MUDS: Cyberspace Communities

This chapter will describe the social structures of several text-based and graphic multi-user domains and the role of the citizen/builder in their communities. The chapter will conclude with a discussion on the representation of the computer and its users through avatars and agents.

The Abstraction of Games vs. MUDs

MUDs are mostly text-based virtual realities - a convoluted way of saying that users read them rather than view them graphically. Action is performed via keyboard while all scenes must be rendered mentally by the players from scraps of text offered in the course of play. Text is an efficient medium - a few words can evoke a rich response in the mind of the user.

Compared with graphic, sensational action games, text MUDs rely more on cognition than sensory perception. Spaces and avatars are not viewed on the screen but in the player's mind. The domains are symbolic extensions of our cognitive space. Action games, to the degree that they appeal to our senses, are more a part of our perceived space.

In the chapter "The scale of abstraction," I presented a scale for artifacts with physical objects at one end and concepts at the other. Neither video games nor text MUDs are physical although they are situated between the two extremes. Even if video games use imaginary settings and characters, they refer to physicality in their action and images. This puts them at a point in our scale near realistic, fictional images in paintings or film. In modeling our perceived space, they rely on our perception for their effect.

Conversely, text MUDs are abstract and cognitive since the characters and scenes are conveyed symbolically rather than sensorially. Yet even they use descriptions of physical objects and places to ground the action. And to that degree they still maintain a connection to the concrete world in the same way that memory palaces used perceived space model conceptual ideas. There is a major difference between MUDs and memory palaces, of course. While memory palaces were idiosyncratic and internal, MUDs are external, shared environments. Conceivably, they are sites of collective memory recalling the accretive memory of CAD models and network news threads. (fig. 87.)

MUDding: Representation and Movement

There is no dynamism involved in "walking" from room to room. Users instantly relocate episodically with a directional command. They can also "teleport" discretely to nearly any place in the MUD by using an @go command - although even the simplest movement in a text MUD involved categorical relocation.

The experience of MUDding is similar to reading a book. Characters converse in a described setting. If a user, Varmint, types "Hey, there!" the computer reconfigures this to "Varmint says, 'Hey, there!'" for other participants. Users and computer conspire to create a living literary fiction - the text equivalent of a masquerade.

The following is an example taken from a session on hte MUD HoloMUCK recorded by Michael Buldo, a former student at the School of Architecture at New Jersey Institute of Technology. Naima and Dex are avatar names of other MUD citizens. "You" is the computer's reference to Mike and is only used on his terminal. Other occupants of the MUD see the name of his avatar, Kilian. Spelling errors reflect speed of interaction since the dialog happens in real time. While all the information below appears on screen, I have differentiated the type to clarify who is saying what. Bold type indicates what the user enters. Italic shows messages sent by other users. Roman type is generated by the computer's MUD program.

01) Naima pages: "They arrived! well wonders never cease."
Time> Tue Oct 31 19:53:04 1995
page naima= we d got them on Monday!!, thanx alot
You page, "we got them on Monday!!, thanx alot" to Naima.
w
You head west...
Main Street (80OW)
This once-desolate section of Main Street is looking busier these days.
To the north, at 800 W. Main St., stands the Red Dragon Inn.
[Obvious exits: north, w, e]
page dex= are u free tommorow between 8Your pager vibrates slightly. and 10..am?? You page, "are u free tommorow between 8 and 10..am??" to Dex.
01) Dex pages: "no, not till tomorrow night"
Time> Tue Oct 31 19:54:33 1995
page dex=awww .. the people comming in to review the project will be on wanna let you meet them ... i might be round tommorw night..dunno
You page, "awww ... the people comming in to review the project will be on wanna let you meet them ... i might be round tommorw night..dunno" to Dex.
w
You head west... Main Street (900V~/ Ohio Avenue You are at the intersection of Main Street and Ohio Avenue, very near the western border of Tanstaafl County. This region is very hilly, and covered with a variety of trees -- oaks, maples, and especially pines. Reality still holds sway in this region, but just to the west, the darkened West Tunnel burrows through a hill and into the unknown regions beyond...

Despite the contrivance of text MUDS, they have an avid following. Novices quickly learn the ropes. MUD and MOO software is easy to learn and offers helpful prompts. Once the MUDding skills are acquired, the process of play becomes transparent.

In their minds users are "there" in the MUD, seeing the spaces and chatting with friends. Players invest themselves in the process, filling in details and suspending disbelief. Perhaps for this reason many are ambivalent - if not hostile - to the idea of visual MUDs. Some users insist that the introduction of graphics would diminish rather than enhance MUDding. They stress that MUDding is a social rather than sensory experience.

The MUD Environment

At times it's a claustrophobic world. For instance, users can't casually see distant views through windows or doors as they might in real life. This requires specific commands. However, unlike normal sight, long-distance viewing, once elected, is not limited by lines of sight and perspective. If I were to "look" at a remote object while MUDding, it would be visible as text regardless of intervening objects or rooms, like gazing into a crystal ball.

Both sight and movement in MUDs appeal to our spatial understanding. Yet their episodic, surreal nature belies the illusion of conventional space. This is a product of MUD technology and of computing in general. The categorical movement and "clairvoyant" vision in MUDs results from accessing stored descriptions. When prayers enter a new space the contents of that space's file is presented on the screen. The movement of the user - via her avatar - is entirely symbolic. Aside from the crafted illusion, it is no more spatial than accessing a file on a computer.

And yet the illusion of space is so compelling that it forms the basis of a MUD's continuity, its chief activities, its "placeness." Users gladly transcend the anomalies of text MUDding to benefit from this illusion and its resulting sense of community.

MUD Social Structures

The relative permanence of MUDs and user identity gives players a chance to familiarize themselves with their environment and each other. Beyond this, it also lets them define their own on-line identity. Domains require players to use avatars. Long-term use of an avatar allows character development and has an impact on the political structure of the MUD as well.

Competition and Cooperation in MUDs

Other players extend their social lives through the MUD environment. Here they make friends, gain an audience, play games and even engage in simulated sex. For many text MUD users socializing is more important than the need to experience the MUD visually. Some believe that visuals would detract from the immediacy of text - the sense that they are engaged in intimate conversation. Relationships blossom quickly in these environments owing to the isolation and unfamiliarity of the surroundings. It's as though participants were trapped together on a desert island.

Risk and Reward

But there is an element of risk in MUDding. The environment is strange and its occupants not always benign. The narrow bandwidth of communication paradoxically increases the risk through its very limitations. For example, in the real world we are used to communicating verbally and visually through gestures. In MUDs - even graphic ones - messages are stripped down to live text hammered out on a keyboard - spontaneous and misspelled, they convey immediacy. But, without the modulation of bodily expression, subtleties of meaning - irony or humor - are lost This all to the detriment of the communication. Risk of offending or being inadvertently insulted attends all transaction and users are forced to use smileys where neither words nor bodies can help.

