The Future of Digital Preservation — Why Formats Die Before Data Does

A file that exists on a hard drive is not the same thing as information that can be read. The distinction seems obvious when stated plainly. It becomes difficult to maintain once the software needed to open that file has been discontinued, the operating system it depended on has been replaced, and the hardware that could run it has been retired from production.

Digital preservation is the effort to keep digital information accessible across generations of technology. It emerged as a field in the mid-1990s when institutions realized that the materials being born digital — emails, databases, word-processed documents, digital images — were more fragile than the paper records they were replacing. A paper document can be read by anyone who knows how to decode the symbols on the page. A digital document requires a chain of working tools: storage media, file-system drivers, operating-system libraries, application software, and sometimes licensing agreements. If any link in that chain breaks, the information becomes inaccessible.

What digital preservation actually means

The field does not focus on preventing physical media from degrading. That problem, often called bit rot, is real but manageable. Hard drives fail, optical discs delaminate, and magnetic tape loses signal. The standard response is refreshing: copy the data to new media at regular intervals. The bits remain unchanged. Only the physical carrier is replaced.

Digital preservation deals with the harder problem of format obsolescence. This occurs when the encoding standard that a file depends on is no longer supported, or when the software capable of interpreting that encoding has been discontinued. The data is intact. The information inside it is not.

A file saved in WordPerfect 5.0 format in 2026 cannot be opened by any modern word processor without conversion tools that may no longer exist. The bytes are there. The structure they form — formatting codes, embedded metadata, field markers — is intelligible only to software designed to read that specific encoding. When that software disappears, the file becomes a collection of numbers that no one can interpret.

Format obsolescence is not a single failure mode. It has several layers. Format obsolescence occurs when the encoding standard is replaced. Software obsolescence occurs when the application that reads the format is discontinued. Hardware obsolescence occurs when the device that could run the application is no longer functional. Context obsolescence occurs when the metadata that explains what the file represents — who created it, what it describes, why it was saved — is lost. Each layer is independently vulnerable.

How institutions try to preserve digital materials

The most widely adopted framework for digital preservation is the Open Archival Information System model, standardized as ISO 14721:2025. It was first published in 2005 and provides a reference architecture for digital archives. The model organizes preservation into three types of information packages. A Submission Information Package (SIP) is the material as submitted by a creator. An Archive Information Package (AIP) is the version stored by the repository, enriched with preservation metadata. A Dissemination Information Package (DIP) is the version delivered to a reader, formatted for consumption.

The OAIS model requires repositories to maintain five categories of metadata: reference metadata that identifies the object, provenance metadata that tracks its origin, context metadata that explains its meaning, fixity metadata that verifies its integrity, and representation metadata that describes how the object should be rendered. The representation metadata is where the format-obsolescence problem is addressed. It must contain enough technical detail to reconstruct the environment needed to read the file, even when the original software is no longer available.

Two main strategies exist for dealing with format obsolescence. Migration converts files from an obsolete format to a current one. A WordPerfect file is converted to PDF or DOCX. The conversion is lossy. Some information — formatting nuances, embedded scripts, field calculations — may not survive the transformation. Migration must be repeated whenever the target format becomes obsolete.

Emulation preserves the original format by recreating the software environment that reads it. A video game saved as an MS-DOS executable is rendered through an emulator that mimics the original hardware and operating system. The file is not converted. The environment around it is rebuilt. Emulation preserves the original user experience. It requires ongoing maintenance because emulators must be updated as the host operating system changes.

The choice between migration and emulation is not purely technical. Migration is simpler to implement at scale. It requires fewer specialized skills and produces files that are immediately usable. Emulation preserves fidelity but requires sustained investment in software development and testing. Institutions often use a hybrid approach: migrate common document formats to archival standards like PDF/A or uncompressed TIFF, and use emulation for complex or specialized formats like software, databases, or multimedia applications.

The role of format registries

Format registries are the infrastructure that makes informed preservation decisions possible. The most widely referenced system is PRONOM, created by The National Archives (UK) between 2002 and 2006. It is described as the only operational public file format registry in the world. PRONOM maps file formats to their technical characteristics, the software that reads them, and the tools that can convert them. Each format is assigned a persistent, unique identifier (PUID) in the format info:pronom/.

