A Chance of Tag Clouds
Pet peeve on the 4th of July weekend. What exactly is this?
For those of you who think it’s performance art, I think you might be right, but let’s get you some facts first. Tags are all the rage right now and it’s a good thing. Google is a fine way to data mine, but it’s slow and and it doesn’t know who I am. Wow, I can’t believe I just said Google is slow.
It is slow. Go look at the popular page on del.icio.us right now. It’s a near real-time list of pages people have added to their del.icio.us bookmarks and it’s ranked by the number of times the page has been added. Show me the equivalent on Google and remember that del.icio.us has a decidedly nerd demographic… and I’m a nerd.
Unlike Google, anyone can add any tag to anything. Go chaos. Thing is, trends emerge after time. Common tags grow in popularity and we again demonstrate that while human beings really want to be unique, we just aren’t.
This brings us back to the photo above — a tag cloud from Flickr. It’s the list of all time most popular tags on Flickr and I know what you’re wondering “Where’s the sex tag?” Yeah, they probably edited that out..
My question remains. What is the purpose of a tag cloud? It’s more interesting than a bulleted list, but as user interface goes it’s a aflutter with clutter. If you tell me this is intended to a casual interface for browsing tags, I’ll buy it, but if it’s intended to more useful than a stumblable interface, I’d like to hear your theories…
To be honest, I'm not sure how useful the tag cloud is on Flickr. Photos aren't the best sort of data here.
As you pointed out, tags are very good at tracking trends, so to find the best use of a cloud you have to think about situations where trends need to be spotted quickly.
An example would be a database of customer complaints, each complaint tagged appropriately to indicate something about its nature. A tag cloud may help support reps to spot trends they otherwise might have missed as tags amass and grow in size.
I don't think it has much use except as a browsing tool.
On the other hand, I think there's a big difference between browsing other people's aggregates (delicious & flickr) and browsing your own stuff.
For example, if I'm looking through my 'documents' folder for a file I know I worked on, it would be awfully useful to be able to, with a few swift clicks, whittle down the selection by what I may have tagged it with.
I wrote a bit about this and then realised that it sounded awfully similar to what MS seems to be doing with WinFS. So who knows...
I think that in general, a simple list sorted by usage would be more easily digested than the alpha-sorted list with visual weighting showing usage. The idea of the sort (IMHO) is that you sort on the most useful facet of the data.
You can file tag clouds (heat maps, weighted lists, folksonomy lists, whatever you want to call them) under 'cool concept, but not very practical'.
A cooler concept might be implementing an XHTML micro-format for weighted tag data. That would at least give individuals (and bots) a fighting chance of processing/displaying the data as needed.
As a side note, these tag clouds are not very accessible either. Here's a more detailed analysis of some of the shortcomings:
http://www.alttags.org/archives/2005/02/27/42/
I don't really see the point of tag clouds beyond being more aesthetically 'interesting' than a standard list. As Kirk wrote, I see them as a neat concept, but not exactly practical.
A bit behind the times, but this caught my interest. I think the tag cloud was an excellent idea. It organizes things in two instantly recognizable ways; alpha, which is great when you know what you're looking for, and intuitively (via visual weight), when you don't.
I'm sure you've had the experience where you're looking for something but you can't remember the right name, or you have another name applied to the concept than what its tagged as, or you forgot what file, or where it is. Your brain can pick out that right word a lot faster than you can, especially if the set of inputs is non-homogenous in appearance, plus the visual weights give you a wierdly human experience with the set of data.
PLUS, and this one should appeal to you, Rands, there is always the chance you might stumble across something totally new, unexpected, and interesting. Clicking on tags on personal content pages like MySpace and OkCupid has led me down some fascinating dark alleys of human psychology. I bet if the tagged data was actually something worthwhile, I'd find even better new input to hold my attention.
And hey, back to the idea at hand, its still in alphabetical order if you DO know what you're looking for.