Even 3D graphic MUDs with humanoid avatars require separate controls to operate limbs from those used in messaging. Avatars may move limbs but the vocabulary of body movement is often limited to walking and waving. Compared to our corporeal language of shrugs, raised eyebrows and wrinkled noses the avatar is effectively mute.

The main source of risk for the MUD player is betrayal - betrayal by others sharing the space or by the illusion of the domain. The former is often a product of the masking effect of the avatars - the temptation is just too great. The latter is more pervasive since the MUD illusion relies on the user's willingness to be taken in. The user buys into MUDding despite its obvious contrivance. All evidence points to its artifice: it is mediated by machine, experienced as text or graphics, vicariously explored by avatar. Disbelief is the price of admission and users leave their doubts at the door. Analogies drawn by Brenda Laurel between computers and theater are extremely valid here.

Skepticism is a defense mechanism, the nagging voice of reason. It's what helps us separate fact from fantasy. It's exactly what we don't want at the theater - or a MUD for that matter. But by abandoning doubt MUD players risk betrayal. This betrayal takes many forms in domains and in the chapter "Navigating cyberspace" we saw some of the ways the spatial illusion of MUDs breaks down. These lapses may be accidents or pranks of programmers. But the rewards of simulated society obviously outweigh the hazards for most players.

Investment in the Community

Although Erikson focused on the maturation process, we also benefit from the moratorium as adults. Play, hobbies, vacations are all part of our on-going, psycho-social regeneration. They refresh and re-animate us. The simulation of life in a multi-user domain is another venue for the experimentation described by Erikson. MUDs provide their users with a safe zone for exploring matters of identity, social behavior and even sexuality. For those readers interested in these issues I highly recommend Sherry Turkle's Life on the screen for a close examination of the psychological issues underlying these simulations.

Finally, unlike many other games, many MUDs are built by their citizenry. Although AberMUDs - a type of multi-user domain - limit the power of construction to the wizards, many MUDs invite their citizens to build their environment and its contents. This, of course, appeals to the participants already adept at programming, but in many cases code writing is quickly learned by aspiring citizens.

But this alone doesn't explain the tenacious hold domains have on the imaginations of their users. MUDding requires an investment by the user in exchange for its benefits. In the chapter "Body Extensions in Space" I described a party in terms of the risk and investment we make by being physically present. Our payoff is in the reinforced friendships and engagement we have at the event. Any involvement, mediated or physical, requires investment of our attention, time and physical activity. This applies to MUD activity as well. Time spent MUDding is time not spent on somethin else. Presence in a MUD is an investment in the simulation.

Construction of (and by) the Community

The investment of experienced MUD players is in their service to the MUD, their construction of a surrogate self - the avatar and the enhancement of the domain. In exchange many players gain achievement and a sense of belonging. Although their work takes place in cyberspace, the feeling is real enough.

For in building spaces and animating them with avatars the citizens create what Ray Oldenberg calls a "third place."². Our daily "real" lives are often polarized between our domestic and work roles. And polarized physically between the places we play them. A third, intermediate space is where we are free to relax, socialize and be ourselves. Multi-user domains are the electronic equivalents of cafés, bars and street corners. These aren't places of solitude but places of engagement where we find friends and talk out ideas.

It's a neutral turf where, ideally, no one dominates. Perhaps for this reason as well as the anonymity of avatars, MUDs are well suited as sites for self-discovery and exploration - the moratorium spaces of the Internet. Third places bring people together, effectively leveling their social status. Since selection of avatars is arbitrary there are few clues indicating status within a MUD. The giant blue oyster you're chatting with could as easily be a fifth-grader as a computer scientist.

Many electronic media require the creation of avatars to represent the participants. These are either analog descriptions of their users - a telephone voice or video image - or, in digital media, they may only have a tenuous connection to their user. In the role-playing environments of MUDs players take on either assigned characters or - in many cases - assume self-created identities. The critical difference between this and a television show is that a MUD avatar is part of an on-going, unscripted performance. The spontaneous interaction with the space and fellow citizens of the domain gives MUDding its life-like immediacy. Envisioning Cyberspace

Elusive Entities: Avatars and Agents

Most video game players have to accept the avatar they are given. Conversely, a MUD player creates her own avatar which remains part of the MUD once she is accepted into the community. A MUD player may log in as often as she likes, using her avatar to participate. In constructing this identity the player invests herself in the MUD, returning periodically to reinforce her presence in the domain. As we would expect from an social setting, the consistency of characters over time is important to their identity in the community. Repeated exposure to someone else in the same environment dimensionalizes them as we learn more about them.

But matters of identity are never simple in MUDS. The avatar is only a mask for the user. It may not resemble her physically at all - not even in the text description (figs. 88, 89, 90). An avatar may take on traits of animals, robots, historic figures, even furniture. A character's ability to teleport - a given in MUDding - is miraculous by everyday standards.

This gap between the user and avatar mask gives rise to many complexities of on-line identity. Nothing prevents a computer user from belonging to several MUDs - effectively distributing herself among several communities and identities. Arguably, this resembles our role-playing in conventional society. I, for instance, am a husband, father, teacher, architect and party-goer - each requiring distinct modes of behavior. But in everyday society I manifest all these roles in one body. On-line, these manifestations - or avatars - may be radically different.³ In most cases, though, it's not a problem since an avatar is usually unique to its MUD - most players limiting themselves to one avatar per domain.

Theoretically one user could have as many avatars as she can manage. Some MUDs encourage this since often MUDs appear underpopulated. A user might, for instance, operate a group of avatars at a time. Simon Birrell, a designer of graphic MUDs, and his colleague Christopher Stangroom used this strategy in creating their MUD, Jekyll & Hyde. A MUD player, rather than taking on a single character, takes on a character group - say a gang of thieves or a team of policemen (fig. 91). The action is accomplished via the screen interface but the major feature is the multiplication of the user through her avatars. The person behind one avatar may be the same as behind another.

Just as one person may wear a number of masks, several people may share the same mask. Logging onto a computer with a borrowed identity is frowned upon but commonplace. The same holds true for a multi-user domain. In fact, given the parallel with video games it comes as no surprise. Since pre-set video games offer only a set staff of avatars, players commonly have to share one another's avatars from game to game. That's not hard given the brief and mortal interaction between participants.

