Introducing the FAIR Principles for research software

In 2016 the publication of The FAIR Guiding Principles for scientific data management and stewardship supported a vision where valuable scientific outputs are made FAIR by becoming more Findable, Accessible, Interoperable and Reusable. From the outset, the FAIR Guiding Principles were intended to be applicable to many kinds of digital assets. Increased understanding of the importance of research software in research has catalysed application of the FAIR Guiding Principles to this type of digital asset. Community-endorsed FAIR principles for research software were released in 2022 by the FAIR for Research Software (FAIR4RS) Working Group (WG), jointly convened by the Research Software Alliance (ReSA), FORCE11, and the Research Data Alliance (RDA). This milestone reflects the maturation of the research community in understanding the benefits of having FAIR research software, and coming together as the FAIR4RS WG to achieve this. The FAIR4RS WG is a global and interdisciplinary community including researchers, software users, developers and maintainers, policy makers, infrastructure support staff, and funders. The FAIR4RS Principles are relevant to any stakeholder in the research community seeking to increase transparency, reproducibility, and reusability of research. This paper highlights their importance and the positive signals of adoption that demonstrate high levels of community support. It also acknowledges that research software and data discoverability is a long-standing challenge with multiple past efforts; in this sense, the FAIR4RS principles provide an umbrella framework integrating aspects of these existing efforts. The paper outlines the principles and examples (Results), their importance and organisational adoption (Discussion), and the community process used to develop them (Methods).

Methods describes the community-driven process used to develop and validate the FAIR4RS Principles. The FAIR4RS Working Group engaged about 500 people from more than 110 organisations across more than 34 countries, including over 240 WG members. Work began with subgroups (July 2020 to March 2021) producing outputs that informed the principles. These were consolidated into a single report and presented for wider community feedback in March 2021. Draft FAIR4RS Principles were released in June 2021 and underwent a formal community review process for one month. The community-validated version (v1.0) was released in May 2022. The WG aligned with and amplified existing efforts to apply aspects of FAIR to research software (since 2017) and sought alignment with FAIR data efforts, leveraging existing community momentum. Maintenance responsibility for the principles transitioned to the RDA Software Source Code Interest Group, with a plan for the community to reconvene in two years to assess potential updates.

• The FAIR4RS Principles adapt the FAIR Guiding Principles specifically to research software, recognising its unique characteristics (executability, composite nature, continuous evolution and versioning, social coding platforms, package managers).
• Definition of research software includes source code files, algorithms, scripts, computational workflows and executables created during the research process or for a research purpose.
• Principles (summarised): F: Software and its metadata are easy to find (globally unique and persistent identifiers; rich metadata; metadata include identifiers; metadata are FAIR, searchable and indexable). A: Software and metadata are retrievable via standardised protocols (retrieval by identifier; open, free, universally implementable protocols; authentication/authorization where necessary; metadata remain accessible even if software is not). I: Software interoperates via standards for data/metadata exchange and APIs (uses domain-relevant standards; includes qualified references to other objects). R: Software is usable and reusable (described with accurate attributes; clear license; detailed provenance; qualified references to other software; meets domain-relevant standards).
• Practical examples of implementation: • Comet (mass spectrometry tool): registered in bio.tools with persistent ID and rich, searchable metadata; downloadable via HTTPS; metadata persists independently; uses proteomics community standards; Apache 2.0 license; development history on GitHub; documents dependencies. • PuReGoMe (COVID-19 Twitter analysis): versioned DOI in Zenodo; registered in Research Software Directory with rich, searchable metadata; downloadable from repository; uses standard file formats (e.g., CSV); Apache 2.0 license; GitHub history; lists dependencies. • gammaShiny (R GUI for gamma): deposited in HAL with persistent identifiers (HAL ID and SWHID); GNU GPL v3.0 license; archived source in Software Heritage with codemeta.json including dependencies.
• Early organisational adoption and alignment efforts underway at ARDC, ELIXIR, Netherlands eScience Center, and ZB MED, indicating broad support and anticipated impact across stakeholders.
• Community engagement scale: about 500 participants from 110+ organisations in 34+ countries contributed to the development and validation of the principles.

The FAIR4RS Principles mark research community maturation in recognising research software as fundamental to research and as a first-class digital object deserving FAIR treatment. While some original FAIR data principles directly translate, software’s unique properties (executability, composite and evolving nature, social coding ecosystems) required revisions and extensions to craft software-appropriate guidance. Unlike data, where FAIRification often occurs at publication to an archive, open source software ideally works toward FAIRness from initial development. Community-endorsed principles catalyse shared practices by enabling participation from diverse stakeholders, increasing awareness of challenges and providing clearer routes to implementation. The FAIR4RS WG identified opportunities for future work emphasising standardisation: identifier authority, metadata vocabularies and properties, software identifiers, domain-relevant community standards for software, and identification targets, all aimed at simplifying adoption. Adoption and impact: The WG has facilitated initial adoption reflecting strong early impact. Benefits span stakeholders, from clearer funder requirements to publisher and institutional guidelines, and alignment for repositories and registries. Maintenance of the principles now resides with the RDA Software Source Code Interest Group, providing a venue for continued community input and adopter reporting, with a planned review in two years. Organisational examples:

• ARDC: Updating co-investment policies to reference FAIR4RS for software outputs; developing guidance and a FAIR research software self-assessment tool; engaging platform projects and broader communities; committing to making ARDC software outputs FAIR.
• ELIXIR: Policy encourages FAIR across research outputs including software; leveraging services like FAIRsharing; aligning the ELIXIR Software Management Plan with FAIR4RS; developing training to increase use of Software Management Plans and understanding of FAIR4RS.
• Netherlands eScience Center: Embedding FAIR4RS in calls and projects via Software Management Plans; training and practical guidelines; updating Five Recommendations for FAIR software tools to align with FAIR4RS; collaborating with NWO on national templates for Software Management Plans aligned with FAIR4RS.
• ZB MED: Extending existing open and FAIR-aligned practices (e.g., Five Recommendations for FAIR Software) to better match FAIR4RS; enhancing dissemination and training so more researchers adopt FAIR4RS.

This paper introduces and contextualises the FAIR4RS Principles, adapting FAIR to the specific nature and ecosystem of research software, and illustrates practical application through case studies and organisational adoption plans. The principles aim to enhance the transparency, reproducibility, and reusability of research by making software findable, accessible, interoperable, and reusable. Early adoption by major organisations signals strong community support and potential for substantial impact across the research ecosystem. Future directions, identified by the community process, include advancing standardisation around software identifiers and authorities, metadata vocabularies and properties, defining domain-relevant community standards for software, and clarifying identification targets. Ongoing maintenance by the RDA Software Source Code Interest Group and a planned community review in two years will help evolve the principles in response to practice and feedback.

The FAIR4RS Principles are explicitly aspirational; they provide goalposts rather than prescriptive requirements. Effective application often depends on the availability and maturity of scholarly infrastructures (e.g., registries, archives, identifier systems), and implementation responsibility typically lies with software owners/creators rather than users. Discoverability of research software remains a long-standing challenge, and while the principles integrate aspects of prior efforts, community practices and standards are still evolving. The work reports early signals of adoption; long-term outcomes and comprehensive assessments of FAIR4RS implementation across domains are not yet available.