Google UX Talk Thoughts, Once Removed | Main | Prohibitions and Restrictions on Items Mailed to the United Kingdom

February 02, 2005

Tag Inversion - When Metadata Isn't

I really should be working, but last night, I found myself staring at the ceiling as I thought about how social free tagging is inverting the standard tagging behavior.

Typically when you tag, you assess some object for it's attributes, and tag it with those attributes. Those tags are invariably metadata -- information ABOUT that object/thing. The tags are derived from the nature of the object themselves.

What we're seeing on Flickr with tags like "sometaithurts" or Technorati's "10placesofmycity" is less about metadata then it is about... community? group-forming? art projects? I don't quite know. But what I do know is that we have an inversion. Instead of people tagging an item based on the qualities of that item, they're instead generating content based on the qualities of the tag.

Whether the tag or the content comes first might seem like a subtle distinction, but I think it's a crucial and important one. These tags are no longer simply keywords that describe something about the content, but instead are the reason for that content to be.

(If this post were to have a bibliography, it would be made up of Adam Mathes' excellent essay on folksonomies, and Dave Weinberger's "cool stuff with tags".)

Posted by peterme at February 2, 2005 04:54 PM


Well, it's clearly "metadata", but it's along the lines that Dan Hill described here.

It's what he's describing in the "esoterica" facet: using a tag-like attribute to make adhoc groupings of otherwise unconnected things. I like his eample of the set of "great 'second-string' guitarists in early-70s US rock bands". The "content" here is the connection between those guitarists. This is exactly "generating content based on the qualities of the tag" as you say, Peter.

It's interesting, "tagging" already seems to have a sort of, er, grammatical style to it. Just like linking to something can provide a sort of knowing sideways reference to something in the middle of a sentence (see any Metafilter post, for example), "tagging" is starting to feel like a grammatical structure. I might associate some things by tagging that I wouldn't want to enumerate as hyperlinks in a blog post: those two grammars are for different things.

Or, put another way, this is why no one bothered to create a blog called "sometaithurts". Part of the point is the effortlessness of just slapping the tag on a photo and knowing what the result will be.

Another nice example are the spontaneous groups of publishers/subsribers that form by simply agreeing to use a specific tag. The "group" has now outlived the community blog it came from.

Posted by: Andrew at February 2, 2005 12:02 PM

Or like this. That is quite a good idea: assign tags to the audience that you know can make use of them, and a collection of content springs into life.

Posted by: Andrew at February 2, 2005 03:41 PM

Often metadata becomes the data. But in the instances you show the value of a tag in a folksonomy allows the user to find information of interest. It is a an emergent term for a adaptive community that allows a community to congregate around the term.

If only every tool were this adaptable.

Posted by: vanderwal at February 2, 2005 08:47 PM

I think what you're seeing is that (1) any system can be gamed (2) any grouping mechanism becomes a communication mechanism. I would speculate (and one could probably get some convincing numbers with a few days work) that most of these tags really are descriptive-of-item tags. Not denying that there are local inversions - just that they're probably interesting in how exceptional they are.

Posted by: Mark Eichin at February 3, 2005 05:08 AM

%syn(Cool|Nice|Rulezz)% %syn(blog,|portal| site ! I)% hope to make %syn(my own|own weblog|my diary)%, not worse than yours ;)

Posted by: Jonathan Riker at February 3, 2005 08:59 AM

What interests me in all of this is the practical implications for someone trying to spread their meme, as instantiated by tagging behavior. So, let's say I want to get people to see the world the way I do. I could generate my own content tagged as I see appropriate. It is probably a requirement that I do at least some of that.

However, if I really want to get the meme out, I will do a lot of tagging of other people's content. Why? Well, part of getting your meme to spread is getting people to see the world differently, in other words, apply different tags to things they may already be familiar with. I gave some thoughts to the practicalities of how to do this with and technorati here.

Posted by: Bud Gibson at February 5, 2005 02:54 PM