Geography

It might surprise many designers to discover that geographers have an interest in virtual worlds. They do. They are interested in spaces, they regard cyberspace as a space, and virtual worlds are an exceptionally good example of the concept in use.

Note, there is a distinction ¹ between space and place. A space is an abstraction that groups objects of a particular type under a set of fixed rules; a place is a region (under adjacency rules) of some space. For example, matter operating under the laws of physics gives us the 3D space we call reality; Athens is a place in this space. Historical events operating under the laws of causality give us an “event space;” the French Revolution is a place in this space. Mathematicians are particularly fond of spaces, and will readily refer to name spaces, function spaces, and object spaces.

Spaces are distinguished from sets by the imposition of their governing axioms. Although it is true to say a place is a subset of its space, these axioms mean there is far more to the relationship than that: Things can be inferred about one place relative to another place in the same space. This is because places are connected to other places under adjacency rules, through which they can influence the character of these other places. Whether the spaces are real or imaginary, discrete or continuous, finite or infinite, it is with the study of the rules that underlie them that geographers are ultimately concerned.

In terms of virtual worlds, this means they are interested in:

• Virtual worlds as places (what they comprise).

• Virtual worlds as instantiated spaces (what comprises them).

• The (hopefully new) forms and relationships underlying virtual worlds.

As places, virtual worlds constitute one small part of the data space known as the Internet ²; other parts include the World Wide Web, email, instant messaging, bulletin boards, and Internet relay chat (IRC). Formally, “cyberspace” covers anything implemented using information and communication technologies, and therefore includes non-Internet forms of interactive digital data transfer, such as television and telephones. However, in practice it is with the conceptualization of these technologies known as the Internet that most of the research in this area concerns.

Geography begins with maps. Maps formalize the space to be mapped, and determine the nature of the adjacency relationships. Maps, like spaces, can be multidimensional; they are only, however, abstractions of the space ³, illustrating solely what is important from the point of view of the cartographer or the person using the map. By making a map of a new place, a geographer can begin to understand the nature of that place and of the space in which it is located.

How easy the Internet is to map depends, therefore, on the viewpoint of the cartographer. Most places on the Internet have a real-world geographical association, in that they are located on servers or operate through routers that have a physical presence in the real world. Thus, one obvious way to map the Internet is to overlay a map of the real world with the locations of major nodes, using data traffic flows to show their connections.

This wouldn’t be a map of the Internet, though; it would be a map of the real-world network that supports it. The Internet itself is a construct in the mind; to map it is to determine those constructs common to the individuals who experience it. Fortunately, most people have sufficient experience of real space to conceptualize cyberspace in similar ways, guided by the underlying structure of the data being presented. Thus, for example, they will regard a web site as a location; a hyperlink to another web site indicates adjacency (a browser’s Back button making it bidirectional, at least temporarily). Distance is identical between nodes, therefore the map will be topological rather than spatial.

On any map of the Internet, virtual worlds are only going to show up as different from other nodes if the cartographer captures and reflects details that are prominent for them—number of simultaneous connections, or period of continuous connection. Even then, they may not have a signature any different from an Internet radio station. A map of the Internet in general only makes statements about the relationships between nodes and about the variables that these relationships consider. It can only respect variables that are meaningful across all nodes (for example, bandwidth, uptime, router connections); the place of virtual worlds in cyberspace can therefore be located, but only in those terms. Unfortunately, interesting though this information may be to geographers and network managers, it’s not something virtual world designers are going to regret not knowing.

Homing in on virtual worlds, though, and looking at how they are constructed, is a different story.

In their two hugely impressive 2001 books ⁴, Martin Dodge and Rob Kitchin lay out the foundations of Cyberspace Geography as a discipline. They map everything: infrastructure, traffic, web sites, email, bulletin boards—even some individual Usenet threads. In so doing, they offer different ways of looking at cyberspace for different purposes. Because virtual worlds use richer, more intricate spatial metaphors than any other Internet social space (IRC, chat rooms, email, and so on), one of these purposes is virtual world design.

Beginning with MUD1 ⁵, Dodge and Kitchin examine a number of virtual worlds, both textual and graphical. They reproduce player-drawn maps ⁶ of varying quality and scale, commenting on the ability of each to convey in a clear manner the salient points that would be of interest to users ⁷. They conclude that there are four main categories of space in a virtual world:

• A core public space, accessible to all characters.

• Private spaces, accessible only to individual characters, but directly connected to the core space.

• Private spaces that are disconnected from the core space—that is, only accessible using teleportation.

• Multi-scale spaces that break the rules of realism—for example, a house inside a grandfather clock.

