Visit for more recent material

Chapter Six: Real-time Tribes/

By Howard Rheingold

Thousands of people in Australia, Austria, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan, Korea, Mexico, the Netherlands, New Zealand, Norway, Spain, Sweden, Switzerland, the United Kingdom, and the United States are joined together at this moment in a cross-cultural grab bag of written conversations known as Internet Relay Chat (IRC). IRC has enabled a global subculture to construct itself from three fundamental elements: artificial but stable identities, quick wit, and the use of words to construct an imagined shared context for conversation. For a student of virtual communities, IRC is an opportunity to observe a critical experiment-in-progress: What are the minimum elements of communication necessary for a group of people to cocreate a sense of community? What kinds of cultures emerge when you remove from human discourse all cultural artifacts except written words?

An artificial but stable identity means that you can never be certain about the flesh-person behind an IRC nickname, but you can be reasonably certain that the person you communicate with today under a specific nickname is the same one who used that nickname yesterday. There's nothing to stop anybody from getting a new nickname and creating a new identity, but both the old and the new nicknames have to be unique. The stability of nicknames is one of the few formally structured social requirements in IRCland; an automatic "Nickserv" program ensures that nobody can use a nickname ("nick") that has been registered by someone else.

Quick wit is necessary because rapidity of response becomes important in this written medium in the same way it is important in face-to-face conversation. IRC is a dynamic form of communication: new comments appear at the bottom of your screen as you watch, and older comments scroll off the top of your screen. Somewhere in the world, a human being has typed those words on a keyboard, no more than a couple seconds ago; if you know the right words to say in response, you can leap into the conversation and make that person and others around the world laugh out loud, grow angry, feel lustful.

Perhaps one attraction of IRC is that the conversation literally continues to move up your screen as you watch. You can treat IRC as a spectator sport, never venturing into the flow. Or you can show the IRC tribe how fast you are with a well-worded rejoinder by keeping up with other participants in a rapid interweaving of cleverly linked comments. IRC is a stream. Many people who work for long hours in front of computer workstations--college students, computer programmers--leave a small "window" on their computer screens tuned in to IRC while they go about their day's work. They have their own automated programs, known as bots, to greet newcomers and say goodbye to people who leave. When they see something interesting going on, from the corner of their eye, they jump in.

The initial absence and subsequent reconstruction of social context is the third fundamental element that IRC enthusiasts use to build their subculture. Without facial expressions, tone of voice, body language, clothing, shared physical environment, or any other contextual cues that signal the physical presence of participants in a social group, IRC participants use words alone to reconstruct contexts in their own image, adding imagined actions (such as "Howard smiles ironically" or "Howard takes offense and it looks like he's going to punch you in the nose") as metadescriptions to the running dialogue. These virtual actions are typographically set apart from words meant as straight dialogue. The "actions" in IRCland are the same as "poses" or "emoting" in MUDs, and serve a similar purpose. They add a modifier to the strict definitions of words, indicating intention, mood, or other contextual cues.

Those thousands of people tuned into IRC at any one time are divided into hundreds of "channels" that Internet users can "join" or "leave" at any time; like Usenet newsgroups, the channels operating at any one time include a rich variety of topics, from the scholarly to the obscene. Farflung business groups, task forces of technical experts, and scholars also use IRC to do real work. I have participated in IRC discussions of electronic publishing, organized by the participants in an electronic mailing list who wanted to add a real-time dimension to our collegial relationship. This variation of CMC wasn't invented specifically for real work, however. IRC was invented as a means of playing with communication, and that remains its most popular use.

IRC is what you get when you strip away everything that normally allows people to understand the unspoken shared assumptions that surround and support their communications, and thus render invisible most of the web of socially mediated definitions that tells us what words and behaviors are supposed to mean in our societies. You can't see people when you are computer-chatting with them; you can't even ascertain their true identities, and you are unlikely ever to run into them on the material plane or recognize them if you do. Chat systems lack the community memory of a BBS or conferencing system or MUD, where there is some record of what was said or done in your absence. Although words are written and broadcast (and thus can be electronically captured, duplicated, and redistributed by others), they aren't formally stored by the chat system. The discourse is ephemeral.

Despite the anonymity and ephemeral nature of their communications, IRC habituees become addicted, form close friendships, fall in love. There is even a Usenet newsgroup--alt.irc.recovery--for IRC addicts and former addicts to testify, support, and debate each other. You sense, even from a brief visit to IRCland, that many of these people have built a kind of community that they would defend as passionately as the most committed WELLite, habitual mailing-list participant, Usenet veteran, or accomplished MUDder.