It's another matter on a MUD however. As we saw earlier, newcomers are commonly assigned re-usable avatars for the duration of their visit. These avatars have default names like Guest 1 or Blue Guest and their user/avatar relationship has no longevity in the MUD. Green Guest, for instance, may represent a host of users in the course of a week. This is not disruptive to the MUD since its citizens recognize the guests' transience. The guest's avatar is like a rented car.

It is more disconcerting if the person behind an established avatar is not consistent. It shakes the illusion of the MUD by calling attention to the gap between the user and her avatar. Consistency of character is important to the illusion of MUD society. Inconsistency, on the other hand, connotes duplicity even schizophrenia - neither being desirable in any social setting.

Unlike its analog counterparts, the digital avatar is an independent artifact within the MUD. It's not just a projection of an individual as is - say - our television anchorman on the nightly news. Instead, an avatar is like a mask through which a user is present in the domain. Sherry Turkle refers to avatars as "personae," the Greek term for mask. Translated it means "through sound" indicating that the masks used in theater through which the voice of the actor was projected. Also implicit is the interdependence between the actor/person and his mask/personae. Every day we take on our roles according to our situation. We express these roles - and their masks - though demeanor, voice, dress and ornament. Arguably these roles are the various threads we weave into the braid of our identity. ⁴

What distinguishes the digital avatar from these more integral roles is its psychic distance from the user. Given the gap between the player and the avatar, users may play roles that bear no resemblance to themselves. By separating from their masks, players bear less responsibility for their actions since consequences are directed at the avatar rather than the user. Although users treat their avatars as personal extensions, avatars are also a kind of shield. This allows the freedom of personal exploration encouraged by MUDding - its moratorium. Just as a MUD space may not be what it seems, neither may its occupants.

One of my former students, George Wharton III, took this ambiguity of identity as the theme of his design for a cyberspace bar. When a player arrives at the bar, she is given a mask through which she can see the other participants (fig. 92). All other avatars are represented as masks floating in the dimly lit space. At the side of the screen a menu offers several other optional masks. This menu is a listing of the other avatars in the space. By selecting another mask the user can see and hear what the current user of that mask is experiencing. The bar scene is populated by masks behind which there may be any number of users. Or none at all.

The idea that there may be no one behind the mask is hard to imagine in the physical world, but it's a fact of life in MUDs. Momentarily abandoned avatars haunt the MUD like zombies, to be awakened when their users return. Others are ownerless and autonomous. Much as an avatar represents a player, a bot - short for robot - may be simply a mask for a computer agent (fig. 93). Unlike many of their video game counterparts, bots may be supported by elaborate programs for interaction. They may range from simple behaviors, like a robo-dog that follows players from room to room, to sophisticated routines that fool players into believing they represent living players. In fact, two famous on-line personalities were actually programs created to simulate human interaction in networked environments.

Michael Mauldin, a researcher at Camegie-Mellon University, programmed Julia, a bot that responds to players with words and gestures. Since she operates in a text environment, she has as much presence on-line as any other avatar. And this is crucial to the illusion. Mauldin has programmed Julia to seek out key words and phrases from other users' text and respond to them from a list of options that apply. Her spontaneous wit is crucial to the illusion of her intelligence. She can dish it out, too:

Ace says, "So why not kiss me, Julia?"
Julia says, "Sorry, I'm not into geeks, Ace."
Ace says, "Julia, I'm not a geek."
Julia says, "Then lose the pocket protector, Ace."⁵

ELIZA, a program originally developed by Joseph Weizenbaum in 1966, was originally created to test a computer's ability to mimic human conversation. The program mirrors the comments of another participant in a conversation. For instance, someone might tell ELIZA, "I am feeling sad." ELIZA's program separates out the "I am," substitutes it with "You are" and precedes the answer with "Why do you tell me that?" The response reads, "Why do you tell me that you are feeling sad?" Despite its clear artifice, ELIZA became a preferred conversation partner. Also, to Weizenbaum's chagrin, she became a hot topic in the Psychiatric community as a possible tool for therapy, an idea that her inventor felt trivialized human interaction and worth. ⁶

Bandwidth as Mask

The Trouble with Human Metaphors

For this reason designers of agent interfaces try to manage users' expectations. Kris Thorisson, a former researcher at MIT's MediaLab and now at Lego Corporation, has designed cartoon-like interfaces for agents. One interface, Gandalf, is a caricature of a Viking - helmet and all (figs. 95, 96, 97). He is disarming, a slightly goofy smile on his face. The user is aware of his attentiveness yet knows not to expect the intelligence of a human. Although Gandalf is quite sophisticated and can converse on specific topics, the user risks no disappointment owing to his representation. Gandalf presents a well-chosen metaphor for interaction.

Another problem with the human metaphor is that it sets up a dialogic relationship with the computer. This is the point brought up by John Walker in his "Through the looking glass" memorandum. ⁷ Agents acting as humans - even cartoons imply that interaction is by conversation. This may not be the best model for getting needed information - if we don't ask the right question, we won't get the right answer. This concern underlies debates between those promoting artificial intelligence (AI) and those in favor of virtual reality. Others besides Walker, notably Jaron Lanier, feel that the computing environment should be a space for discovery rather than a limiting channel of dialog. In this view AI is seen to be reductive, deterministic and a poor model for accessing information.

The Legacy of Artificial Intelligence

Over the past twenty years the model of human cognition changed from being monolithic to being comprised of many parts that together - it was believed - formed identity and produced thought. In this view, our behavior is the complex product of many relatively simple, iterative processes. While this model continues to be controversial among psychologists, the computing community has used it effectively to advance artificial intelligence research. This model vies against the top-down model of cognition that argued for one complex algorithm defining all behavior. Instead the bottom-up model proposes that complex behavior is the interaction of many concurrent simple processes. The impression of behavior and thought is the emergent property of these many subsystems in action.

Rodney Brooks, a researcher at the Massachusetts Institute of Technology, is well known for implementing this principle in several small robots - some smaller than a breadbox. Each robot is made up of relatively simple software mechanisms that, in concert, create complex behavior. Patti Maes, his former colleague, has applied these ideas to producing software agents that perform tasks ranging from scheduling to taking polls on the lntemet

Agents are subject to similar criticism as against AI. One of the more intriguing twists on the older arguments addresses the morality of the human metaphor. While the original model of AI inspired fear of being dominated by expert systems, the agency model does the opposite. Should computing be based on having others do our work for us? Are the simple agents that sort our mail and search the lnternet no more than electronic slaves? Has our cultural and technological evolution only led us to repeat the very worst of our mistakes?

The problem, of course, lies in the metaphor. No one denies that computers - and their software - are created to serve our needs. It's when we put a human face on computer agents that the problems begin. It doesn't matter to the computer what representation data takes, whether ambient or anthropomorphic.

