The recent revelations about America’s security agencies have pushed the word ‘metadata’, hitherto familiar only to computer scientists and librarians, into the limelight. Thankfully, I haven’t yet come across one of my least favourite collocations, familiar from dozens of PowerPoint presentations and at least 75,000 webpages. The commonplace definition of metadata as ‘literally “data about data”‘ is true only if we believe, like Humpty Dumpty, that words mean whatever we want them to mean.
Metadata is data. In an oversimplified sense it is data about data. But not literally.
The Ancient Greek prefix ‘meta-‘ means ‘with’, ‘after’, or ‘between’. In the first century AD Aristotle’s lectures on the principles and causes of change were gathered together as ‘τὰ φυσικά’ (ta phusika, i.e. ‘on nature’). His treatises on causation, matter and God as the first mover came next in the arrangement, and were thus known as τὰ μετὰ τὰ φυσικά (ta meta ta phusika, i.e. ‘[the writings] after “on nature”‘), ‘metaphysica’ in Latin. The word became associated with the topics under discussion, and metaphysics ended up as the branch of philosophy dealing with first principles.
During the twentieth-century ‘meta-‘ began to be applied to certain words, in an analogy with philosophy, with the sense of ‘beyond’, ‘above’, or ‘at a higher level’. With this implication, it was first appended to data sets in 1969. But the prefix does not, and never has, imposed any self-referential quality over the word which follows it.