Technorati Tags - Not Quite Meta Data

Technorati announces a refactor of its presentation of tag data. I've not yet looked at it, but something in the blog post caught my eye:

The beauty of tags is that they’re metadata: data about data. What does that mean? Tags actually describe their subject, as opposed to, say, keywords, which just occur within them.

If you do a search on Technorati for 'technorati tags' you'll get hits for portions of posts which contain tag information. In fact, once you do this search, it is pretty easy to construct a search which will produce results that hit tags - not the content of the post. In fact, tags as implemented by the tag microformat, are just markup on text that appears in blogs.

A good blog index will roll up many forms of categorizing blog posts including tags, categories, etc. One thing which blog search engines seem to be uniformly bad at is filtering out tags. This is why tracking discussion about Technorati itself is so hard - all those searches for 'Technorati' that hit their tags! I know one could exclude that phrase - I'm just trying to make a point.

In my mind, tags as they have been envisioned as a microformat to annotate textual content in blog posts is a bit of a mess. The fact that you are annotating what can only be considered object data - not meta data - is a fundamental problem, and contrary to the claim that Technorati makes in the above quoted post. On the other hand, tagging non textual data is less problematic as there is no confusion between an image and a textual tag.

Think of it this way - if tags really are meta data, then why do they appear in search results snippets as object data?

I've commented before on my thoughts on tags.

Comments

[P-Air]

You're obviously dead-on Matt. The tagging infrastructure really works best for audio & video where word descriptors cannot be easily inferred. Until people can broadcast their thoughts to a computer in order to define the images they're looking for, tags seem like the best solution. As for text, there are plenty of text analysis technologies working on addressing issues here, that I don't see much value in adding tags for such.

To highlight your point on Technorati's tagging issues, I recently wrote a post about how a section of my blog that displays headlines fm a Technorati tag feed are themselves being indexed into Technorati...doh! (http://direwolff.wordpress.com/2006/12/10/technorati-self-references-itself/).

[Pat Hall]

But sometimes people do use the words "Technorati" and "tags" in their posts. What then? I think this argument is based on an edge case. After all:

http://www.technorati.com/tag/technorati+tags

Works as advertised. Text is text, it will always be "a bit of a mess," because there is no conceivable way (Microformatopia notwithstanding) to get all users to consistently mark what's "meta" and what isn't; and because, given that state of affairs, tags are still just text at heart.

The boundary between keywords and tags is fuzzy, and will stay that way. But that doesn't mean that the central uses of tags (which, I dare say, are not searching the full text search interface for "Technorati tags") is not useful.

[Greg Gershman]

In part because T'rati tags were really just meant to get people to link back to T'rati.