Chat systems that enable one person to send typed words directly to the screen of another person who is logged onto the same system date back to the first time-sharing computers of the 1960s. In this regard, they are probably the oldest form of CMC, predating electronic mail. By storing electronic messages in a private computer mailbox until the recipient logs in and reads it, electronic mail differentiated itself from existing chat programs that required both participants to be logged into the computer at the same time. But the need for a kind of direct conversational form of CMC continued to flourish as asynchronous many-to-many written media emerged. The CB service on CompuServe is probably the oldest, continuously operating, commercial chat service in the world. Minitel's messageries in France are into their second decade of a wildly popular, national, government-supported, written-word chat system. Fujitsu's Habitat combines a cartoonland graphical representation of participants and their environment with a synchronous chat system. IRC is Internet's pioneering multi-user chat system. IRC is the corner pub, the caf‚, the common room--the "great good place" of the Net.

Most computers connected to Internet run a program called "talk" that enables people on different host computers to communicate screen-to-screen simultaneously. If I suspect that a person I know is online at a different site, I can issue a chat request through Internet. If that person is logged in, even if the computer is on the other side of the world, my chat request will go directly to the recipient's screen for immediate attention (rather than going to an e-mailbox for later attention). If the recipient agrees to chat, we can communicate by typing to each other's screens in real time. I see my words on the bottom half of the screen, and the other person's words on the top. It's the only conversational medium in which both participants can talk and listen simultaneously, which can be liberating for typographically loquacious types who are likely to be found online. But "talk" involves only one-to-one communication.

In 1988, Jarkko Oikarinen at the University of Oulu, Finland, wrote the original IRC program, a multi-user, synchronous communications tool designed to work over Internet. First it was tested on a local community of twenty users and then installed throughout the Finnish national network and ultimately the Scandinavian portion of Internet. IRC--the software needed to access the medium, as well as word of its attractions--propagated throughout the wider Internet by the end of 1988. By the early 1990s, there were hundreds of channels and thousands of people chatting across the Net, twenty-four hours a day. Eventually, scientists and scholars began to get the word that IRC was a way to convene informal discussions among geographically distant colleagues, but the continuing popularity of IRC appears to be primarily a function of its appeal as a psychological and intellectual playground.

Why should taxpayers support anyone's playground? It's a good question, and it goes to the heart of the current debate over privatization of the Net. It pays to keep in mind a bit of history when addressing this question, however, because playgrounds have played a critically important role in the evolution of computer technology. The very earliest creators of interactive computing also created the enormously popular Spacewar game in the early 1960s, which turned expensive research computers into powerful versions of what would come to be known in the 1970s as video games. At top computer science research centers, the very people who were supposed to be creating the next generation of computer technology were staying up late to blast each other with proton torpedoes on primitive graphic display screens. Note that most attempts to eliminate Spacewar from computer centers failed. When administrators shut down Spacewar activities at their centers, they found that programmer productivity went down instead of up. Spacewar lived on despite bureaucratic opposition because its enthusiasts were also their research projects' best programmers.

The government-sponsored researchers who built the first ARPANET started using government resources to exchange e-mail about science fiction. ARPA administrators wisely invested in building even more powerful CMC systems when the SF-LOVERS list began to eat up more and more of the communication capacity of the system. In computer technology, playgrounds often are where real innovations emerge. Those who would characterize IRC as a frighteningly cerebral and mechanically dehumanized form of interaction that has no conceivable socially redeeming value would do well to recall the cases of Spacewar and SF-LOVERS.

IRC does not fit well with conventional theories of human communication because CMC technology makes possible something that human communicators could not do previously--a geographically dispersed group of people now can use the written word as a conversational medium. So much of what scientists and scholars know about human communication involves physical presence or even potential physical presence, both totally absent from IRC. The telephone has more physical presence, more of a direct sense of the living being behind the words. Words, and the elegance of expression and timing that accompany their use, exist in a purely disembodied state in IRC.

Subcultures based on chat programs constitute a rich, uncharted domain for social research. Like Amy Bruckman, who opened the phenomenon of MUDs as a legitimate topic of media research, and Marc Smith, who investigated online cultures from the perspective of sociology, Elizabeth M. Reid was a graduate student who found her way to the Net and realized that cyberspace is a living laboratory for some of the hottest theoretical controversies in her discipline, history. "Electropolis: Communication and Community on Internet Relay Chat" was Reid's honors thesis in 1991 for the University of Melbourne. The full text was widely circulated through the Net. Reid's central thesis is that "IRC is essentially a playground. Within its domain people are free to experiment with different forms of communication and self-representation." From that communication playground, Reid claims, IRC habituees have evolved rules, rituals, and communication styles that qualify them as a real culture according to criteria defined by prominent social scientists.

Reid's thesis derives from IRC's reversal of the role of social context in shaping conversation and community. In the material world, social conventions are built into houses and schools and offices, signaled by modes of dress and codes of etiquette, posture, accent, tone of voice, and hundreds of other symbolic cues that let people guess accurately how to behave in a particular social situation or society. People learn how to adjust their behavior to conform with a learned mental model of conventional behavior. Until the era of electric communications media, almost all the cues people used to ascertain social context in communications were more physical than verbal. In IRC, however, participants react to a world stripped of nonverbal context by re-creating the context that has been lost. They do this by using written words to describe how they would act and how the environment would appear in a shared mental model of a wholly constructed world. People can "swoon" or "charge to the rescue" or "look askance," and they can do it from horseback, the back of a limo, or the far side of the moon.

