The user wants to browse content by popularity in a visually appealing way.
A tag cloud is a list of tags, where the font size of each tag is larger or bigger depending on its weight. Weight in tag clouds can be represented in three different ways:
There are several opinions on how tags should be ordered. Examples of ways to order tags are:
Tag clouds helps visualize semantic fields; how some categories have greater importance than others.
It can also help give an impression of what content is to be found on a site and which categories of content the site is focused on.
I’d be interested in seeing some commentary about how people use this, how often they use it, what it does that other things don’t, and, particularly, how it justfies the relatively large amount of screen real estate it uses.
Personally, looking back, I just realized that I click on something in a tag cloud perhaps once per year, and usually ignore them completely.
Tag clouds don’t work and don’t really make sense. Why? Because there is absolutely no reason to say that a visitor is more interested in, or more likely to want to click on, or needs help finding, things that others have found interesting.
No one searches for things on Google because other people search for them.
A tag cloud would make more sense if the visual clues said something interesting (there are new items here, this is super cool, this is classic stuff, etc).
I am aware that many developers and designers want to implement tag clouds just because it is popular, but tag clouds aren’t very useful. Especially when you have a large number of tags. It looks messy, and doesn’t provide any information that can affect user’s choice.
Just imagine the Flickr example placed in sidebar width of 200px.
I prefer lists, such as Categories (with number of posts), most popular posts, most rated posts, etc.
I’ve always been a firm believer that the size of a control should be directly proportional to its intrinsic value. In order to make a tag cloud truly useful for its size – tags would have to be limited to a particular context… and cross correlated with an extensive user profile. In the end, the specificity needed to ascribe meaning and purpose to the navigation potential of a ‘cloud’ would negate its nature and scope (i.e. Stumble Upon). The best bang for the UI buck would be implementing something similar to the iTunes feature – ‘People who bought this album also bought…’ with a list of a half dozen items listed by popularity.
I have used a tagging system to store all the files on my laptop for years and I have found it to be extremely useful for finding files without having to remember where something is stored.
Tags are great because they allow content to be grouped together without having to be physically stored in the same place. For example, I can have a presentation that is tagged with “presentation” “sales” and “work”. When I click the “work” tag everything I have tagged with work shows up.
What I feel most tag clouds fail at is the fact that they don’t allow users to search using multiple tags. So in my previous example, show me everything that is tagged with “work” and “presentation”. Having the ability to preform multi-tag searches is the key to tag clouds being more flexible and scalable than other forms of file browsers (like Miler Columns and File Trees)