Just posted this to Sig-IA in reply to somebody wanting some examples of good tag clouds (see also my earlier venture). I’m sure the following will be wonderfully arcane in about 10 years time.
I was looking at movietally.com the other day. While it’s not exactly a shining example of good design overall, the use of the tag cloud struck me as particularly good when applied to the movie pages.
The cloud here is much more effective than reading a synopsis. Through the power of user-generated tags, I can also get clues about the film I would not otherwise obtain (eg lots of users tagging a film as “boring” or “left wing”).
The del.icio.us tag cloud is a lesser example, but still good: The cloud tells you instantly that del.icio.us users are all geeks. No “baseball”, “makeup” or “upholstery” here. Job done.*
A more common example though would be for news sites where you can get a feel for the kinds of stories that are prevalent at any one time. I’ve always felt these to be of more marginal use since they tend not to be using user-generated tags, so the insight they provide is limited – they’re serving more as another type of navigation. The same applies to tag clouds on blogs – only as effective as tags and often don’t reveal very much. I see the Observer blogs have dropped their tag clouds in favour of simple category navigation. The IA in charge of that site (Ben Hammersley) is a bit of a god in my opinion (and smokes a pipe), and I can see that’s a good move.
On a slightly more analytical note about the size of the tags in the cloud, I would think most people would not parse out more than about four attributes: present (in the cloud), important, less important, insignificant and absent (the tag is not in the cloud). Differentiating the sizes by more than that would be futile.
Can’t find any other good examples. Maybe they’re mostly awful.
* Since I wrote this, I realise that this is somewhat facile. In order to benefit from a tag cloud, you need to understand the system that generates it. In this case, it’s the tagging of links by millions of people. That’s not always the case. I think I should have made this aspect clearer.