The story you are about to read is true. Only the URLs have been changed to protect the innocent.

A brief discussion of Wurman

Wurman claims that there are only five possible organizations of data: category, time, location, alphabet, and continuum. He sounded far too sure of himself, so we tried poking holes in his theory. As we tried to find other possible organizations though, we kept realizing that Wurman could look at our organization as either categorization or continuum. The trouble is, these two organizations are a sort of catch-all; anything that is discrete can be seen as categorization, and otherwise it's a continuum. In fact, alphabetical order, time, and location are all just extremely useful continuums. (Actually, Christian decided alphabetical order is not a continuum, but is in fact a dust with fractal dimension log(25)/log(26). After much debate we decided he was right, but we threw Oreos at him anyway.)

While it might be possible to force most organizational systems into one of Wurman's categories, this doesn't seem very useful. We came up with quite a list of ways to organize information on the web, each suited to certain types of data and uses of that data. These include orderings by:

Categorization

Continuum

However, we did come up with one more from of organization that did not fit into any of Wurman's categories: randomness. An example of this type of organization would be the French swimsuit-picture page "Femmes femmes femmes je vous aime!" which serves up a new random picture each time the page is accessed. While this system is certainly not useful if you're looking for something in particular, it keeps a high interest in the service over time due to curiosity of what picture will be chosen.

Jumping around the net

Categorization

The categorization of data by hierarchical subject area is useful for directed browsing: browsing when you aren't sure exactly what you're looking for (or what's out there) but have general areas of interest. By being led through hierarchical pages (each focusing on more specific subject areas) the user is led to progressively focus his search, such that by the time a leaf node is reached the user has decided on a specific item of interest. This method is not nearly as useful if the user knows what he or she wants exactly, because it often takes several steps to reach a leaf node. A good example of this type of subject categorization is The WWWVirtual Library subject catalogue. The ontology of hierarchical subjects does limit the kind of actions the user can perform and the kind of data we can classify. For example, a hierarchy implies that our information can be intuitively clumped into large groups, and that these groups in turn can be further classified. Also, the fact that our structure is hierarchical implies that each subject fits only into one superclass. For example, though it is certainly possible for a page on biochemistry experiements to be pointed to by both the chemistry and biology subjects, this breaks the model we started forming of a tree fanning out with information at the leaves.

Guided Tour

Organizing data in a guided tour model is useful when there is a strong linear thread running through the data (such as in a story, or when data relies on previous data to be understood). In these cases a guided tour can supply the user with that main thread, without restricting the user from occasionally going off on tangents that interest him. A guided tour fits nicely with (but doesn't require) a geographic conceptual model, since we tend to think of guided tours in cities or through exhibits of some sort. A good example of this merging is in the Soviet Archives Exhibit, which is based on a very complete "virtual museum" metaphor.

Location

While location has always been a useful organizational tool for atlases, guidebooks, and the like, it is an especially nice tool when mixed with the ability to click on a map or picture to gain more information. This sort of organization is really only useful when location is the most salient or the most important feature of the data. When location is the most salient feature in data (for example, in gaining information about states in the Union), it might actually be easier to search for a data point by location than by alphabetical listing. When the location is the most important feature, the organization of the data actually gives extra information that would be lost otherwise. For example, in deciding what college to go to, a student might be very concerned how far away from home each university is. By placing universities on a map, the user is not aided in finding a specific university so much as he is guided to specific universities based on that organization. A good example of a geographic location metaphor is the European Home Page.