Reid's theory is that IRC participants use the lack of context and geographical separation to create alternative communities, with written-word versions of several of the essential tools face-to-face communities use to encourage solidarity: "Both positive and negative methods of sustaining community are developed on IRC. Computer-mediated rewards and punishments are developed, and complex rituals have evolved to keep users within the IRC `fold' and to regulate the use of authority." The main taboo is using a computer hack or other subterfuge to adopt the nickname of another user. Reid cited heartfelt "confessions" on a Usenet newsgroup devoted to IRC enthusiasts as the example of an expiation ritual for a subculture member who has committed its ultimate taboo. Reid noted that channel operators (chanops) who administer individual channels, and IRC operators (opers) who voluntarily keep the IRC service running on the Net, have special powers that can affect users. Chanops can kick people off a channel; opers can ban people logging in from specific Internet sites from participating in IRC. But these powers are never used without explanation, even though there is no official governing body that requires such explanation.

Violating the sanctity of nicknames is a taboo because it attacks one of the fundamental forces that holds the IRC culture together--a minimum certainty about the identity of all participants in discourse. According to Reid, "The uniqueness of names, their consistent use, and respect for--and expectation of--their integrity, is crucial to the development of online communities." Public confession, Reid points out, is a ritual that reaffirms the taboo and allows the confessor to rejoin the fold. Explanation messages for chanops and opers are voluntary rituals that constrain the power of authorities while upholding loyalty to the hierarchy of power that maintains IRC's minimum of social order.

These aspects of IRC behavior noted by Reid fit several characteristics of culture as defined by contemporary anthropologist Clifford Geertz. Culture, according to Geertz, is "a set of control mechanisms--plans, recipes, rules, instructions (what computer engineers call `programs')--for the governing of behavior." Referring to Geertz's definition, Reid concludes, "In this sense the users of IRC constitute a culture, a community," because they have devised elaborate social mechanisms for addressing the problems posed by the medium.

Taboos and expiations are negative social control mechanisms. The IRC culture's most powerful positive mechanism for maintaining solidarity is peer recognition. The ability to use words to create context by describing imagined actions and environments is one key to high social status in IRC culture. Those people who are most skilled at creating the illusions of shared context (and thus those who most strongly uphold the elements of the imagined community) are those who are most rewarded with words of recognition and expressions of affection by their peers. Like a relative newcomer to a pub or other informal public space, you know you've arrived in the social hierarchy of an IRC channel when the regulars begin to greet you heartily when you arrive, instead of sending you the kind of canned greetings that bots provide. Personal attention is a currency in IRC: everyone is on stage who wants to be, everyone is the audience, and everyone is a critic.

The notion of mastery that Sherry Turkle and Amy Bruckman mention in regard to the popularity of MUDs among college students generalizes to IRC. The written composition of poses to accompany actions requires creativity, quick thinking, imagination, and either a literary sensibility or the style of a stand-up comedian. Many IRC participants, like many MUDders, are undergraduates. Some, who might not find their skills of quick description and witty rejoinder--a combination of applied daydreaming and advanced bull session--fully appreciated in other parts of their lives, find that they are higher in the IRC social hierarchy. The regulars in an IRC channel might be the first "in-group" that ever accepted some of these people socially.

During informal monitoring of the alt.irc.recovery newsgroups, I noticed frequent confessions that constant IRC use led chat addicts to spend less time with physical friends in the immediate geographic neighborhood; in the long run, this reclusiveness led to loneliness even more acute when they weren't logged onto the one community in which things went smoothly. Like the question of MUD addiction, any question about proper ways to use a communications medium leads to deeper questions about what our society considers to be proper ways to spend time. If a lonely person chooses to spend many hours a day in an imaginary society, typing witticisms with strangers on other continents, is that good or bad? By what authority should any person or institution have the right to intervene in such cases?

The lack of systematic observations of real case histories, beyond the anecdotal evidence of Usenet, makes any conclusions about the dangers of IRC addiction premature. Enough people think IRC is dangerous to their emotional health to keep an online support group alive; that constitutes evidence that a social problem might exist. We need to learn more about the dimensions of the problem. The serious questions about the possible negative effects of this medium sharpen the need for serious scientific attention to the social dimensions of cyberspace. Banning MUDs or IRC activity is one natural bureaucratic impulse in response to tales of addiction. Do we need to create a new kind of online cultural policing mechanism in response to this problem, or would effective prevention and treatment of obsessive communication behavior, on a case-by-case basis, be a wiser strategy?

The most popular continuing IRC channels are $SGhottub, an ongoing flirtation space, mostly heterosexual, and $SGgblf (gays, bisexuals, lesbians, and friends). Sexy chat and chat about sex, net.sleazing and gender-switching, are as well known in the IRC community as they are in MUDland. And just as there are by now dozens of cases of MUDders who met in a fantasy world and eventually married, although they lived on different continents when they met, there are similar cases in IRC where informal chat led eventually to matrimony. I spotted this electronic wedding notice in a Usenet discussion of IRC, formatted electronically to simulate the more traditional kind of announcement:

In deference to the circumstances of their meeting,

Karl "poptart" Kleinpaste


Deborah "KarlsGirl" Brown

are pleased to announce to the IRC community

their engagement,

which became a fact on

the evening of 7 August 1992.

