worked out how old formats operate by studying them.
Layer Three: The Environment Disappears
The third layer is the machinery around the data. Software needs an operating system, and operating systems need particular hardware. A game written for a specific home computer may rely on timing quirks of its processor and display. A business program might need a copy-protection device, a particular printer, or a network service that no longer exists.
Software that depends on online servers is especially exposed. When the server is switched off, the program may no longer work even though every file is intact. This is a growing concern for modern apps, which makes the problem different from the one faced by owners of older, standalone software.
What Helps
There is no one cure, but a few habits make a large difference.
- Keep multiple copies on different kinds of media in different places.
- Check copies regularly, using checksums where possible, and replace anything that shows errors.
- Migrate before it is too late. Copying data to current media while the old drive still works is much easier than rescuing it later. This is the idea behind media migration.
- Prefer well-documented formats, and keep a copy in an open format alongside any special one.
- Record context: what the files are, which software made them, and what settings were used.
- Keep the means to run the software, either through original hardware or through emulation software, where that is legal and practical.
Rescue Is Sometimes Possible
Failure is not always final. Specialists can sometimes recover data from damaged disks, read tapes by carefully cleaning them or reconstruct a file from partial information. The results are uncertain, and the process may be slow and costly, so it is better regarded as a last resort than a plan.
The overall lesson is that digital preservation is an ongoing activity, not a one-time act of saving. Data lasts only as long as someone keeps caring for it, which is true of old paper records too, but digital material demands attention on a much shorter schedule.