The prevalence of private spaces will, of course, be related to the number of people able to build them ⁸; there will be more in a MOO than in a DikuMUD, for example ⁹. In builder-heavy worlds, spaces more distant from the core in spatial terms will be more distant in thematic terms, too.

Dodge and Kitchin examine other ways to map virtual worlds, apart from this “physical” approach. They reproduce maps created from data gathered by a LambdaMOO bot ¹⁰ concerning the social interactions of virtual worlds; these enable the trained eye to pick out cliques and other social groupings among players.

An important consideration for any kind of space is navigation: How do you get from one location to another (whether this “location” is a point, a person, or a web page)? This is, after all, one of the primary reasons people use maps. The classic work on navigation by individuals is Kevin Lynch’s The Image of the City ¹¹, which describes how people build up a cognitive map of their environment and use it (among other things) to get from place to place. Lynch demonstrated that this “environmental image” is constructed in terms of five features:

• Nodes are distinct locations people can enter

• Paths are channels connecting nodes, along which people travel.

• Edges are borders that are difficult or impossible to cross.

• Districts are parts of the environment that share a common theme, look, or character. “The old town” or “the docks” would count as districts.

• Landmarks are strong reference points. The more important an object is, the more likely it is to be a point of reference. Being visible from far away is, in this context, considerably important.

Note these vary from person to person and from context to context. A path for a motorist may be an edge for a pedestrian; a junction may be a landmark for one, but not the other.

Although Lynch’s work concerned real-world cities, its principles can often be extended to environments in general. This was demonstrated categorically for textual virtual worlds in Andreas Dieberger’s 1994 Ph.D. thesis ¹², using results from two questionnaire surveys undertaken by Jolanda Tromp ¹³. Players of textual worlds have particular problems in that

• There are holes in the virtual world, where there “ought to be” rooms but there aren’t.

• Rooms are of different sizes, which players perceive only through their descriptions. A room that would take up several squares on a grid layout can still seem “small” to players.

• Exits are limited to eight compass points, plus: non-standard contextual exits such as “out;” hidden exits you have to work to discover; teleport exits that look like normal ones.

• There are unconnected rooms accessible only by teleport. Some exist for special purposes (for example, the virtual world’s mailing system), but this doesn’t have to be the case.

• Mazes that are fiendishly difficult to solve are not fiendishly difficult to construct.

• Z-axis movement (steps, ladders, cliffs, gradients, and so on) are hard to realize effectively.

• Distance is perceived in terms of time taken to traverse between rooms. Although this is related to the number of rooms passed through, inconveniences like doors and monsters can make a journey seem longer.

Nevertheless, even given these problems, Lynch’s model applies. Nodes are rooms. Major landmarks are rooms containing some special functionality or (very interestingly) ones where there tend to be other players. Minor landmarks are often where people find unidirectional movement ends—for example, the place where if you keep going east you suddenly find a wall that stops further travel in that direction. Paths are room exits or chains of rooms that get you from one landmark to another. Districts are used, but with theme being more important than look. Edges are only implied through room descriptions and the lack of exits at the “edge,” but rooms that have them will usually be on paths or act as landmarks.

Bearing this in mind, therefore, designers of virtual worlds should be able to construct their virtual worlds in such a way as to make them easier for players to navigate. Obvious nodes, paths, edges, districts, and (especially) landmarks should be provided to give players a sense of where they are and how to get to where they want to go; a sense of place, in other words.

Although this work was undertaken for textual worlds, Dieberger noted that graphical worlds (which have few of the “unrealistic” spatial problems that textual worlds do) were even more amenable to this kind of analysis. Designers should not simply assume people will find things to use as landmarks: They should follow the lead of urban planners and actually provide them. Furthermore, they should provide them in a structured manner, rather than as an accident of geography ¹⁴. In textual virtual worlds, they can go one step further and provide landmark-seeking commands ¹⁵.

Unrestricted teleportation is perhaps the greatest threat to the geographical integrity of a virtual world, as it breaks all the adjacency rules that make a space anything other than a set of points. Designers should therefore be careful to ensure it isn’t overdone. The other major problem (which affects textual worlds more than graphical ones) is the virtual world’s geography should be relatively stable. Experienced players don’t want to have to check all their maps every time there’s an update “just in case” the designers added a few new rooms somewhere off the beaten track, and inexperienced players will think you’re playing with their heads.

Geographers look at how space appears and how people (through their maps) visualize it; although they can offer up general principles relating to what makes a geography understandable and useful, they are not immediately concerned with designing spaces themselves. This may change if geographers seize on virtual worlds as test-beds for their theories of geography; geographers are already being called on to advise in the construction of geographically correct virtual worlds ¹⁶.

There is, however, a group of people for whom the design of space is precisely their main interest: architects.