Content, data, metadata
June 1st, 2007 by David Weinberger
Rob Styles usefully disambiguates these. Since I have rightfully been charged with willful abuse of the term “metadata,” I found his operational definition encouraging:
 Metadata is different from content and data because most of the time we don’t actually want it. We step over it, often not noticing, on our way to the content we really wanted.
And now that just about anything can be used to find anything else — the content of a book can be used to find the year it was published, once that content is online and indexed — we all just got smarter.
Styles’ point is superb, but we need to look at the obvious (and perhaps not so obvious) exceptions. E.g., playlists – which might have begun as means to the end of accessing songs, but then become examples of taste that can be shared, compared, used for social identification, etc.
Or someone’s top ten list of films – all the “content” in the list is factual – merely the titles of actual films. But by saying “this is an expression of my taste,” one endows the grouping with something of value and possibly uniqueness, which some might claim as copyrightable.
Metadata, in its function, needs to be invisible, minimal, to not obtrude. But since everything is miscellaneous and reversible, one can no longer say with confidence that something is merely metadata.
Oh well, we had nice neat divisions for a day or so. Thanks a whole heck of a lot, Tom!
But seriously: Good points, of course.
Doesn’t it seem that we’ve totally flipped the Cartesian idea that being clear and distinct is the mark of truth? If it’s clear and distinct, we just haven’t thought about it enough.
Dave: The Wikipedia definition of “metadata” is one-dimensional, utilitarian, and short-sighted. Someday, we may understand metadata as the essence of how data comes into being – not the other way around.
As I see it, Metadata is an ontology – it exemplifies the //process// of dynamic and conceptual reasoning. Metadata seeks better questions instead of better answers. Metadata is liminal, en route, emerging, becoming. Data is fixed, static. Metadata is process. Data is product.
David, your suggestion that we’ve upset the appledescartes – “flipped” it – might take more than a comment box to address. But I do think there’s one distinction that might help clarify some of the discussion of the book, including some of the interesting and thoughtful tuggings at the idea of tags and taxononies going on at the Well. That’s the distinction between metadata that restricts itself to a careful description of the properties of the data to which it points – and does no more – call it descriptive meta-data – and on the other hand, metadata that interprets the data, evaluates or judges it. The difference between saying the book has 278 pages, and saying it “gives Aristotle a swift kick downstairs.” The former sort fulfills the narrow definition of metadata. The latter actually compresses statements about the meaning and value of the book into tags, one or few-word signifiers that cannot possibly do justice to the meaning they are supposed to convey.
The former sort are the silent, obsequious metadata, the latter are the vociferous crowd labelling, catcalling, pointing and whooping it up in the sloppiest possible manner.
To the extent that both of these can be said to be metadata, general statements about metadata will tend to be wobbly and might defy coming to some systematic, top-down understanding of the argument of the book. The book, then, appears to emulate its argument, and vice-versa.