A wedding is planned for May 1993,

its location being anyone's guess just now,

following Miss Brown's graduation from

Case Western Reserve University School of Law.

The couple will be residing in

Pittsburgh, Pennsylvania


I sent e-mail to Karl "poptart" Kleinpaste, asking about the circumstances, and he replied:

About 4 months ago, Debbie was looking for someone to chat with, scanning through a /names list. She said hi to me because (as she said later) I had a "cute, non-scary" nickname. We chatted a little bit just casually then and for a few days, then it became a pretty much every-day occurrence. We had a lot in common, and after a month or so, I invited her to visit me here in Pittsburgh. (She lives in Cleveland.) She came down for the first time over Memorial Day weekend. We've managed to spend every weekend since together, alternating which one of us travels to the other's city, as well as continuing to talk daily in IRC or, more recently, on the phone. We went out for a special dinner last evening, I asked her to marry me, and she said yes. We'll be getting married in May of next year, probably.

We're waiting until next year due to logistics; she's entering her 3rd and last year of law school. But now she's shifting her plans to take the Pennsylvania bar instead of the Ohio bar, and will be looking for a position with a Pittsburgh firm.

Just goes to show, and contrary to the opinions of many, relationships that start in IRC really can work out after all . . .

Love and matrimony--net.romances--might represent one of the more socially attractive examples of IRC as a communications medium. Other examples are less positive. Sara Kiesler and others who have studied the social effects of CMC in business groups have noted that the lack of social context cues encourages both positive and negative disinhibition. Normally shy people react by speaking up, and people who would never shout at others or hurl insults in a physical gathering sometimes behave that way online. Reid notes that regulars on IRC channels tend to be uninhibited about self-disclosure, discussing serious problems in their real lives with their online friends. Such channels are self-regulating in the sense that newcomers who take advantage of this openness and trust by insulting others can be "killed" by the chanops. And IRC includes a mechanism for countering the insularity of established channels with cliques of regulars who define the criteria for membership. Any IRC user can create a new channel, and thus become a chanop, at any time. There are private channels, as well, but the majority of traffic these days is in the public channels.

Antisocial behavior is not rare in IRC. Racist and homophobic outbreaks are regular events. Hostile individuals can program their bots to flood a channel by sending out a stream of characters to IRC. It's very annoying when you are devoting part of your desktop computer screen and part of your attention to a conversation on an IRC channel and someone's bot somewhere in the world starts sending profanities or scriptures or dictionaries or gibberish to everyone else's screen. The same lack of social feedback that lowers inhibitions enough to promote self-disclosure among groups of people can also lower inhibitions enough for individuals to disrupt those groups and sometimes tear the delicate fabric of trust that has been carefully woven over months of conversation among disembodied strangers.

Reid points out that IRC is a global cultural phenomenon, with high potential for citizen-to-citizen communication across national and ideological boundaries:

It is not uncommon for IRC channels to contain no two people from the same country. With the encouragement of intimacy between users and the tendency for conventional social mores to be ignored on IRC, it becomes possible for people to investigate the differences between their cultures. No matter on how superficial a level that might be, the encouragement of what can only be called friendship between people of disparate cultural backgrounds helps to destroy any sense of intolerance that each may have for the other's culture and to foster a sense of cross-cultural community.

Just as the stories of addiction are evidence of a social problem that deserves closer attention, the possibility of a new tool for fostering cross-cultural understanding is worth a closer look as well.

The most widely cited instance of cross-cultural dialogue on IRC is the use of the medium during the Gulf War. An Internet link had been set up in Kuwait before the Iraqi invasion, and the link stayed up for a week after radio and television broadcasts ceased. Kuwaiti students used IRC for eyewitness reports, and so did Israelis. According to Reid, "I am told that users from the two countries often interacted with very few disagreements and mostly with sympathy for each other's position and outlook." There has been some discussion of the role of the Net in the Gulf War, via Usenet.

From: frechett$HPspot.Colorado.EDU (-$SIRunaway Daemon$SI-)

Subject: Re: IRC Folk History

The channel at the time was called $SGreport. It was moderated by a bot first run by lynx and then by myself. Basically it was a moderated channel and a number of people around the world would come onto the channel and listen. If someone had something to report then they would change their nickname to something like NBC or CBS or AP. etc. . representing some news reporting agency. (Just people listening to the news though, not real representatives of that agency) Those people would get chanop and could report the news as it came in.

One of the notable things I recall was an Israeli on the channel (was in the Wall Street Journal too I believe) who had a terminal in the sealed room and would talk about an incoming scud alert and then would come back a few minutes later typing at us from in the sealed room.

