libraries | play | information | media | policy | culture

2008-03-25

Clay Shirky Is Overrated

It's to Clay Shirky's credit that after three years people still talk about his piece Ontology Is Overrated: Categories, Links, and Tags. It makes sense we still talk about it because Shirky is clearly smart, is a good writer, and makes a compelling argument. His piece certainly made a few generations of catalogers, indexers, and taxonomists lurch forward and spit up their morning coffee—which is never a bad thing. Keeps us from getting complacent.

There are a handful of good critiques of Shirky's argument, most notably from Gene Smith, who takes on the politicized nature of what should otherwise be a methodological argument, and Peter Merholz, who takes exception with Shirky's classification of what classifications are.

However, there are two fundamental flaws in Shirky's argument that I've not seen addressed, one philosophical and one technological. First, philosophically Shirky takes umbrage with the bias imposed by top-down classification schemes. He's certainly right about that. However, bottom-up classification schemes are just as culturally imperialistic.

We are, after all, talking about tags applied by an Anglophone society. Our use of English, which is no less a classification system than any offered by the Library of Congress, informs how we view the world. As McLuhan rightly noted, English's phonemic system has a profound effect on how we learn, relate to each other, and relate to the world, as opposed to how a people who grow up in a morphemic linguistic system, such as the Chinese, view the world. Neither is better than the other, but they are different.

Just because a majority of people who can read and write English apply a certain tag to, say, a picture of Fenway Park doesn't make it inherently better or any less imposed. What about the minority that doesn't think of that word at all when they see Fenway? At least with an ontology, everyone has to learn the same lingua franca to access the same material. Which is why, incidentally, Latin, French, and now English have all taken turns as the languages of diplomacy.

Shirky's argument is also—and has always been—creaky technologically. He neglects to consider, either from intention or ignorance, that the entire Web, which enables his dream of democratic tagging, is built on structured metadata, which is why it will always operate optimally amidst structure.

Web 1.0? Built on Java, which solidified an entire system of programming, called object-oriented programming, whose power lies in their application of classes and objects. By the way, classes and objects, along with relationships, form the holy trinity of ontologies. Oh, and Web 1.0 was also built on the back of database technology. So, for that matter, is 2.0 and whatever 3.0 will be. Is there anything more structured and more precisely labeled than a database table?

Web 2.0? Built on XML, which allows individual systems to label their data however they want, but insists that they provide crosswalks that enable other idiosyncratic systems to talk to them and use their data.

And Web 3.0 might yet be built on RDF, which creates a lingua franca for predicate logic that will enable systems to make inferences, thus creating a "semantic Web." The result, if anyone actually pulls it off, would be a mash-up heaven for all who venture on the Web, whether they know script or not.

Wouldn't that make Shirky happier than unfettered tagging ever could?

2 comments:

thisishere said...

The odd thing about Shirky's stuff is the way that his Web appears to basically be the Web of Bits of Lint from my Pocket, and a Shiny Penny, Shared with a Friend and a Friend of Theirs.

In "Overrated", he derides the Library of Congress for overwritting his personal book organisation in favour of their prescribed taxonomy, and in another piece he talks about how it's not a failing of LiveJournal to make the interests "movies" and "cinema" completely distinct terms, because they mean different things to different people.

Many of his "organisations without organisations" look like small groups of like-minded individuals talking to themselves, basically. I caught one of his talks in England promoting his new book, but the audience was highly non-tech, and I don't think asking a q' about these now that terms like Linked Data and things like mashups have begun to prove both interesting and useful to actual users.

librarian@play said...

Precisely. His version of the Web is is niched and tribal. Which is not entirely a bad thing. I actually do like the notion of folksonomies: Why not let people play in their own corner of the Web and use it to organize their digital lint?

But if that's the only vision for the Web, then it'll actually make us less informed rather than more. In that vision, what's the incentive to explore beyond Things That Interest Me or People Who Agree With Me?

I don't need or want to sift through everybody's lint. That doesn't mean it shouldn't be there, or that some of it isn't valuable. But there needs to be multiple ways to float valuable information to the top.

Folksonomies are one way; classifications, another. Both exert a kind of tyranny over information.

Great pocket metaphor, btw.