Software can take many forms, each telling us something about its designers. Presenting the user with agents to do his bidding is different from immersing him in an information environment. Each scenario empowers the user in different ways. In the former the user delegates his work, in the latter he navigates a space created for his use, letting him discover and learn.

In the end, however, the space and its agent occupants are masks for the machine. Whether the machine presents us with characters or ambience should ideally make no difference so long as our needs are met. Just as there are times when text is more appropriate than graphics, so too are there times when an anthropomorphic interface is preferred to a spatial one. The selection of one over another is often a pragmatic decision based on resources and desired information. ⁸

The debates surrounding the use of agency have less to do with its function than the entailments of its metaphor. Whether or not they were manifested as icons or text, agents were understood to be software entities with purpose and, at times, personality. In their software simplicity they were not too far removed from the automated creatures that roamed the IRCs and multi-user domains. Their anthropomorphic association stuck, resulting in some heated debates that only recently have begun to cool.

What the controversy about agents reveals is how cultural and political values enter what originally were discussions of technology. The stakes have been raised. As computers have become more sophisticated and accessible, they have become a cultural as well as technological issue.

Conclusion

1. I will focus on text MUDs in this discussion because they have a longer history of use than graphic MUDs and because they share many qualities of space and categorical movement associated with the Internet and World Wide Web.

2. Oldenberg, Ray. 1989. The great good place: Cafes, coffeeshops, community centers, beauty parlors, general stores, bars, hangouts, and how they get you through the day. New York: Paragon House.

3. Turkle, Sherry. 1984. The second self: Computers and the human spirit. New York: Simon and Shuster. p.294.

4. Other metaphors for identity could include houses with many rooms. LambdaMOO has an avatar called "Tree" who is actually a MUD in itself. Visitors can enter the avatar and explore it like a domain.

5. Turkle, 1995, op. cit., p.89.

6. Turkle, 1995, op. cit., pp. 105-106.

7. Walker, John. 1990. "Through the looking glass," in The art of human-computer interface design. ed. B. Laurel. Reading, Mass.: Addison-Wesley. p.443.

8. Laurel, Brenda. 1990. "Interface agents: Metaphors with character," in The art of human-computer interface design. ed. B. Laurel. Reading, Mass.: Addison-Wesley. p.362.

MUD Spatial Structure

This chapter presents the implicit spatial structure of MUDs. It introduces logical adjacency models (LAMS) as tools for understanding the cognitive structure of on-line social space. We will also see how spatial anomalies in MUDs reveal inconsistencies in the spatial reference of domains. And finally, we will conclude with a discussion of how motion in MUDs determines their form in the way that circulation forms our physical communities.

The organization of each MUD varies according to its themes and history. Its spaces can be modeled by simply noting directions used to get from point to point. For instance, if a room is north of another, it maps as a cube above another cube. A room to the west is mapped to the left and so on. Sometimes rooms are not accessed directionally, but by invoking their name or number. These spaces are indicated with spheres located arbitrarily so long as the connections are maintained in the diagram (fig. 98). In studies I have conducted with students at the New Jersey Institute of Technology and the University of Michigan we have mapped several text and graphic MUDs using this method to uncover their logical structure.

The product of these efforts, called logical adjacency models (LAMs), were surprisingly complex. Because of their node/connector construction they resemble large, molecular models. But LAMs are necessarily incomplete because of the size and ephemeral nature of the MUDS. Larger MUDS, like LambdaMOO, can contain thousands of rooms. Mapping such a MUD in its entirety is almost impossible since the number of rooms change by the minute. Entire sections of a MUD can disappear or transform overnight. While Rome wasn't built in a day, Columbia University's graphic MUD model of Rome effectively was. After months of preparation, the MUD came into being immediately after its server was connected to the network.

Adding to the difficulty of mapping, some rooms can't be mapped logically because a MUD may contradict itself. For example, in real-life we expect reversibility. If I leave a space by going west, I should be able to return by going east. As a rule this is true of MUDs, however in DreamMOO such an attempt at re-entry may unexpectedly lead to a third room.

The ball-and-stick modeling of MUDs works well so long as the MUD's implicit structure complies with the model's assumptions. But that's not always the case. Nested spaces like those found in MUDs like The Chatting Room or Diversity University are linked by including one another rather than bridging between. The forced rigidity of movement in HoloMUCK requires the construction of intermediate stepping-stone "rooms." These have no other purpose than to relate changing perspectives as a player moves down a street.

The logical adjacency structure of each MUD has a distinct form, like a fingerprint. Often MUDs begin as a verbal diagram of a neighborhood (Jay's House), an existing town (The Chatting Zone), or even the Earth (MeridianMOO). A MUD administrator will determine themes, rules of play and often designs the preliminary spaces. Once in place, citizens of the MUD are often invited to build their own rooms, objects and buildings. Over time the configuration of the domain evolves to where not even its operators - called wizards - know its current shape.

This is a participatory construction, a kind of architecture without architects. There are constraints, however. The degree of freedom enjoyed by citizens is determined by their stature in the community and the MUD's administrators. Some MUDs like HoloMUCK or Jay's House have stringent codes enforcing the realism of proposed additions. In Jay's House, for instance, the builder of some cliffs outside the neighborhood had to change the description of their mineral composition to that of the actual cliffs of the referent community. The irony is that the only material builders use is memory - the allocated computer memory of the MUD.

MUD Citizen as Builder

Getting memory and status in a MUD is a symbolic and social matter - often a result of who you know. This has obvious parallels in conventional society. The more time a player invests in a particular MUD, the more exposure she has to others and the MUD administration. This vicarious investment pays off in memory and in special powers granted by the wizards.

Citizen-builders rarely write code from scratch. In object-oriented MUDs - called MOOs - each object is usually a modification of a parent object. Every object shares this "genetic" characteristic whether it is an avatar, room, or robot. It is inherent in object-oriented programming, a code-writing method that lets programmers assemble software from code modules called objects. There is an implicit evolution of objects as one generation succeeds the next.

Modifications are accomplished by editing the code underlying a parent object. Say I wanted to create a bartender robot. I could take the code for a waiter robot as a template and change some characteristics. Whereas the waiter would ask, "May I take your order?," the bartender might ask, "What'll ya have?." Editing attributes is easier than creating new behaviors, and many objects are simply redefined versions of earlier models.

MUD Spaces and Societal Roles

The consistent backdrop of the MUD space - however abstract - acts as a stage set for dramas played out by its members. It influences and sustains thematic interaction simply by being there. There are clear parallels in how our physical environment affects our behavior. As we may each play separate roles in our society, society provides us with places to act them out. Barroom behavior is not appropriate in a church, nor is preaching at a party. We go to different places to play our roles. MUDs emulate these spaces, by hosting different behaviors.