I also recall one of the Isreali's [sic mentioning to us when he heard one of the first scuds hit that he heard Isreali jets scrambling and flying west. This was a bit of information I never did hear through CNN, AP or the network news.

A little research on the Net led me to the archives where the logs of the Gulf War IRC sessions are kept. Reading through a transcript certainly re-creates that feeling of news hunger that surrounds the fog of war during the early hours of a major conflict. Here are a few selected excerpts from the IRC channel that preceded the $SGreport channel. Ironically, these remarks took place among participants in the $SGpeace channel:


IRC Log started Thu Jan 17 01:03

$SAumfonta6$RL bombsd are droppiong in baghdad

$SAhstanley$RL jeesh

$SAspamgod$RL am i getting through?


$SOBreeze$SO has CNN on

$SAArkie$RL this channel name is fairly oxymorninical [sic.

$SAspam$RL koala* 20 minutes since first attack

$SAspamgod$RL thanks

$SAhstanley$RL who's in baghdad here?

$SAMark$RL scoop? hah!

$SAumfonta6$RL bombs are happening

$SAMosiah$RL bombs hitting the net!

$SACaptainJ$RL 2 major explosions-1 near major comm center.

$SAumfonta6$RL phones are going down

$SAstealth$RL Attack started at 2:30am baghdad time.

$SALipstick$RL I dont think bagdahd is on the net $SI

$SABOY$RL No arab countrys on the net..

$SADatawolf$RL hold for net diagnostic..

$SAPhaedrus$RL welost the finnish d00ds


$SAstealth$RL Boy and Ely are on from israel.


$SAstealth$RL Got your gas mask ready, Boy?

$SACaptainJ$RL AA-fire still going up-very random (CNN)

$SA$SABOY$RL steal: yes .. and we just finished silling the house from gas. .

$SAAlexander$RL WAR!!!!!!!!!!!!!!!!!!!!!! WAR!!!!!!!!!!!!!!!!!!

!!!! WAR!!!!!!!!!!!!!!!!!!!!!! WAR!!!!!!!!!!!!!!!!!!!!!!

WAR!!!!!!!!!!!!!!!!!!!!!! WAR!!!!!!!!!!!!!!!!!!!!!! WAR!!!


$SOHamlet$SO The liberation of Kuwait has begun

$SABertin$RL Hallo bardo Willkommen in $SGpeace. Hallo chris Willkommen in $SGpeace.

$SA$SAWill$RL u guys lets please keep this channel english only

Like MUDs, IRC has the potential to develop into a more serious communications medium. Can chat be a tool as well as a toy? Jupiter and MediaMOO are attempts to create MUDs without the dragons, substituting informal discourse among intellectual colleagues for symbolic combat between imaginary adventurers. In IRC, the equivalent to taking away the dragons would be for colleagues who share a specific interest to meet on an IRC channel at a specified time every day or every week for the kind of interaction that is best done synchronously--brainstorming is the primary example of the kind of group interaction IRC could facilitate, if used adroitly.

At the point where groups in academia or private enterprise start using MUDs and IRC channels as communication tools, the way they started using word processing in the 1970s and e-mail in the 1980s, these CMC playgrounds converge on the discipline known as computer-supported cooperative work (CSCW). CSCW has been the focus of international, interdisciplinary conferences, scholarly journals, and corporate consultants for the past several years. It's part of a movement among many information systems specialists in the business world to embrace the notion of groupware--a term first coined by hard-core CMC revolutionaries.

Computer-supported cooperative play, however, is an online dimension in its own right.

Habitat: Computer-Mediated Play

When I first encountered the phrase "computer-supported cooperative play" in a scientific paper from the Kyoto Institute of Technology, several ideas that had been floating around in my head began to fit together. CSCW is promising, but it is a strictly utilitarian perspective. If industrial civilization has a taboo so engrained in the culture that few even recognize it as a taboo, play may be it. Play is somehow not quite proper behavior if you aren't an infant or a professional athlete. People are expected to spend their time "constructively." Working, educating oneself, parenting, even organized commercial recreation such as watching television or taking an elaborately planned vacation, are considered constructive. Playing around, however, is not considered constructive. Yet the history of CMC reveals that people often use the new medium to do just that. And some educational psychologists, most notably Jean Piaget, claim that play is the way humans learn best.

Atsuya Yoshida and Jun Kakuta, of the Department of Information Technology at the Kyoto Institute of Technology, proposed the term "computer-supported cooperative play" in their study "People Who Live in an On-line Virtual World":

We would name the Habitat type of communication as a playful communication. Historically, we have the example of the playful communication such as a wall painting in a cave and a drumming behavior to neighboring villages at a festival, while Indian's signal fire and military flag signals are the examples of task-oriented communications. Playful communication could be a key concept in the next generation human interface architecture and the Habitat presents a good model for the study of the concept of playful communication, that is, computer supported cooperative play.

