Why there’s no such thing as metadata
I like Jean-Francois Lyotard – mainly because of his name, I think. And I tend to think of him every time I hear people talking about metadata. Lyotard had a problem with meta-narrative – grand narratives that pull us all together and make sense of the world. Lyotard said this class of thing, stories about stories didn’t really exist anymore. There shouldn’t be grand narratives that we treat as special and authoritative – there are just narratives. He argued that metanarratives should give way to petits récits, ‘localised’ narratives that focus on individuals and interpretations – they can be shared, but their status is less than a universal truth.
I mention this because when people define metadata they often say it’s ‘data about data’, and this definition makes me think of Jean-Francois and Bertrand Russel and his ‘turtles all the way down’.
I don’t like the idea of ‘data about data’ – I think it’s bordering on tautology. It subjects data to an inferior status beside the content. It might even be taken to imply that it’s possible to produce meta-metadata. Wouldn’t we all be better off if we just started talking about data?
What’s at stake?
I’ve said it before, words matter. Approaching data as if it were metadata – separate and distinct from the good stuff, relegates its status. Data becomes a secondary thought all too often. But as web technologies evolve, the data is going to be the most important part of the web. We all like the idea of automation, but rarely does a discussion of data make it into a content strategy.
The more we expect of automation, the less we can ignore the role data plays in the modern web.
The internet of the future should be like an iceberg, the visible surface is only a fraction of the story. This has common-sense benefits. Progressive disclosure allows us to build a web that is useable. But the good stuff isn’t always on the surface. Data allows us to build and remould web sites and services. It allows us to re-use content. Having a comprehensive approach to data solves problems that weren’t foreseen. It just makes sense.
Drop the meta
The limited scope that a ‘data about data’ mindset encourages is another reason to drop the meta. Data should be linked. We should be spending time to order and structure our data so that it can be shared. Modeled, structured data allows this.
Structured data enables communication, both within and between systems. It empowers content owners, giving them more options and greater potential to play with their content. We can build models, ontologies to describe our content. And we can do this from multiple perspectives, complementary representations. We can maybe do what Lyotard was talking about. As he said:
A self does not amount to much, but no self is an island; each exists in a fabric of relations that is now more complex and mobile than ever before.
Maybe if we stop treating data as something less than a body, if we spend the time and effort required to understand our content and the ways it can be described, we’ll be in a better position to power the web of the future.