There are many MUDs without fictional themes. These may be based on physical models, perhaps the hometown of the founding wizard or the physical site of the MUD's server. For instance, The Chatting Zone maps Ipswich, Ireland, hometown of its founder. While MeridianMOO represents the entire planet, users enter it in "Norway," the native land of its wizard. Oddly, the MUD's server is located in Morristown, New Jersey.

Basing MUDs on actual physical models is an expeditious first step in starting the domain. It saves the wizard the effort of creating spaces from scratch and lets her make a "home" of the domain. JupiterMOO, developed by Pavel Curtis, is based on the layout of Xerox PARC in Palo Alto, California. Its mapping is so accurate that citizens of the MUD who visit the physical facility have no trouble finding their way around the campus. LambdaMOO, still one of the largest operating MUDs, was originally based on Curtis' personal apartment. His bedroom closet is still the point of entry. In another case, MediaMOO incorporated the architecture of MIT's MediaLab.

Using a physical prototype is also a basis for reference. A conventional urban space provides a familiar spatial framework fornavigation. Streets, buildings and places keep directions simple and memorable since they are experiences common to all users.

These elements were also common to earlier memory palaces. They too relied on a familiar - architectural - organization. However, unlike those purely cognitive and idiosyncratic spaces, MUDs are shared through technology. In fact, the primary purpose of MUDs is to support social interaction. The wizards' selection of realistic physical prototypes in starting MUDs provides the MUD with a palette of ready-made social spaces, sets for various activities: classrooms, houses, bars, museums and the like. While a thematic MUD might determine the behavior of its characters, MUDs based on existing communities let users "be themselves" and select settings based on momentary need - the way a city provides a variety of optional backdrops (figs. 99, 100, 101, 102).

Linked and Unlinked Spaces

This produces spaces with non-directional links to the main structure. Most private spaces - some quite elaborate - can only be entered by teleportation. And then only at the host's invitation. Teleportation requires the address of a destination. If a casual user doesn't know of a "hovering" private space, it's unlikely that he will find it simply by navigating the MUD. He would have to make a lucky guess using the @go command or get assistance from a colleague. Due to this opacity, accurately mapping a MUD structure can be a frustrating experience.

Politics and the Spatial Structure of MUDs

The LAMs of more tightly controlled MUDs are rigorously geometrical when compared with more democratic domains. In contrast, the looser structure of democratic communities makes them initially harder to navigate since there is no overarching order. In MediaMOO, for instance, organizing elements like Curtis Commons were added later to provide orientation for users. This is the electronic equivalent of urban projects in Baroque Rome or Napoleon III's Paris. Larger public spaces and boulevards were carved out of the medieval city to provide spatial structure for larger-scale organization.

As in cities, MUDs need orientational spaces once they become large, particularly if they lack any other geometrical structure. In democratically-run MUDs like MediaMOO, these spaces are created spontaneously by community members. Centrally controlled MUDs must rely on their administrators' experience for their organization.

The McGill University MUD, HoloMUCK, illustrates the extremes of administrative control. Originally its wizards had developed Flux, a MUD that placed minimal restrictions on new construction. As it evolved, its configuration became increasingly complex. Navigation in the MUD depended more and more on teleportation as the order of a larger MUD structure dissipated. Eventually the administrators felt the illogical nature of the spaces confused the players, rendering the MUD unusable.

Flux was reincarnated as HoloMUCK (fig. 103) using geometry clearly derived from a generic Canadian small town. Two main roads intersect to provide the center of a gridded community, a river bisects the town laterally. The wizards placed strict and strongly enforced rules on construction within the main structure of the community, TANSTAAFL. ¹ As in Jay's Place, HoloMUCK's planning stresses the realism of the domain setting. If a closet were revealed to house aircraft, the wizards would not allow its construction within the main structure.

If the failure of its predecessor was due to spontaneity, HoloMUCK now suffers from its stifling restrictions. Even movement is affected by the demand for realism. Players have to move through stepping-stone spaces that describe changing perspectives as they navigate the MUD. This exacerbates the already anomalous, categoric movement in a text MUD.

The peculiar interior quality of MUD spaces presents them as a set of connected rooms. However, HoloMUCK's spaces are described as both external and internal. And this poses a problem for those attempting realism in text. If, for example, we walk down an actual street the perspective of our view changes with each step. The church on the far corner soon becomes the church to our immediate right. How does the designer convey this dynamism using categoric motion?

In many text MUDs a public space like a street is simply another room with suggested exits. In HoloMUCK a street is a series of separate spaces disposed linearly with exits to the right and left. If players were given instructions to go to the third building on the right, they would have to go through three of the street's sub-spaces, then exit to the right. This methodical plodding, of course, resembles the process of moving down the street. But the deliberate incremental movement between spaces calls attention to the artifice, undermining the credibility of the MUD environment

HoloMUCK's wizards have tried to alleviate the over-regimentation of the domain by allowing builders free reign outside the city limits. Lying outside the main structure is a freezone where spaces may follow any or no logic at all. As a result most new construction lies outside the rigorous and isolated core of the community.

Spatial Anomalies

For example, the popular notion of "surfing the Web" alludes to dynamic motion despite its categoric, discrete nature. It is simply the illusion of Web pages' adjacencies regardless of server locations. I can "go to" a Web site in Tokyo or Sydney no matter where I am physically. The sense of being somewhere else while logged onto a network is the illusion that underlies Gibson's cyberspace. These illusions of motion and engagement are not limited to computing networks. As we have seen, film, television, art and literature have the power to transport us as well.

Text MUDs have a range of anomalies involving space, identity and movement. As mentioned, a space described as "outdoors" is simply another room in the MUD's diagram. A building in a MUD would have to have a link to this space in order to have an exterior - unlike buildings we experience physically. Things are not what they appear to be. A room's description may have nothing to do with its use. The bedroom closet of LambdaMOO can house dozens of users at a time. The sewer system of The Chatting Zone turns out to provide shortcuts below the rigorous mapping of Ipswich, Ireland. Nor is identity safe from ambiguity - social identity in MUDs is itself part of an on-going masquerade.

Many MUD anomalies are products of movement. HoloMUCK is only one example among many. Motion in a domain - using cardinal directions - is the result of text commands, not physical relocation (fig. 104). Users are subject to the logic of the MUD and its designers - neither of which may be trusted. "Black holes" occur when users can't leave a space unless they teleport to a destination or log off the system. It's hard to imagine a physical counterpart to these anomalies since they are a product of categoric motion.