The Habitat that Yoshida and Kakuta discussed as an example of an online community devoted to cooperative play is a project of the Japanese computer-aerospace-electronics giant Fujitsu. Fujitsu's Habitat is based on an older, pioneering, graphical virtual community first developed by Lucasfilm Games and an early commercial online service in America, QuantumLink Communications. I observed the Japanese version of Habitat when I visited Fujitsu's research-and-development facility in Kawasaki, but my first encounter with Habitat's original architects took place years earlier. In Austin, Texas, in 1990, at the First Conference on Cyberspace, I met the two programmers who created the first large-scale, multi-user, commercial virtual playground.

In their address to the conference, and the paper they later published, "The Lessons of Lucasfilm's Habitat," Chip Morningstar and F. Randall Farmer recounted their experience as the designers and managers of a virtual community that used computer graphics as well as words to support an online society of tens of thousands. Much of that conference in Austin was devoted to discussions of virtual-reality environments in which people wear special goggles and gloves to experience the illusion of sensory immersion in the virtual world via three-dimensional computer graphics. Randy Farmer and Chip Morningstar stood out in that high-tech crowd because the cyberspace they had created used a very inexpensive home computer, often called a toy computer, and a cartoonlike two-dimensional representation to create their kind of virtual world. Farmer and Morningstar had one kind of experience that the 3-D graphics enthusiasts did not have, however--the system they had designed, Habitat, had been used by tens of thousands of people.

In the early 1980s, LucasArts Entertainment began a series of ambitious research ventures through the Lucasfilm Games division. Morningstar and Farmer were handed the tempting and vexing task of designing a graphic virtual community for a large population that was using computationally puny Commodore-64 computers. Before the Macintosh came along in 1984, the Commodore-64 was the world's most popular home computer. The "64" stood for the 64 thousand bytes of main memory. Today's desktop computers use millions of bytes of main memory, which is why the C-64 is considered a toy. But Lucasfilm Games had made a business deal with QuantumLink Communications, an online service that connected C-64 users via modems.

Morningstar and Farmer were given the design constraints of a population of users numbering in the tens of thousands, hundreds of whom would be online at any one time, each of whom would have a laughably small amount of communication bandwidth and home computers that were barely more intelligent than hand-held calculators. Within these constraints, they created a graphical chat system that enabled tens of thousands of participants to play games, build societies, engage in politics, fight and even "kill" each other, start religions, wage wars, fall in love, get married, and create markets and economies. The central metalesson the system designers had learned from Habitat was that "detailed central planning is impossible; don't even try." Create the tools for users to build their own society, Morningstar and Farmer concluded, and let the users tell you what they want to do, because that is what the users of an online communication system will do, no matter how hard you try to structure some other purpose into the tool.

The C-64 computers acted as the front end to the Habitat system. When you logged onto the service, the front end would create an animated computer graphic model of Habitat on your screen. The user interacts with the Habitat world by moving around a joystick, and by means of a cartoonlike representation known as an Avatar. Avatars are mostly, although not exclusively, humanoid, and have heads and bodies that can move around the space depicted on the screen. The Avatars of multiple users are visible on the screen, as well as external objects and parts of the environment, like coins or guns or trees. The thousands of individual users were linked to far more powerful computers that constituted the back end to the Habitat system. The back end kept track of who was where and which objects were lying around--the "world model." When you type something on your keyboard, the back end takes the message from your desktop computer and displays it to anyone else who is in the same region of Habitat by means of a speech balloon that appears above your Avatar's head. If you pick up or drop an object, the back end moves the object in everyone else's world model.

One thing that Avatars, unlike IRC users, can do in a real-time chat is use gestures such as nodding and eye movement to indicate unspoken feedback while others are speaking. In normal conversation, these phatic utterances and gestures are an important if invisible part of navigating from topic to topic or understanding unspoken nuances. The conversation does seem more alive in a sense, when you can move your eyes toward someone's Avatar when it says something interesting. Besides moving their heads and changing their facial expressions, Avatars can move from place to place and pick up, put down, and manipulate objects. Dozens of classes of objects, including examples such as Tokens, act as Habitat currency, a Change-o-matic for changing an Avatar's gender, a club, gun, or knife for protection in some of the rougher neighborhoods, drugs to cure wounds.

Habitat was designed with a spatial metaphor in mind. Each user can see a region of Habitat at one time on the screen. The Avatar can move to one of four adjacent regions from any region by moving through a door, or use a teleport booth or device to move instantly to a distant region. Habitat was designed to hold twenty thousand regions. Each user was given a private region, known as Turf, to store objects, hide out, and invite guests.

Morningstar and Farmer reported that their software engineering backgrounds led them to think of the design of Habitat in terms of central planning. For twenty thousand Avatars, there had to be twenty thousand houses, clustered into towns and cities. Thoroughfares, wilderness areas, and shopping centers were needed. "We, attempting to play the role of omniscient central planners, were swamped," the world-designers reported. When the issue of entertainment and recreation facilities--activities that would presumably keep Habitat users coming back for more--arose, the designers explicitly rejected the game structure used by the first MUDs, with experience points and hierarchical grades and wizard powers:

The idea behind our world was precisely that it did not come with a fixed set of objectives for its inhabitants, but rather provided a broad palette of possible activities from which the players could choose, driven by their own internal inclinations. It was our intent to provide a variety of possible experiences, ranging from events with established rules and goals (a treasure hunt, for example) to activities propelled by the players' personal motivations (starting a business, running the newspaper) to completely free-form, purely existential activities (hanging out with friends and conversing).

