Tagging and Relevancy

I was reading this great article by Thomas Vander Wal on the recent boom of tagging content, and it got me thinking about my own opinions on metadata, classifications, taxonomies and content relevancy in general.


Now I’ll freely admit that the first time I noticed tags reaching more of a critical mass (del.icio.us, Flickr, Technorati) my initial reaction was a negative one: metadata soup! Although tag clouds certainly have visual appeal on a UI level, most (IMHO) also seem to suffer from what early Photoshop and UI wunderkind Kai Krause referred to as ’747 syndrome’- an interface/usability pattern that overwhelms more than informs (i.e. the console of a 747 jetliner) and requires steep investment from the user to browse and/or use effectively. I don’t want to parse a cloud to find something quickly- and can get really annoyed when I visit sites/blogs that only use tags, as although tags feel like navigation, in practice it’s more like exploration (IMHO). Sometimes exploring content is fun, of course- but I’m not always in a ‘nose around’ kind of mood. Who really is?On the plus side, the greatest benefit I can see for tagging schemes are that they expose a given content item to larger, cross-community relevancy patterns outside the myopic site/application/resource it lives within. Local hierarchies simply can’t compete on that level. Very cool end result when navigating this shared ‘tagspace’, but will different interpretations of tag terms (and the ever-growing morass of tags shared across communities) dilute the long-term effectivity of this new browsing meme? For example, the tag ‘service’ could mean public utilities to a consumer, wsdl links to a developer, military branches to a government official, et al- interpretation definitely affects relevancy. Does a given tag/keyword mean the same to you as it does to author X on site Y as interpreted by consumer Z?’Folksonomy’ efforts offer a twist – i.e. letting content’s consumers categorize your data – but this further subjects classification of content to third-party interpretation and/or misinterpretation. There are benefits to ‘folksonomic classification’ of course- one would tend to trust data from known sources/quantities (and misinterpretation could possibly be construed as a sign that the content itself has flaws or unclear focus) but I’d still prefer the content owner- who ostensibly should understand their content’s intent and purpose more clearly than anyone else – be the primary first step in the process.At the end of the day, when I mull over Thomas’ article and whether it’s worth the effort to jump on the hype-wagon and start implementing tag-based categorization myself, I still have doubts these ‘open social taxonomies’ will ultimately help consumers to navigate and target content effectively. I just keep stalling on the dilution problem- without a way to at least standardize tags across resources (example- ‘auto-fill’ UI elements that suggest standardized strings/terms based on your initial keystrokes)- I can’t help but feel that tags will be negatively impacted as more and more interpretations of what a given ‘tag’ actually represents are entered into the system. Maybe I’ll implement tags here soon, but honestly I’ll probably watch things develop a bit more first and make a call later. Updating my crappy pumpkin-colored Movable Type default template here will be the first use of any free time, honestly. ;-)But enough braindumping- what do YOU think about tagging?Passing fad riding on a big hype-wave, or navigational meme of the future?