Friday, January 12, 2007
Tags and Metadata
On Monday Dave Sifry of Technorati posted Happy Taggiversary, an entry marking the second anniversary of Technorati tags. In it he announced the launch of tag pages, a kind of 10x10 of semantic tags assigned to various blog entries around the web for this hour. Instead of 10x10's keyword/picture relay, we get a cloud of the tags themselves.
I'm interested in the response to Sifry offered by Matthew Hurst at Data Mining. Hurst contends that tags, whether assigned by authors or by third-parties, constitute object data rather than metadata. Because search engines easily conflate the semantic content of tags for the semantic content of a blog entry itself, tags are more appropriately identified as object data. Hurst differentiates textual objects from non-textual objects; for the latter, semantic tags are less likely to be confused with the object itself, as with an image, for example.
The questions, then: Are tags metadata? Or are they object data? Are they both? Do they function or perform differently for textual objects than for non-textual objects (i.e., iconic, sonic, or filmic objects)? Are tags always/ever distinct from keywords (a confluence of which appear in the text itself)? Why might it be significant to distinguish keywords (capta?) from tags--to hold them apart, if momentarily?
In one manner of thinking, much of this rolls back to just how strict we want to be with metadata as a concept. Metadata: data about data, yeah? Or data that re-associates or re-assembles other data and things. Or data that, in and of itself, interrelates. I don't have well-formed answers just yet, but I'm inclined to accept that tags are metadata, particularly when tags are understood to be those contingent wrappers (as Vander Wal explains it) that shuttle new media objects into still-developing relationships. Yet, as a microform, where tags have a 1:1 relationship with the thing named, they can be understood as object data, too. The interest in tagging practices, in tags as authored, and in folksonomies, however, might not be as pronounced if tags were object data alone. Because tags bear out something like a 1+1:n relationship with the thing named, to my mind, they do something else, something more, something meta-.
Posted by Derek Mueller at January 12, 2007 10:00 PM
to Methods
I think the answer to whether the tag is metadata or object data or both depends on the activity within and the purpose for which it is being used. It may sound simplistic, but tags only "shuttle new media objects into still-developing relationships" when they're being used to do that. From the perspective of someone searching Google for information about, say, genre, or podcasts (ewm tags), they may perform an entirely different function, like leading someone to one--but not the whole constellation--of your so-tagged blog entries.
In this view, tags' uses and functions are malleable, dynamic, rendering them open to appropriation by and within very different systems of activty.
Posted by: Lance at January 15, 2007 8:11 PMI'm with you on the both/and, Lance. Tags can function as object data and as metadata. Why not? But I'm less settled on the degree to which intentionality has much to do with tipping that function in one direction or the other. But you seem to be applying intentionality to the search: what am I looking for? I can imagine a scenario in which using the searcher's sense of purpose would become too teleological--constrained by the end. In their function as metadata, tags, I think, potentially circumvent that purpose-driven model and instead hold it open to still-developing relationships, to associations which have not yet been traced and may never be. They're latent and potentially relevant.
Posted by: Derek at January 15, 2007 8:24 PMGood point, Derek. The network logic that tags-as-metadata both reflect and sustain may index a massive, more or less self-organized system (the blogosphere, e.g.) that'as an autopoietic process--might as easily be said to 'use' us by getting us to create and maintain it as much as we use it to serve our needs. In this sense, the system could be considered operationally closed, in which case people's reasons for using it (or for following a tag where it leads) aren't very important, if at all. It would also be latent, as you say, since the system/network would not be defined by any static structural qualities but, rather, simply by the production and maintenance of relationships and distinctions, irrespective of the form they take (I'm sort of channeling Luhmann here).
But by the same token, it is the purposive use of an artifact like a tag (it is an artifact in the sense that, despite its being 'virtual,' it must always be materially mediated) that both reflects and sustains other immeasurably complex, nested, and interrelated activity systems ranging in scale from, say, an individual classroom to the whole enterprise of American higher education. And precisely which system is being sustained by the use depends utterly on the perceived purpose for which it is being used. In that sense, then, what tags 'are' depends on how they're purposed--after all, the Riverside Shakespeare is a great read, but it is an equally great doorstop and a not-half-bad portable writing surface.
Maybe it just depends on what angle you come at the question from.
Sorry for the long comment.
Re: the angle. Yep, I think so too. It changes significantly depending on whether we frame tags as artifacts or as agents (or both). Just how visceral can a tag be? And still all the various species of tags swim in the more or less similar code-brine of digitized sociability.
I'm not sure I follow your last example about the Riverside Shakespeare. It seems like the use-value of a tag is shifting into the domain of affordances. I mean that the book-as-doorstop or book-as-writing-surface is more like an affordance than a purpose generated through tagging, right? Guess I should think on just how it is that tags repurpose objects and thereby involve them in the (discrete? closed?) systems you mention.
Oh, and no concerns about the long comment. I appreciate the chance to work/play through this stuff.