Some anomalies are unique to specific MUDs. These include variations on the black-hole anomaly. In BayMOO, for instance, the prison island of Alcatraz is represented by a loop of connected spaces (fig. 105). One enters the island from "San Francisco" and proceeds through the spaces one at a time. Appropriately, the player can't get out. Each room only leads to the next and none return to the mainland. Another example is in Alpha World, a graphic MUD, where a group of five spaces interconnect. Each room has a teleportation terminal that looks like an elevator. Each terminal has four destinations, but the user discovers that they are only the other four rooms. The five rooms are part of a cluster forming a pyramid. There is no way out by using the teleportation devices.

While Alpha World's anomaly was clearly designed as a trap, many anomalies may not be deliberate. Programming - or digging - a new MUD space in a MOO requires simple but specific programming. Entering the MUD command @dig "Living Room" creates a space called "Living Room." This space is unconnected to any other, effectively hovering in Limbo. The builder can only access it categorically by teleportation.

If a designer in the "Kitchen" wants to create a room with links to the Kitchen, the command would read @dig w/e to "Living Room." This creates a Living Room that is entered by going west from the Kitchen. It is exited by going east from the Living Room, returning the user to the Kitchen.

A designer can accidentally create a black hole by using the command @dig w to "Living Room." This leaves out the return leg of the previous command, stranding the player in the Living Room. There is no way out except by teleporting to a destination or logging out of the MUD.

In MOOs the tools for creating a spatial environment include @dig, @add exit, @add entrance, @exits, @entrances among others. Each has its own syntax and entailments and - predictably - coding errors that produce accidental oddities in the MUD. Regardless of their origin these paradoxical spaces have few precedents in our perceived world. If anything, they are more allied with the fantasy world we inherited from childhood. Similarly, we suspend our belief when we enjoy a movie or a novel. Disembodied relocation and point of view are easily accepted in these media without risk of losing the audience.

Since anomalies are an accepted part of text MUD culture, designers often employ them to direct traffic within the environment as well as to create elaborate traps. A room with appropriate exits can be used to bypass a tangle of connected spaces, for instance. Other rooms are actually vehicles, examples include buses, trains, airplanes and MeridianMOO's QE2 - a fully outfitted cruise ship. A room hanging disconnected in a MUD is effectively private since no casual user is even aware of its existence.

Problems of Design

Unsuccessful rooms are like unsuccessful Web sites. Once built, they are rarely modified. Visitors may "hit" on a space once or twice, but without novelty or companionship to engage them they rarely return. Wizards encourage builders to innovate, making their objects and spaces entertaining, but the talent and ambition of users varies widely and unless construction is checked by an administrator, the MUD risks wasting memory on superfluous construction. Even in successful MUDs many citizens are only familiar with a handful of spaces. Many have not explored the main structure since their first few visits.

MUD social activity often centers on the entry, where users begin their sessions. It often appears as a lobby, town square or visitor center. The area immediately around the entry is also populated but occupancy drops off sharply thereafter. Experienced MUD users often prefer teleporting to their destinations rather than sequentially moving through the labyrinth of rooms. Once they arrive at their destinations they often move spatially, browsing the adjoining rooms (fig. 106). This movement resembles our earlier recollection our childhood bedroom. We moved categorically in recalling the space, then we moved dynamically within it.

The apparent emptiness of MUDs is exacerbated by privatization. As mentioned, most private spaces often float free, unlinked to the main structure. The Chatting Zone and the University of MOO, among others, have a predominance of rooms where private socializing occurs. Many citizens are in the public space only momentarily before they teleport to their rooms. From there they secretly monitor MUD activity or work on their spaces.

This depletes the public spaces of a population and turns them into virtual fish bowls. The MUD space appears empty since there often aren't enough users logged on to support this stratification. Only by using a categorical command like @who can the average user find out who is currently logged on.

Solving Problems by Spatialization

The graphic representation of a domain might offer solutions to these problems. If a visitor can "see" the extent of a MUD she may be more inclined to explore it. The text blinds users to distant spaces, blinkering their experience. This limits users to sequentially plodding from room to room.

Teleportation is preferred to spatial movement once the terrain is familiar and destinations are known. But socializing seems to relate inversely to teleportation - it's hard to meet people on the street if everyone is teleporting.

Another resolution to the social stratification may incorporate all private spaces into the main MUD structure. Also limiting access to these spaces to spatial movement may improve social interaction. Finally, increasing the number of entries to the domain would shorten the distance to subsequent destinations. If more than one entry is used, each will serve as a node of activity, creating the equivalents of neighborhood pubs and hangouts.

Random access at these points could stimulate exploration and interaction. Once the main entry has a critical mass of occupants, additional visitors could be let in elsewhere to spread activity to lesser frequented areas. This could revitalize the community.

Visual MUDs

Most WOOs refer to our perceived world by incorporating ground planes, sky, horizon, light sources and 3D objects. Significantly, while these items are tacit in a text MUD - our minds fill them in - in graphic MUDs these form the basis of the environment, the game board. This replication of our earthbound, physical world implies that the player is present in "human"form. Of course, the user's avatar may take any shape but many principles of human perception still apply. For instance the default orientation of the sky is above the ground plane and the limited range of view implies frontality and vertical posture. Standard viewing heights of a few feet imply a scale similar to that of an average person.

Predictably these environments enc ourage dynamic, spatial movement. Displays of these spaces - such as those of Van Gogh TVs Worlds Within (figs. 107, 108, 109) include navigational controls.² Others offer virtual joysticks, standard in the VRML browser, as though the avatar were some form of vehicle. Movement appeals to the user's perception - perspectives change as the player navigates the space. In many ways the action resembles that of a conventional video game, although at a slower rate.

Action games like Ouake have elements of MUDs built into them but, to enhance their speed, the images are low resolution. In large graphic MUDs the resolution is also limited but the complexity of the environment slows the interaction down. To convey dynamic motion, new perspectives have to be quickly regenerated to portray the new viewing angles. The more complex the environment - the higher the polygon count - the longer a view will take to render. If the user moves too quickly he may notice a lag in the image. In an immersive environment this could, over time, lead to disorientation and headaches.

Programmers of 3D MUDs keep the polygon count down by simplifying the construction of objects and using bit-mapped surfaces which show details as two-dimensional graphics. Consequently the worlds created have a schematic, cartoonish quality that belies their realism.

Another technique surrounds the user with a zone of resolution, called a horizon, comparable to a bodily zone of privacy. In Alpha World, for instance, only the buildings in a radius of roughly one hundred feet are rendered. However, beyond them, a player can see remote sky and mountains many miles away. As the user moves in the space, the zone of resolution keeps pace, generating objects and buildings along the way. To the player it appears as though the objects magically pop into view. This is a parallel to the episodic movement in text MUDs.

