Mapping virtual spaces

please click on titles or pictures to see more information

The Image of the City

In thinking about Lynch's introductory chapter, I began to think about what makes a city legible. Looking at my own experience, I realized that one of the most powerful visualizations of a city is the map of it's subway system. Living in Boston, this isn't readily apparent, since (at least until recently) the official T map included no information about the geography of the area that it serves. However, growing up in New York City, this relationship is very visible.

Below are scans of a handful of maps from the subways serving those cities. It is clear which provide a grounding in the geography of the area, although New York's subway map makes this link clearest. An interesting study was conducted (in a manner similar to that discussed in Milgram's piece) that asked Bostonians to map Boston. Most respondents could only place a couple of the main attractions of the city. Further, the Charles river and Massachusetts Avenue were very distorted.

Boston	New York	Montreal	Washington, D.C.	London

Were this same experiment performed on New York City residents, I suspect the result would be rather different, due in a large part to the situation of the MTA Subway lines display overlaying a map of the City's streets. NYC is arranged more "intelligently", with Streets and Avenues running perpendicular to each other and numbered orderly; This regularity, however, makes the layout of the City so generic that it is useful mostly to chart courses or to place addresses of unknown places, and doesn't take into account most of the landmarks present in the City. The streets may as well be a supplemental grid system which is used in the same way latitude and longitude are used to pinpoint spots on a map. It is the integration of the Subway lines and the geography of the City on the map of the MTA that allows New Yorkers to form coherent and accurate mental maps of their home.

Indeed, it is precisely this lack of coherency between the map of the mass transit lines in Boston and the geography of the area that prevents Bostonians from forming coherent and accurate mental maps. Just looking at the T map below, we can see how distorted is its geography.

Somewhere in between lies the map of the London Underground, which maintains some link with the geography of the London area, while making the map compact (compactness is presumably the reason for the form of T maps). Spending only three days in London with this Underground map, I was able to form some basic picture of the city itself, linking my interaction with the city (restaurants, theaters, bars, etc.) with the geography of the city.

Subway maps are a way to represent what Lynch calls the "apparent clarity" or "legibility" of a cityscape. When well executed (as New York City's is and Boston's is not) they provide the user with a substrate on which to build a mental representation or map of the city portrayed.

Lynch speaks also about getting lost. He interprets the state of being lost as more than just "simple geographical uncertainty". For him, it is disastrous. Digging a bit deeper into the meaning of being lost, we find that its emotional anxiety is heavily linked to a sense that we cannot find our way home. We don't feel terrific anxiety being lost in distant places, so long as we know a way to get back to our hotel rooms. We do feel this intense emotion even close to home when we cannot find our way, and have no visible recourse to finding our way there. We even feel this anxiety vicariously when watching actors in movies. Yet the emotion of being lost is completely absent in cyberspace, perhaps because we never really leave home, either physically or virtually (we are always browsing from our desktop).

Vivid physical settings also form the basis for "collective memories of group communication". The placement of a social situation in a physical setting within a larger area can be a powerful way to remember and think about that situation. This requires a mental model of the area and its surroundings, and if that area is mapped mentally within a much larger space, such as Times Square within New York City, then encompassing map can be effected by the emotions and events that unfold in specific areas of the larger map. Times Square evokes a certain feeling and contains memories that can be linked to other similar memories (temporally or spatially) that occurred elsewhere within the larger map of New York City. "As long as [you] can fit reality to the diagram, [you have] a clue to the relatedness of things." Paths can be traced from feeling to feeling within the larger map and future experiences can be predicted by their expected location within that map.

Providing such a well formed structure and map within the domain of the World Wide Web would allow people to navigate the space in the same way people decide where to go in the real world from maps. As each specific area is explored and experienced, familiarity with that area increases, and navigation would become intuitive.

