Tag types

I’ve been thinking about information architecture a lot recently, in terms of how important it is for web sites (even simple web sites), what IA features a CMS can and should offer. Web design has gone through several major iterations over the last 15 years, each of these adding a new layer of sophistication, or, at least, as new aspect of endeavour, as technology, budgets and, most importantly, expectations, from both users and clients, increase. IA has always something that has sat at the fringe of most web design projects (moving closer to the centre the bigger the project, obviously), but it is my opinion that it could be something that takes a much more central rôle in the next big iteration of the web.

But I’m not planning on talking in detail about IA today (although I am planning a series of posts). Today, we’re talking tags.

What’s it all about?

With any luck, the connection between tags and IA is obvious. I’m not 100% sure when tags were first introduced and became popular on web sites, but I’m guessing it was around five or six years ago. After an explosion of popularity, especially associated with social media and UGC, enthusiasm seems to have waned somewhat, although I know you still see tagging around a lot (hey, I’ve even got a tag cloud on my sidebar at the moment!)

So what happened?

Don’t get me wrong—I’m sure tags are here to stay. I think the main thing that’s changed is that site owners, developers and designers are now a lot more realistic about their promise. Tags (especially user generated tags, known in IA circles as ‘folksonomies’, i.e. crowd sourced taxonomies), are clearly not the be all and end all of organising large amount of information*, for a whole bunch of reasons.

I think there’s a general idea floating around that collectively, people will always come up with the best, most optimal solutions possible. Nothing could be further from the truth**. Although individuals may not behave this way, large groups tend to produce bland results, without much in the way of real insight or depth. This is reflected in crowd sourced tags, often in the form of a classic long tail—massive weighting on very broad, fairly meaningless tags, followed by a huge long tail of tags far too specific to be useful in most situations. People also often use the same tag to mean different things (such is most language) or describe the same things with completely different tags. So you end up with a bit of a mess.

However, all is not lost…

I don’t think tags are a write off at all. As is, they remain useful in certain areas, such as***,

• reinforcing search and indexation (especially image search)
• displaying related content (à la many blogs or last.fm’s music streaming service)
• giving a quick overview of a site’s topical content (with a tag cloud)

However, as an efficient, meaningful way of organising content, I think they are pretty much broken. But we can change that!

Strategies for change

The one that springs to mind first is administrator moderation. I think that’s useful, but not particular interesting (and for large, crowd sourced tag systems, not very practical). What I’m much more interested in is the idea of having different types of tag, making not all tags created equal after all. A simple idea, but probably quite powerful, I think (maybe it has already been tested, I don’t know).

There are a whole bunch of different ways you could differentiate tags in a system. The simplest would be to distinguish between tags entered by administrators and users (and possibly between different levels of user, if you have them). But there would be no need to stop there. You could make distinctions based on,

• tag length
• number of words or tokens (if you allow them)
• the presence of certain words
• correct spelling or obvious shorthand
• the presence of punctuation
• the number of similar tags already in the system (this could act as a direct or reverse correlation)
• how many times a given user has already entered that tag
• etc.

The next question would be what you actually do with these distinctions when you have them. I guess what we are trying to do here is create some kind of score of usefulness for a particular tag, but that in itself is not useful unless we put it to use somehow. Again, something obvious would be to weight tags differently according to their score when they are used in search or picking out relevant content. Any thoughts on any other uses?

Summing up

I should be clear—although it may not have been apparent earlier in this post, I am a great fan of tag systems, especially crowd sourced systems, and I believe that they have a great potential (both for making the life of the  information architect simpler and in offering a more useful experience to the user). I’m excited about this idea of scoring tags or creating types of tag, as I think it could help that potential become more of a reality.

Renaissance already has a tag system, based on simple tags, and can produce all the usual things to do with it (tag clouds, etc.) I hope to start to develop these news ideas with it soon, however… and we’ll see what happens.

[Don't forgot to tune in next week for the first guest post on this blog. Fran Alexander, taxonomist extraordinaire and writer of the excellent taxonomy and information architecture blog VocabControl, will be waxing lyrical about IA.]

** Yes, specific, well chosen groups of people can achieve amazing results working together—but not the mob.

*** I’m sure there are many others—comments please!

No related posts.

Tagged as: Information architecture, structure, tags, user experience