First published in 2016, in an article that has now received over 6000 citations, the FAIR (Findability, Accessibility, Interoperability, Reusability) Principles have had a significant influence in policy, practice and thinking about research data management and stewardship. The catchy mnemonic has been effective, but the article and related work convey an important message. The fundamental purpose of the FAIR principles is to provide guidelines such that data and metadata relevant to all kinds of research outputs are machine readable and machine actionable. The vision is one in which research outputs can be visited online and at vast scale and the potential of machine assisted analysis can be realised, but with data and metadata that are sufficiently reliable so as to reduce error and quantify uncertainty.