Forming cyberspace into cities where information and online spaces are arranged like city street would allow people to use familiar navigation and mental model/map building techniques. This structure for our online world may not be attainable; we should, however, use our analysis of physical spaces to guide our initial forays into arranging and mapping our online worlds. Both are man made, and though they obey different physical models, because of their origins, they must be in some way modelable.

Notes on Mapping the Net

The notion of tribal cyberspace maps is a compelling one, at least in the short term. They don't provide for easy orientation, since their scales are often very different from reality, however, they provide an easy way to represent the relationships between places, and in this respect aid navigation. They are, unfortunately, almost impossible to generate algorithmically.

Moving from tribal maps to cartesian maps of cyberspace allows much easier orientation as well as navigation, however almost every heuristic for creating these maps binds too closely to the underlying form of the network. Ideally, since that which we call cyberspace exists entirely removed from the organization of the underlying network, our maps should also be derived from the content and the "sites", and not the computers and connections from and through which the information that makes up cyberspace flows.

Mapping the virtual geography of the World-Wide Web

Creating a map of the Web algorithmically is certainly a daunting task, and one that is not intuitive (not yet at least). Part of the problem is that there are so many documents that there are no clear metrics for ranking documents that are satisfactory for every document. Using the connections between documents as a base for sorting them in N dimensional space seems to produce a workable map of the space when collapsed into 2 or 2 1/2 dimensions, yet the result doesn't seem satisfying.

When looking at a space, and defining a map for it, we as humans allow two aspects of the space to compete against one another: regularity, which allows the map to be usable and the space navigable, and salience of important features, which allows us to find and use important landmarks to orient navigation. Girardin's self-organizing maps put too much emphasis on the former of these criteria, and (seemingly) no emphasis on the latter.

Since each element is equally weighted in terms of any metric of importance, there are no landmarks in Girardin's maps, and so they become useless for orientation. The map is so abstracted from the content of the web pages which they attempt to visualize that it is useless as a navigation tool. All that is available to us is some sense of the overall organization of the structure, which doesn't qualify as a map.

Mapping the Media Lab Website

There are a few key aspects of map making that allow maps to be useful for navigation and recognition of parts of a space:

• The map must sift out most of the unimportant aspects of the space, and extract only the useful data for the scale at which it is used. For example, a map of New York City may contain icons of important landmarks, but leaves out every other building in the city; including them brings no added benefit to the map of the City, and only adds confusion and clutter. However, a map of Times Square may show the buildings in outline form, so that orientation is facilitated.

• The map must provide some connection to the medium which it is depicting. This allows the reader to image the space, and form a mental representation that links the map to the space. Subway maps should show some of the geography of the area, so that the reader can connect the form of the city with the form of the subway lines. In a virtual space, this becomes more difficult, as there is no natural medium (linking the map to the physical layout of the network is useless since the physical network has little bearing on the form of the virtual network).



Each section of the city represents a research area within the Media Lab. Each building is a project. The green area in the center is a park, where all of the personal pages on the Media Lab website reside. Icons of pages are shown pointing to their respective "buildings", marking important landmarks.

The heavy colored lines represent "subway" lines, which are paths through the website following a particular idea or concept in research. By following a subway line, a user can visit all of the web pages that cover research on technology for children, for example. Each major project has a "stop", which is represented by the black dots along that subway line. White circles represent locations where multiple subway lines, and therefore research concepts, intersect. Sponsors are shown as a highway that encircles the city, representing their supportive nature as well as their fringe interaction with the Lab.

Note that the locations of the different areas of research were not randomly assigned, but rather placed in locations that would correspond to recognizable areas of New York City. For example, NIF was place around Madison or Park avenue, where many professionals work. Digital Life was placed on the Upper East Side, where many higher income persons and families live.

One side question: Why do maps of the Internet or of Websites always have so many arrows? I understand that they represent links between pages, but they just add clutter, especially when there are more than a few arrows on the map. Besides, maps aren't supposed to be just directions, but rather show the structure of an area, so that a person can pick out their own way through the space, not necessarily following a single line through the area.

page last updated 3/27/2001