After putting weeks into designing and building highly structured treasure hunts for the first Habitat users, Morningstar and Farmer began to understand that the users themselves would use the tools they provided to create activities unplanned or unsuspected by the system's original designers:

Again and again we found that activities based on often unconscious assumptions about player behavior had completely unexpected outcomes (when they were not simply outright failures). It was clear that we were not in control. The more people we involved in something, the less in control we were. We could influence things, we could set up interesting situations, we could provide opportunities for things to happen, but we could not dictate the outcome. . . . Propelled by these experiences, we shifted into a style of operations in which we let the players themselves drive the direction of the design. This proved far more effective. Instead of trying to push the community in the direction we thought it should go, an exercise rather like herding mice, we tried to observe what people were doing and aid them in it. We became facilitators as much as we were designers and implementors. This often meant adding new features and new regions to the system at a frantic pace, but almost all of what we added was used and appreciated since it was well matched to people's needs and desires.

Avatars can die in Habitat, and weapons are allowed in certain areas because the designers "felt that players should be able to materially affect each other in ways that went beyond simply talking, ways that required real moral choices to be made by the participants. We recognized the age old story-teller's dictum that conflict is the essence of drama." In Habitat, death means that the Avatar is teleported to his or her home region, head in hands, divested of all possessions carried at the time of death. In many MUDs, death is equally impermanent. Morningstar and Farmer echoed MUD1 designer Richard Bartle's observations about the way people took even imaginary online death seriously: "Nevertheless, the death metaphor had a profound effect on people's perceptions."

Because the designers imposed very few rules on Avatar social behavior, Habitat players were forced into social debates. One core issue regarded the ontological status of Avatars. Is an Avatar a personal projection of the player, entitled to be treated respectfully? Or is an Avatar no more real than a video game character? Polls of the population revealed it to be about equally split on the issue. The arguments were brought into focus when a group of players, out to make a point, started using their freedom to shoot many other Avatars. Two interesting results emerged from the great debate about gunplay that ensued. First, the players decided to outlaw guns in cities but leave them legal outside city limits. That decision led to online elections for a sheriff, because you are obliged to enforce a law once you make one. The other intriguing result was the founding of the Order of the Holy Walnut, the first Habitat church. The founder, a Greek Orthodox priest in real life, required his disciples not to carry weapons or steal. "His church became quite popular," Morningstar and Farmer noted, "and he became a very highly respected member of the Habitat community."

The designers built the elements of an economy into Habitat, which is even more interesting given that they are now employed by American Information Exchange (Amix), a company that is creating an online knowledge marketplace where contractors and clients can exchange information, graphics, and software for real money. In Habitat, there were Tokens, which your Avatar could earn by spending time in Habitat to supplement the 2,000 it was given when it was "hatched." You could buy a new head for your Avatar, or various objects that are useful in Habitat, from automatic online vending machines. An interesting situation occurred one night when some users learned that a Vendroid in one region was inadvertently selling Dolls for 75 Tokens, and a Pawn Machine in another region was redeeming Dolls for 100 Tokens. Worse, the same Vendroid was selling Crystal Balls for 18,000 Tokens and the Pawn Machine was redeeming them for 30,000 Tokens.

When the designers were offline, a couple of users invested all their money in Dolls and shuttled between the machines until they could afford to trade in Crystal Balls. The next morning, when Morningstar and Farmer logged on, their monitoring programs told them that the money supply in the Habitat database had quintupled overnight! When they tracked down the entrepreneurs and fixed the bug, they found that the users didn't want to give the money back. The powers that be decided not to force the issue, and conviviality was not only restored but boosted when the newly rich players then used their amassed Tokens to underwrite a series of games for the other players on the system.

By 1991, when Morningstar and Farmer reported on their experience, the North American version of Habitat, called Club Caribe, had been running on QuantumLink for almost two years, and sustained a population of some fifteen thousand users. By 1993, QuantumLink had become America On-Line, a Macintosh-oriented commercial online service, and the number of active C-64 users had dwindled. In 1993, Farmer reported to me via e-mail that "Club Caribe is still up and running on QuantumLink, the ever-shrinking Commodore-64 corner of America On-line. CC is completely managed by the customers themselves. There are only a few thousand QuantumLink customers remaining, and some hundreds of them remain active in Club Caribe."

