The Yew Tree

Abstract: In order to reflect the way people actually use the internet, websites need to cluster their information in customisable ways, rather than arranging it in a hierarchy or imposing grouping on the basis of perceived need. This can be achieved with XML.

Information, like so many other human activities, is structured.  Data, by contrast, is meaningless - it is only structure and metadata which give it meaning.  For example, the string http://www.useit.com only has meaning because we know that http://www. means a website address (so we have metadata about the data string).  Those in the know might recognise it as the URL of Jakob Nielsen's website about usability.  

There are two ways to turn mere data into information: to structure it, and to give it meaningful labels.  But the structure and labels we choose depend very much on our world-view: and when designing websites, you can't guarantee that all visitors will have the same world-view and agenda as the designer.

Recently I was looking at a website which clearly had its own agenda very much in mind, and didn't much care about that of the user.  (Its agenda seemed to be to repeat itself as much as possible in order to ensure its legal responsibilities were covered, without regard for the unfortunate user actually trying to locate information in all of this verbiage.)

So how should information be structured?  There are two models here: the hierarchy and the cluster.  Many websites are hierarchically structured, partly due to the nature of the technology, but mainly because a hierarchy is easier to visualise, because it is an image we see all the time.  Many technologies have clustered information structures (e.g. the Nokia mobile phone menu).   The webring is another good example of clustered information.

Hierarchy:
the hierarchy in action, source: http://classes.kumc.edu/general/wbmodule/anatomy_hp.htm 		Cluster (webring):
webring example, source: http://salveregina.virtualave.net/webring/webring_announce.html

In the hierarchical structure, you have to dig deep to find information, and if you choose the wrong pathway, it is difficult to retrace your steps. (Think for example of the last time you phoned a public body to get money out of them - "Oh no, you don't want this department, you want that department over there...")

vertical boundaries, lateral links

In the clustered structure, information of a similar nature is grouped together, which increases the likelihood of finding the information you want. An example of this is websites which organise their material according to user groupings (e.g. staff, students, visitors). Another thing that is noticeable about this structure is that it is very lateral - links join information across vertical boundaries.

What does this have to do with inclusionality?

Some of the insights offered by inclusionality are:

Actually the internet is inherently non-hierarchical, but we consistently attempt to impose hierarchical structures upon it and within it. The maps at the Internet Mapping Project site were produced by sending a packet along the physical network via Internet Service Providers, which naturally produces a hierarchy (K Tesson & A Rayner, unpublished). What it does not do is find the hyperlinks linking sites together, which would look much more like a web than a hierarchy. Nor does it help us to visualise the clustering of information that we get when using a search engine. Interestingly, the algorithm that Google uses to rank pages includes how many other pages link to that page (see http://www.google.com/technology/index.html).

Jakob Nielsen points out that there is a difference between customisation and personalisation:

Customization
is under direct user control: the user explicitly selects between certain options (a "portal" site with headlines from the New York Times or from the Wall St. Journal; enter ticker symbols for the stocks you want to track)
Personalization
is driven by the computer which tries to serve up individualized pages to the user based on some form of model of that user's needs.

Clearly customisation is dynamic and reciprocal - the user looks at a page, selects the information they require, and then sets it as a preference (e.g. the BBC weather forecast, where you can specify your local region, but change it if you want the weather forecast for somewhere else). Personalisation, on the other hand, is a conceptual frame imposed by the website designer trying to second-guess your needs.

Some portal software is based on the principle of personalisation, some on customisation, and some on both at once (e.g. a portal-style intranet might show you information on the basis of your employee status, as well as allowing you to select or prioritise the information you want to see). Michael Sippey discusses the implications of this at theobvious.com.

XML and dynamic websites

XML is the ideal technology for dynamic websites. (A dynamic website is one you can interact with, e.g. it has a search interface, is database-driven, has a web forum, message board, or guest book; or it is customisable, whereas a static website is one without any interactive features.) Not only is its structure nested (as you can nest elements within other elements), but it is also holey (as you can omit optional elements) and the metadata is located alongside the data (in the form of the element name). In addition, because the data is separated from the formatting, it can be presented it in a variety of ways, which the user could customise for their own needs - e.g. by selecting a different style of display, choosing to show or hide areas of the page, or rearranging the page in a different order. XML is also fractal (similar at the global and local levels) - another inclusional characteristic.

People

The really exciting thing about the internet is the way it connects people with other people. Static websites are all very well, but the most exciting aspect of the web is email and online forums, where you can meet and interact with people from all around the world. It is this community aspect of the web that is well served by the possibility of dynamic interaction with the web page (without which you couldn't have online forums, for example). The emergence of virtual communities and the possibility of complex interaction with them is what makes the web so fascinating. (See A Brief History of the Future: the origins of the internet by John Naughton for a discussion of this.) It is in this context that the concept of the sharing circle comes into its own - "where each member listens respectfully, not to contrived argument to win a debate, but to the heartfelt sharing of unique personal experience, without comment from the others, as the talking stick is passed from person to person, and the ‘holographic’ imagery of the way the world ‘really is’ takes form." (Ted Lumley) There is a tendency to seek out like-minded others in order to confirm our own opinions - but in the sharing circle, respect for and celebration of difference emerge as a corollary of the way in which the discussion takes place - and even though the web forum appears linear, it is actually circular in the sense that every participant can be heard.

Thanks to the inclusionality email sharing circle for the ideas about inclusionality and hierarchy which prompted me to make the connection to web usability issues and write this article; and thanks to Simon Cox for introducing me to the world of usability. Thanks to Ted and Simon for feedback on this page.

Yvonne Aburrow, 17 April 2003