PRONOM works by matching internal byte patterns — format signatures — against a database of known encodings. The companion tool DROID performs automated batch identification of file formats by scanning stored files and comparing their byte signatures to PRONOM entries. This is how institutions inventory what they have before deciding what to preserve.

The Library of Congress maintains a parallel effort: format sustainability ratings for its digital holdings. Each format is evaluated on criteria including external dependency, documentation quality, format knowledge, number of active users, and availability of conversion tools. Formats are rated as sustainable, at-risk, or obsolete. The ratings are not static. A format moves from sustainable to at-risk as its supporting ecosystem shrinks.

These registries are not preservation tools themselves. They are decision-support systems. They tell an institution what formats are in their collection, which ones are at risk, and what conversion paths exist. The actual work of preservation — migration, emulation, metadata enrichment — must still be performed by humans with tools that may also become obsolete.

The digital dark age

The concept of a digital dark age describes a potential future gap in historical documentation caused by the fragility of digital records. The term gained traction after a 2003 report by the Society of American Archivists titled “Electronic Records: Who Will Archive the Living Memory?” The report warned that electronic records are “highly vulnerable to technological obsolescence” and require continuous technical support to remain readable.

Unlike paper manuscripts, which can be examined by anyone who can read the language, digital files vanish without compatible hardware or programs. Vinton Cerf has noted that “memories stored in files that use older technology are becoming harder to access.” The problem is not that the data disappears. It is that the chain of tools required to read it becomes incomplete.

The digital dark age is not inevitable. Institutions like the Library of Congress, the UK National Archives, and the EU-funded PLANETS project (2002–2010) have built infrastructure to counter format obsolescence. The InterPARES research initiative, launched in 1999, has produced frameworks for authentic digital records across jurisdictions. The National Digital Stewardship Alliance (NDSA) in the United States develops preservation standards and provides education and training.

But the infrastructure is fragile in its own right. It depends on funding, institutional priorities, and the continued availability of specialized skills. The tools used to preserve digital records — format identification software, emulation frameworks, metadata systems — are themselves digital and subject to the same obsolescence pressures as the materials they protect.

What the Arctic World Archive reveals about the problem

The Arctic World Archive, opened on March 27, 2017, in Svalbard, stores data as QR codes on silver halide film buried in permafrost. The GitHub Archive Program has placed 186 reels in the archive, containing open-source code from platforms like GitHub. Film is chosen because it does not require power, specialized hardware, or proprietary software to read. The assumption is that a future civilization capable of reading optical micrographs will also be capable of decoding QR codes.

The Arctic World Archive solves the format-obsolescence problem by choosing a format that is maximally simple and universally decodable. QR codes are a known quantity. They do not depend on application software. They do not require specific operating systems. They only require optical reading equipment and a decoder that implements the ISO/IEC 18004 standard.

This is a valid strategy for specific kinds of data — source code, text documents, structured records — where the content can be reduced to a sequence of bytes with minimal metadata. It does not work for complex, interactive, or dynamic content. A video game, a scientific simulation, or a database with live queries cannot be preserved on film. The film can store the data. It cannot preserve the behavior that gives the data meaning.

The limits of open formats

A common assumption in digital preservation is that open, non-proprietary formats are inherently more sustainable than proprietary ones. The reasoning is sound. An open format is published, documented, and usable without licensing restrictions. Anyone can build software to read it. A proprietary format depends on the continued support of the organization that controls it.

The assumption is not always correct. Open formats can become obsolete when the community that maintains them dissolves. The format specification may be frozen at a version that no longer matches how people actually use it. Proprietary formats, by contrast, may remain accessible as long as the controlling company has a commercial incentive to maintain compatibility. The specification is not public, but the software is.

Another complication is that an open format is only as sustainable as the tools that read it. A format specification published in 1998 does not help if no modern application implements it. The Library of Congress format sustainability ratings account for this by evaluating not just the format itself but its active user base, documentation quality, and conversion tool availability. A format with a published specification but no active tooling scores poorly on these criteria.

