Open Source Software

Open source software is software distributed under a licence that grants users the rights to freely use, study, modify, and redistribute both the compiled software and its source code. The Open Source Initiative (OSI) maintains the Open Source Definition (OSD), the widely-adopted formal standard: a licence must meet all ten of its criteria before software distributed under it qualifies as “open source” in this precise sense, not merely code that happens to be publicly viewable.

The Open Source Definition’s ten criteria

Per the OSI, the ten criteria are: (1) free redistribution — no royalty or resale restriction; (2) source code — must be included, in a form a programmer would prefer to modify, not deliberately obscured; (3) derived works — modifications and derivative works must be permitted under the same licence terms; (4) integrity of the author’s source code — a licence may require modified versions to carry a different name or version number, but may not forbid distributing modifications as patches; (5) no discrimination against persons or groups; (6) no discrimination against fields of endeavour — a licence cannot, for example, forbid commercial use or restrict use to non-commercial research; (7) distribution of licence — rights attach automatically to every recipient, with no separate licence to sign; (8) licence must not be specific to a product — rights survive extraction of the software from the distribution it came in; (9) licence must not restrict other software — a licence cannot impose terms on other software distributed alongside it; and (10) licence must be technology-neutral. A licence that fails even one criterion — most often the non-discrimination clauses — does not qualify, whatever its marketing calls it.

Open source vs. free software vs. source-available

The Free Software Foundation’s (FSF) Free Software Definition predates the OSD and expresses closely related ideas through four essential freedoms: to run the program for any purpose (freedom 0), to study and modify it (freedom 1), to redistribute copies (freedom 2), and to distribute modified versions (freedom 3). In practice the great majority of licences satisfying the OSD also satisfy the FSF’s freedoms and vice versa — MIT, Apache License 2.0, and the GPL family are recognised by both bodies — so “open source” and “free (libre) software” (often combined as FOSS or FLOSS) describe substantially overlapping sets of licences. The real difference is framing, not licence coverage: OSI emphasises the practical, collaborative-development case for open source, while the FSF frames it as a matter of user freedom and ethics.

“Source-available” is a distinct and narrower category. The source code is published and can be read, but the licence imposes at least one restriction the OSD prohibits — typically a field-of-use limitation (no production use, no competing service) or a delayed conversion to a genuinely open licence after a fixed term. The Business Source License and the Server Side Public License are commonly cited examples: both publish source code but are not OSI-approved, because each restricts use in ways criterion 6 (no discrimination against fields of endeavour) does not permit. Source-available code can be read and learned from, but it does not carry the redistribution and modification rights “open source” implies, and should not be described as open source software.

Relevance to open science and research software sharing

Releasing research software under an OSI-approved licence is one of the concrete, verifiable open-science practices increasingly expected alongside open data and open access publishing. See open code for the closely related, narrower practice of publishing a specific study’s analysis code for computational reproducibility. The FAIR4RS Principles extend the FAIR (Findable, Accessible, Interoperable, Reusable) framework to research software specifically, and an unambiguous open-source licence is a practical precondition for the “Reusable” criterion — software with no licence, or an unclear one, cannot be safely reused by other researchers regardless of whether the code itself is visible. Recognition of this work is tracked separately from the licence itself: see open-software contribution for how open-source research code authorship is credited (the CRediT “Software” role, formal software citation, Research Software Engineer career frameworks) and software paper for the citable-publication route some researchers use alongside a code release.

Licence choice for research software differs from licence choice for research data and text. Creative Commons licences are the default for data and narrative outputs, but Creative Commons itself recommends against applying its standard licences to software — they are incompatible with major open-source licences and do not address patent rights or source-code-specific concerns — endorsing CC0 for public-domain dedication of code instead and directing software licensing toward OSI-approved licences such as MIT, Apache 2.0, or the GPL family.

For the institutional decision of whether a university technology transfer office should release research software under an open-source licence, pursue traditional IP licensing, or use a dual-licensing structure, see Open Source Software Licensing in University Technology Transfer, which covers that commercialisation-strategy question in depth — this entry covers only the general concept of what makes software “open source.” A related but distinct concept is the Open Source AI Definition (OSAID), OSI’s 2024 extension of the same open-source framing to AI models, which additionally requires disclosure of training data information, not just code and weights.

References

• Open Source Initiative, “The Open Source Definition” (opensource.org/osd).
• Free Software Foundation, “What is Free Software?” (gnu.org/philosophy/free-sw.html).
• FAIR for Research Software (FAIR4RS) Working Group, “FAIR Principles for Research Software (FAIR4RS Principles)” (2022).