Explainer · Plain-language
What is FAIR data?
FAIR is a set of four principles — Findable, Accessible, Interoperable, Reusable — published in 2016 to make research data machine-actionable. Most major funders now require FAIR-aligned data management plans for funded research.
What the letters mean
Findable: data has a persistent identifier and rich metadata. Accessible: data is retrievable via standard protocols (e.g. HTTPS) and metadata persists even if data does not. Interoperable: data uses formal, accessible, shared vocabulary. Reusable: data has clear provenance, licence, and community standards.
Implementation in practice
In practice, FAIR means: DOI for your dataset (via DataCite), rich machine-readable metadata, standard formats, controlled vocabularies, clear open licence (typically CC-BY or CC0), and deposit in a recognised repository.
FAIR vs Open
FAIR is about machine-actionability, not openness. Data can be FAIR without being open — sensitive medical data can have a DOI, metadata, and access protocol while restricting actual data behind authorisation. "As open as possible, as closed as necessary."
FAIR4RS — for research software
A 2022 extension applies FAIR principles to research software — Software Heritage IDs, FAIR4RS Working Group principles, citable software releases via Zenodo.
Key facts
At a glance
- Published: 2016 (Wilkinson et al., Nature Scientific Data)
- Steward: GO FAIR Initiative, Research Data Alliance
- Mandated by: NIH DMSP, Horizon Europe DMP, UKRI, ARC, NHMRC, and most major funders
- Extensions: FAIR4RS (research software), CARE (Indigenous data)
Common misconceptions
What people often get wrong
Often heard: FAIR means data must be open.
Actually: No — FAIR is about machine-actionability, not openness. Sensitive data can be FAIR without being open.
Often heard: FAIR is just for genomic / large-scale data.
Actually: FAIR applies to all research data — including qualitative interviews, survey data, microscopy images, software.
Going deeper