This product of computational necessity recalls the somewhat self-centered world of childhood. The world of WOOs responds to the player, it's. not objective or concrete. Just as a child might imagine a world created just for her, WOO players can see one under construction responsive to their movement and point of view. The moon in the sky does follow you.

Despite WOO's at dynamic movement, users still use categoric motion. Logging in and out of a domain still involves the categorical movement of the lnternet. Given the science fictional nature of many worlds, teleportation is popular as well. As in text MUDs, WOO players can forego plodding from space to space and instantly arrive at their destination.

Dimensional Space in Graphic MUDs

Other MUDs, like World Chat and Alpha World are more committed to rendering a three-dimensional environment. Unlike text MUDs, space in these worlds is transparent. Despite the zones of resolution, it's possible to see many objects and spaces at once, whereas text MUDs limit users to one space at a time.

While "outdoors" is simply another room in a MUD, in a WOO it is an all-embracing volume - more like our perceived space. In a text MUD a player may be "outdoors" but still be only offered a finite number of directional cues. A player standing "outdoors" in the open space of a WOO is implicitly connected to all buildings and rooms in the domain as well as all points within the outdoor space itself, theoretically infinite in number.

For this reason, a WOO maps better as plans, like those used by architects and urban planners, rather than the flowchart method used in LAMs. Only when the world space or room is connected to another through categorical, episodic motion can it be mapped easily using the LAM method.

Movement in 3D domains is not limited to teleportation and the ground plane. Flying is another popular way to get around. It's also an easy way to spot the artifice and inconsistency of the simulation. A common way to create a horizon in a 3D MUD is to surround the ground plane with a cylindrical scrim that displays horizon and sky. Seen from the ground plane this is fairly convincing since the user doesn't see the top of the cylinder. Flying, however, changes the perspective, revealing the curvature of the cylinder (fig. 110).

The effects of the cylinder can be seen from the ground as well. Some time back some students and I went to the suburbs of Alpha World. Alpha World is actually made up of several thematic worlds but the main space resembles a city placed on a flat, green plane. Over the past few years the citizens have built hundreds of buildings, roads, and houses from an available kit of parts or by inserting models of their own. However, despite its size, the city of AlphaWorld is finite. Beyond the last buildings the gridded ground plane pushes out toward the horizon. The view is pleasant - whitecapped mountains under a clear sky. At the foot of the mountains the ground plane meets a lake that extends beyond our view to the right and left. Getting to the beach is a problem, though.

As we approached the mountains the ground plane kept extending. New chunks of grass grid appeared as our zone of resolution advanced with us. The edge of the plane - the beach - was always another hundred feet away. It turned out that there was no lake at all. The blue at the foot of the mountains was the gap between the ground plane and the cylindrical scrim. Flying above or below the ground plane revealed the model's artifice and the curve of the horizon.

Movement and the Shape of Communities

A view from above AlphaWorld presents the viewer with a plan that resembles Los Angeles or some other urban sprawl. Houses, buildings and swimming pools litter the ground lane. Despite a grid of streets at the city center, roads are intermittent and disconnected - sometimes stopping after a few hundred feet. Often they lead nowhere, fragments isolated at the edge of the community.

And yet, despite this discontinuity, AlphaWorld has a structure. Beyond its core, the city extends in eight directions. From above, the plan resembles a star with eight rays diminishing as they project from the center. The rays extend in the four cardinal directions as well as northeast, northwest, southeast and southwest. Oddly, no roads project diagonally from the middle of the city. Indeed few diagonal roads exist at all. What would lead the citizens to build such a symmetry despite their otherwise haphazard, idiosyncratic construction? MUD construction was apparently left to the individual with little interference from the administrators. The absence of diagonal roads also indicates that the rays of construction weren't planned by the wizards.

Despite the absence of diagonal roads, the mystery is resolved by examining movement within the domain. A citizen can build anywhere that he finds an open, unclaimed piece of ground plane. AlphaWorld's core configuration reveals the grid used in laying out the original community. This area is dense with construction including residences and public buildings. It also contains the entry point of the community - where visitors log in.

Dynamic movement around the community takes time. If one is not inclined to roam the space, teleportation provides a quick alternative. Teleportation is also accurate. Users can enter an "address" and simply wind up at the destination with no guess work. This is accomplished by entering coordinates in the cardinal directions or distances in any of the four subsidiary directions.

Apparently, as the community ougrew its rectilinear core, it became more difficult for new builders to find empty plots for construction. Roaming around the grid of streets was laborious and unfruitful. Since space was plentiful at the edge of the core, builders would teleport there using any of the eight directions. Upon arriving they would wander around, find an available site and build.

Of course, a builder proud of her work would want to return periodically to enjoy or modify her project. Situating a structure at an arbitrary point would be hard to recall later. Consequently, a prudent builder would choose a site whose Cartesian coordinates were easy to remember - preferably one with the same x and y values or either x or y value being zero. Similarly, a simple distance in one of the secondary directions could pinpoint the construction site. The resulting pattern of AlphaWorld's development closely follows the options provided by teleportation (fig. 111).

Movement within a community, physical or otherwise, is critical to its form and reflects its administration. The shapes and locations of cities from ancient to modem times have been largely determined by their mode of transportation. Bruges and Amsterdam are characterized by the canals used for moving goods and people. Similarly, Houston and Los Angeles are inconceivable without the automobile. We have seen that LAMs of text MUDs are shaped by directional and categorical movement designed into their spaces. Like other dynamic, urban and on-line communities the configuration of AlphaWorld is based on its preferred mode of movement, teleportation.

Conclusion

In the next chapter we will see how the illusions of cyberspace can overlap with actual spaces and people. This is a theme we will pursue throughout the rest of this book.

1. The acronym stands for "There Ain't No Such Thing As A Free Lunch," perhaps a reference to the need for order to maintain a community.

2. Worlds Within also supports voice communication, text-based chat and collaboration through a shared whiteboard, where up to 8 people can work together on one file and store it in the environment.

Merging Physical and Mediated Realities

This chapter describes the bridging of cognitive and perceived space through mediation. It presents collaboratories and work environments that incorporate physical and cyberspaces and will discuss the human and social issues that attend them.

So far we have seen abstract, on-line communities as independent of their physical counterparts. But what if the fantasy were taken away? What if the theme park became themeless? How would MUDs work if people used their real names, the roles they played - themselves? What if the spaces of the MUD reflected conditions of their physical counterparts? What bridges would link our electronic, cognitive space to our perceived world?

While the prospects of mixing illusion and physical space seem troubling, in fact we already use our technology this way. We accept that our telephone partner appears to be next to us. Teleconferencing similarly brings others into our range regardless of their location. Even without technology, our processes of perception and cognition are so complex that separating illusion from reality is itself daunting (figs. 112,113,114).