What preservation reveals about information itself

Digital preservation forces a distinction between data and information. Data is the sequence of bytes. Information is what those bytes mean when interpreted through the correct framework. A JPEG file is data. An image of a sunset is information. The JPEG encoding is the bridge between them. If the bridge collapses, the data remains but the information is gone.

OAIS captures this distinction through its concept of the designated community — the group of users for whom the archive is intended. The designated community has “knowledge of the domain,” meaning they understand the context needed to interpret the preserved object. A bioinformatics researcher reading a genomic dataset has different domain knowledge than a musicologist reading a digitized score. The same bytes can carry different information for different communities.

Preservation is not about keeping bytes intact. It is about keeping the bridge between bytes and meaning intact. That bridge is made of software, metadata, documentation, and shared understanding. All of it is vulnerable. None of it is permanent.

What remains unresolved

The fundamental tension in digital preservation is between fidelity and feasibility. Emulation preserves the original format and experience but requires sustained development effort. Migration produces immediately usable files but loses information. The OAIS model provides the metadata framework to support both approaches, but metadata itself must be preserved.

A more structural question is who bears responsibility for digital preservation. The Society of American Archivists asked this explicitly in 2003: who will archive the living memory? The question is not answered by technology alone. Preservation requires funding, institutional commitment, and a shared understanding that digital materials deserve the same long-term consideration as paper records. Those conditions are not guaranteed.

The tools used for preservation are themselves subject to format obsolescence. PRONOM is maintained by The National Archives (UK). If that institution’s preservation program loses funding, the registry stops updating. DROID is Java-based software. If Java becomes obsolete, DROID becomes obsolete. The infrastructure is only as durable as the organizations that maintain it.

There is no permanent solution to format obsolescence. Every format will eventually become at-risk. Every piece of software will eventually be discontinued. The best that preservation can do is slow the decay, maintain multiple strategies simultaneously, and accept that some information will be lost regardless of the effort invested.

The alternative is silence — the digital equivalent of a library whose books are stored in a language no one can read. Silence favors those who control the current format standards. It disadvantages anyone whose records were created in a system that is no longer supported. A field that defines access through public, documented standards is a field that can be studied, challenged, and improved. A field that defines access through proprietary ecosystems controlled by commercial entities is less transparent and less reviewable.

The tension is unavoidable. Information must be encoded. If it is encoded in a format, the format will determine whether the information survives. The alternative is to store nothing at all, which is itself a kind of preservation decision — one that favors the present over the future.

Primary sources

• Open Archival Information System (OAIS) reference model. ISO 14721:2025. First published 2005. International standard for digital archive architecture, including SIP, DIP, and AIP information packages and five categories of preservation metadata.
• The National Archives (UK). PRONOM technical registry. Public file format registry released 2002–2006. Provides PUID identifiers, byte-signature matching, and format-to-software mapping. Companion tool DROID enables automated batch format identification.
• Society of American Archivists. “Electronic Records: Who Will Archive the Living Memory?” 2003 report. Warned that electronic records are highly vulnerable to technological obsolescence and require continuous technical support.
• Library of Congress. Digital preservation holdings and format sustainability ratings. Evaluates formats on external dependency, documentation, format knowledge, active users, and conversion tool availability. Ratings rated as sustainable, at-risk, or obsolete.
• InterPARES research initiative. Launched 1999. Frameworks for authentic digital records across jurisdictions.
• PLANETS project (2002–2010). EU-funded digital preservation research that developed format identification and migration strategies.
• Arctic World Archive. Opened March 27, 2017, in Svalbard. Stores data as QR codes on silver halide film in permafrost. GitHub Archive Program has deposited 186 reels.
• NDSA (National Digital Stewardship Alliance). Develops preservation standards and provides education and training. Succeeds elements of NDIIPP (National Digital Information Infrastructure and Preservation Program).
• Cerf, V. Statements on digital preservation. Noted that “memories stored in files that use older technology are becoming harder to access.”