Code is a key research output — let’s preserve, reference, and value it

Today, Nature published a commentary that stems from key challenges on which many of us have worked for years: a clear call to stop treating software as an afterthought in research and to record, share, and value it alongside articles and data. (Nature permalink)

Over the past decade, a remarkable community across disciplines—researchers, librarians, infrastructure builders, funders, and publishers—has come together to make this shift real.

Why software matters for science

Software isn’t just “data” or supplementary material; it embodies methods and results with a level of precision that text alone cannot capture. If we want research to be reusable and verifiable over decades, we must preserve software, give it stable references, describe it properly and credit its authors—systematically.

This is why we see so many emerging policies that mention source code: journals requires authors to make available the code used in the articles (but often forget to explain how, or propose cumbersome solutions), funders want to see the software produced released under an Open Source licence, Open Science mandates include software, national awards recognize it.

Policy to practice: the Software & Source Codes College

Within the Software and Source Codes College, we have been working for years to implement the software chapter of France’s National Plan for Open Science, around five different areas:

• WG1: Identification and Promotion of Software Production in the Research and Higher Education Sector
• WG2: Tools and Technical & Social Best Practices
• WG3: Valorization and Sustainability
• WG4: National, European and International Coordination and Networking
• WG5: Recognition and Career Development

It is a unique opportinity to contribute to the definition and dissemination of best practices for development, governance, sustainability, referencing, preservation and recognition of research software—and connect them to operational tools that people can adopt today.

From vision to infrastructure: Software Heritage and the SWHID ISO standard

The well known Nosek's strategy for culture change (make it possible, make it easy, make it normative, make it rewarding, make it required) remebers us that policy alone does not suffice, it is important to provide the proper infrastructure that make it possible to implement it.

For source code we have good news: some ten years ago we launched Software Heritage with Inria and in partnership with UNESCO to collect, preserve, and share all publicly available source code for the long term, as a common good for research, industry, culture, and education. Today, the archive safeguards hundreds of millions of projects and tens of billions of unique source files, offering a uniform, technology-neutral view of public code and its development history. Yes, one open, shared, non profit infrastructure for all fields of endeavour, so you do not need to sift through thousands of institutional repositories like what happens for Open Access articles. And it is replicated in a mirror network, to set this precious knowledge beyond the reach of accident (paraphrasing the famous letter by Thomas Jefferson).

Preservation alone isn’t enough; we also need stable, verifiable references for software at all levels of granularity (file, directory, commit, release, snapshot). That’s what the Software Hash Identifier (SWHID) delivers: an intrinsic, cryptographic, platform-independent identifier that anyone can compute and verify.

In April 2025, SWHID became ISO/IEC 18670, a major step that anchors software citation and traceability in an open, global standard—independent of any single platform or vendor.

Make it easy... make it useful!

We worked hard to make it super easy to archive, reference, describe and cite source code in Software Heritage, in a uniform way, no matter what platforms you use to develop or distribute code (did you know we are tracking over 5000 of them in the archive?). It is a matter of a few clicks to get the code archived, and reference it, with full flexibility (a revision, a directory, a file, a fragment of code).

But easy is not enough: there are many easy things we do not do, just because they bring us no clear value. So we worked to make it useful. For this, we started collaborating with HAL (the french national Open Access portal), to create a simple workflow that allows a french researcher, group, laboratory or institution to get automatically the list of the codes they produce, with proper presentation and affiliations (see this example for the LIRMM laboratory); yes, now with a few clicks french researchers, laboratories and institution save a lot of hours previously spent creating and maintaining manually resumes or activity reports.

What you can do now (and how we can help)

Here are concrete steps every researcher, lab, repository, and publisher can take—supported by Software Heritage services and community guidelines:

1. Archive your code (automatically and on demand).
2. Reference precisely with SWHIDs.
3. Describe your software for credit and reuse.
4. Adopt the shared guidance and tools.

These practices are simple to adopt, and—when done collectively—transform reproducibility, attribution, and long-term access for everyone.

Acknowledgments

My deep thanks to the colleagues who have made this journey possible—from the members of the Software and Source Code College, to the team building Software Heritage and the ambassadors helping the community using it, to librarians and repository teams, to publishers and funders embracing software citation. Your work is turning open-science principles into everyday reality for research software.

Let’s make software preserved, referenced, and valued. Join us: adopt the practices above, integrate SWHIDs into your workflows, and help us scale a universal, trustworthy infrastructure for software in science.

#OpenScience #ResearchSoftware #SoftwareCitation #SWHID #ISO18670 #SoftwarePreservation #OpenResearch @UNESCO