A favorite smartphone game becomes unplayable on the day its service ends. A useful app disappears from the store and will not run on a new phone. A website shows a page-not-found message. These experiences are common.
A book can be opened again if it remains on a shelf, even after people stop reading it. A digital work does not always survive in the same way. When a company switches off its computers or the required device becomes obsolete, even a work used around the world can disappear.
Apps and games are more than convenient tools. They contain the technology of their time and clues about how people played, worked, and thought. Preserving digital works matters if people in the future are to understand our society.
A program preserves a path of human decisions
Software is a collection of instructions that makes an app or game work. Source code is the text that records those instructions in a form people can read and change.
Source code can show the order in which a program handles a task. Names given to features and notes written by developers may remain. If the history of many edits survives, it may also show how a problem was fixed and which idea was selected.
Older mobile phones, for example, had small screens and slow network connections. Developers had to choose what to display in limited space and find ways to reduce the information sent over the network. Source code can preserve the answers people chose under those limits.
A finished app can show us how it looked and moved. It may not explain why it took that form. Preserving source code and its history makes it easier to study both the technology and its creators’ decisions.
Why does something that can be copied still disappear?
A digital file can be copied easily without changing it. Long-term preservation is still difficult.
The United Nations Educational, Scientific and Cultural Organization is an international body that works to protect and expand education, science, and culture. It is known as UNESCO. In 2003, UNESCO adopted the Charter on the Preservation of Digital Heritage to protect digital materials worth passing to the future.
The charter says that software and web pages can be part of digital heritage. It also warns that digital materials are easy to lose. Hardware and software become obsolete quickly. Responsibility for preservation may be unclear. Methods for preservation continue to change. Digital materials are not guaranteed to remain if no one acts.
A file alone may not run
Saving a program file does not always make it possible to run the program.
First, the operating system for which it was made matters. An operating system is the foundation that controls a phone or computer and allows apps to work. Windows, macOS, and Android are examples. An app may fail when the type or version of its operating system changes.
A tool is also needed to turn source code into a form a computer can execute. This tool is called a compiler. A different compiler or compiler version may fail to produce the same result.
A program may also require other software parts, a particular game console, a sensor, or another server on the internet. An online game’s app can remain on a player’s device while its original way of playing becomes impossible after the server connecting players is switched off.
The Library of Congress publishes guidance on formats that are preferred for preservation. For software, it gives priority to source code that people can read. It also asks for a record of the operating system, version, and build number of the compiler that was used. A dedicated game console may require the hardware itself or a system that imitates it on another computer. Such an imitation system is called an emulator.
Saving a file is therefore only the first step. Source code, edit history, manuals, required software parts, build instructions, and the environment needed to run it should be preserved together. Even when perfect reproduction is impossible, recording what is missing can help future research.
An effort to collect source code from around the world
An international initiative collects, preserves, and makes publicly available source code that is already public. It is called Software Heritage.
Software Heritage automatically visits development websites and archives source code. Users can also ask it to save a project. It collects not only current files but also their origin and the history of how they changed.
Archived source code receives a special identifier connected to its contents. An ordinary web address stops working when its site disappears. A content-based identifier makes it easier to check whether code is the same even if its storage location changes. It can identify the exact version used in research or help compare an older version with a newer one.
UNESCO works with Software Heritage and treats source code as a record of human knowledge and ingenuity. A 2019 document called the Paris Call explains that making software is intellectual work and that source code contains knowledge.
Here, heritage does not mean only displaying an old object in a museum. It means a record that our society passes on so people in the future can learn, conduct research, and make new things.
Not everything can be freely preserved and published
Not all source code is public. Some is protected as a company secret or by copyright. It may contain personal information. Some information should not be widely released for safety reasons.
Online services often depend on user data and systems operated by other companies. Anyone outside the operator may be unable to preserve the whole service. The fact that something is worth preserving does not mean it can immediately be published for everyone. UNESCO’s charter calls both for access to materials and for protection of people’s rights and confidential information.
It is also difficult to preserve every piece of software in the same detail. Choices must consider social influence, new technology, regional and language differences, and the risk of disappearance. The process for making those choices should be recorded. Otherwise only products from famous companies may remain in history while work made by small communities and individuals disappears.
What should be preserved when AI writes code?
AI is being used to write source code more often. The amount of code that could be preserved will therefore grow quickly. At the same time, finished code alone makes it harder to know who decided its purpose, what people checked, and which alternatives were rejected.
This does not mean that every intermediate result from AI should be saved without limits. The released source code and its change history should come first. They can be accompanied by the conditions for use, the method used to check correctness, important design decisions, and a record of where AI was used. If an input contains personal or confidential information, protecting it comes first.
There is a problem with the claim that future AI will rebuild old code, so preserving it now is unnecessary. AI cannot reliably recover a fact that has been lost. It may create code that behaves similarly, but that is not proof of the code that was actually used or what its creator thought at the time. Even if AI helps reconstruction in the future, original material will still be needed for comparison.
The number of archived files is not the only measure that matters. Which version was preserved? Who made it and where did it come from? What rights govern its use? How much of the environment needed to run it was recorded? Are we preserving the ideas of many kinds of people, not only famous works? When apps and games are passed on with this information, they can become cultural heritage from which people in the future can learn.