However, the third place¹ of multi-user domains merges the two. These spaces and avatars often float free of physical location and identity. Few MUD users know the true identities of their fellow citizens. While this is unnecessary for those who accept a clear division between the fantasy of the MUD and real life, the linkage between the user, his avatar and society is subtle and a potential resource for the creation of new kinds of mediated communities.

This third place, rather than solely being a moratorium space, of virtual space, a CAD model and a live video. could become a place of engagement where users gather to exchange ideas. In his proposal for a simulation of the physical world, Gelernter describes how those who use it would be better informed citizens of the world. Although Gelernter's proposal is not a multi-user domain as we have described it, it is spatial and shared by many as a common resource, like Wladek Fuch's VCNet. By bringing the information of the real world into cyberspace, the real world may be better managed by such an enlightened populace.

Political Implications

Without physical boundaries, a domain can host citizens connecting from other countries. Already the politics of cyberspace have begun to affect "real" politics. Attempts by the German government to bring suit against Compuserve for trafficking sexually explicit material shows how terrestrial politics are challenged by those of cyberspace. Given trends in technology and global politics we can imagine that Internet sites specific to interests could become distributed homelands regardless of where their citizens live. These become the secular equivalents of other places of mind - Valhalla, Hades, Paradise - shared by many yet inhabited by none.

But the crucial difference is that, unlike their predecessors, these environments are manifest to the senses. Some domains are also information resources, forming an abstract agora for a distributed global population. These domains are actually a curious mix of memory palace and agora, for despite their ephemerality, they contain fixed objects and mnemonic structures. Not surprisingly, computer memory is the very substance of their being. As we have seen, aspects of MUDs often parallel the perceived qualities of our world. We use it both as site and subject of our public lives.

Collaboratories: On-line Work Environments

These workplaces without walls are called collaboratories, a term coined by William Wulf in 1989. They are characterized by their ubiquity and mediation through computer networks letting researchers in various fields - physics, astronomy, engineering - share resources, access to instrumentation and comraderie. For example, the Upper Atmospheric Research Collaboratory - UARC - whose server is located in Michigan, lets scientists share instruments in the study of "space weather," magnetic storms that disrupt transmission of power and communication. Such environmental monitoring is also the focus of the Collaboratory for Environmental and Molecular Sciences based at the Pacific Northwest National Laboratory in Richland, Washington.

Gary Olson, a cognitive scientist at the University of Michigan and creator of UARC, believes that the potential for collaboratories extends beyond research into the corporate world. Collaboratories are increasingly seen as media for the distributed work of multinational corporations. Automobile and furniture manufacturers as well as high-technology companies, he says, are highly interested in this.³

If the focus of many collaboratories was originally access to instrumentation and data, the inclusion of chat boxes and virtual rooms has turned them into on-line social environments. While not conventional MUDs, collaboratories share many of their attributes. Their intimacy and collegiality has a socially leveling effect - typical of MUDs - which lowers the barriers to dialog. A student logging into a collaboratory may have an otherwise unlikely conversation with a senior researcher.

This leveling of hierarchy and lowered barriers found in online environments is also useful in a corporate setting. Lisa Kimball, a director at Metasys in Washington,D. C., recalls a project for a client that was undergoing a merger with another company. In an effort to manage the stress of reorganization, the corporation retained Metasys, a provider of on-line corporate environments. Metasys created a multi-user domain in which various rooms housed discussion groups. Corporate employees and management would log in to converse, leave messages and vent their frustrations. They found that the moratorium of the domain facilitated discussions and that, in real life, the employees would pursue their on-line dialogs in the company hallways and offices. The overlap of the virtual onto the real helped the company through a time of emotional and organizational stress.

MUDs have found their way into the corporate environment. Chiat-Day, the advertising firm, created a graphic domain for its employees when it set up its operations in Manhattan in 1993. Designed by Art and Technology's Jeet Singh, it incorporated messaging and chat spaces, including a bar where employees could socialize whether or not they were concurrently on-site.

More recently Fore Systems, a developer of computer ATM networks, installed a graphic MUD to disseminate technical support to its non-technical staff. Using a client-based software from The Palace Inc., Fore created a 24-room intranet environment based in part on the physical layout of its campus. James Carlin, manager of Fore's marketing help desk, believes that information gathered during on-line dialog can be used to enhance their customer services. For instance, the ideas generated can be incorporated into reference files.

At this time only twenty percent of Fore's sales staff is physically present at its Pittsburgh headquarters. The company looks toward expanding this service into extranets so that its remote sales force can benefit as well. While graphic chat environments are rare on intranets, their use on the Internet is rising. ⁴

Reality Checks

Relationships between mediated workers and their on-site counterparts are also important. A study conducted by Eleana Rocco at the University of Michigan showed that groups that had face-to-face contact prior to on-line collaboration displayed more cooperation than those deprived of it. She had subjects play a computer game that required collaboration as well as competition. Groups that physically met for five to ten minutes prior to going on-line established a level of trust that the control group lacked. ⁵

Gary Olson and his colleague Tom Finholt have closely monitored interaction in collaborators domains. They have found that, though the topics of conversation are the same as they would be in real life, the quality of ideas is higher Though they are not sure why this is, they note that spoken conversation is often comprised of short statements. Speakers often don't have time to develop ideas before the conversation moves on. The presence of text may compel correspondents to consider their thoughts before committing them to the screen. Apparently the increased flow of ideas is a chief advantage of these corporate and research domains. The medium seems to demand it.

While on-line environments offer much, they will never replace the innately physical aspects of social life. We have seen in earlier chapters the complex role our bodies play during social interaction. Despite advances in human-computer interface design, we are accustomed to a quality of communication that no computer can yet convey. Though we may exchange information in a collaborators, we can't shake hands or feel each other's presence.

Conclusion

In the next and final section of the book we will pursue this link between physical and cyberspaces. My aim in reconciling them is not to minimize their differences so much as to point out their collaborative strengths. Each, however distinct, extends us beyond ourselves to each other. Reconciling them may reveal new ways of sharing our common experience.

1. Here I refer to Ray Oidenberg's concept of an informal social place.

2. Curtis, Pave], and David Nichols. 1993. "MUDs grow up: Social virtual reality in the real world." Paper presented at Cyberconf 3 in Austin, Texas. Curtis has since founded PlaceWare, Inc., which develops software to support professional MUDs.

3. Ross-Flanagan, Nancy. 1998. "The virtues (and vices) of virtual colleagues." MIT Technology Review. March/April. pp. 54-59.

4. Murphy, Kathleen. "Graphical chat makes move from games to intranets." lnternet World, vol. 4, Issue 7, Feb. 23, 1998.

5. Ross-Flanagan, op. cit., pp.58-59.