The experiences of Habitat's designers should be required reading for anyone involved in the design or management of commercial online systems. One of the earliest companies to heed Morningstar and Farmer's advice was Fujitsu, which bought Habitat from Lucasfilm Games and redesigned it to use in Japan. Instead of the computationally antique C-64s, Fujitsu adapted Habitat for its new personal media hardware, the FM Towns, a computer that includes a CD-ROM drive--a mechanism for storing and retrieving large amounts of sound and/or images. This machine is a prototype for the multimedia desktop devices that Fujitsu and other companies are planning for the world market in the late 1990s.

The complex graphics capabilities necessary to make any on-screen depiction look more realistic than a crude cartoon figure was far beyond the capabilities of the C-64 and other early personal computers. But the graphics and information-shuffling capabilities of next-generation personal computers--the much-ballyhooed multimedia revolution that combines sound, text, graphics, and video on one affordable unit--have taken an unusually long time to get off the ground. In many ways, like the very first personal computers, the multimedia machines that Fujitsu, Apple, and others are preparing for market are a solution looking for problems to solve. The idea of connecting multimedia capabilities with an online audience, building on the pioneering work of Habitat, made sense to Fujitsu's managers. The push to ultra-high-speed fiber-optic communication in Japan by NTT, Fujitsu, and others, and the continuing evolution of personal computers, was creating a need for real applications that could demonstrate to people who are already information- and technology-saturated that they need to use all the new capabilities available to them. On the scale that these companies operate, buying Habitat, converting it for use on a new computer, and setting up a new online service was a cost-effective research-and-development strategy.

In March 1992, I visited Fujitsu's research-and-development center in Kawasaki, outside Tokyo. Their researchers asked me questions about the online world as I saw it, and they showed me Fujitsu Habitat as they had implemented it. Fujitsu Habitat started operation in February 1990, using the Fujitsu-owned NIFTY-serve network to connect FM-Towns owners with modems to the central computer (the software for Habitat has since been translated to other popular Japanese computers). NIFTY-serve is a commercial service that provides e-mail, chat, and other online services to about 360,000 Japanese users. By April 1992, the virtual community of Populopolis--the official name for Fujitsu Habitat's society--had a total population of 6,200. By the standards of mass media, those figures aren't large enough to be visible. By virtual community standards, Populopolis took two years to reach a population that required seven years for the WELL.

I believe that Fujitsu managers are on their way to discovering, as did Morningstar and Farmer, that the Habitat users themselves, given sufficient tools and freedom, will create their own culture. The multimedia machines are still a small minority of personal computer sales, even in Japan. According to e-mail from Farmer in 1993, "Fujitsu Habitat is still growing daily, with over 7,000 customers currently registered. About 10% of these are `highly active' customers, making up the core of the community. The Macintosh frontend software shipped in March, 1993 in Japan, and, as with the previously released FMTowns, FMR, and PC-9801 compatible versions, the new version should cause a sharp increase in new customer registration. Fujitsu is currently considering an American version."

Some of the most interesting questions had to do with whether or not Habitat showed any potential for a more sophisticated, more finely tuned, more technically powerful version in the future. User behavior is always important in the Japanese research-and-development cycle. Early prototypes of many kinds of different consumer devices traditionally are made available to Japanese consumers, and only those devices that succeed well at home are considered for the world market. Prototypes and proof of concept are another important fixture in the Japanese consumer electronics research-and-development culture. The seven-thousand-and-growing population of Populopolis is a commercial and cultural laboratory for experimenting with cyberspace design--Japanese style. The highly visual nature of Japanese communications is sometimes mentioned as one reason why the text-only nature of the Net is not well suited to the wider Japanese population. Given the tools and the freedom to use them, the citizens of Populopolis have a crack at designing a more visually oriented online culture.

Atsuya Yoshida and Jun Kakuta from the Kyoto Institute of Technology, supported by Fujitsu, formally investigated Habitat "users' social and behavioral characteristics in an on-line virtual world." They chose to investigate the social structure of the system by selecting a novice user of e-mail and chat in Japanese University NETwork (JUNET), a male graduate student of information technology, twenty-four years old, and introducing him to Fujitsu Habitat. For the eighty-one days they observed their experimental subject, "Mr. T.," Yoshida and Kakuta reported that

Mr. T. showed magnificent social adaptation to the visual network society of Populopolis. From March 16, 1993, when the experimental observation was started, the daily access time increased rapidly. We would call this phenomenon `addiction.' Mr. T. smoothly got into the network group and became active to join an on-line election of a president of the town-block association and an on-line discussion about Osaka dialect, and also joined an off-line meeting at an early stage.

Yoshida and Kakuta also observed that Mr. T. and other Habitat users "showed the tendency to use the language specific to themselves, such as special sign language and terms which could not be understandable for the person without knowing the networks and Populopolis." Part of this special sign language was the use of emoticons, or typographic conventions to denote smiling or frowning. The Net version of the smiley face consists of a colon, a dash, and a close-parenthesis symbol that looks like a face if you turn your head sideways: :-) but the Fujitsu Habitat version looks more like an abstract facial representation: (\k \k ). Analysis of conversations in Populopolis revealed that 45 percent of the total number of utterances contained sign language.

Nonverbal context is particularly important in Japan, where gesture, posture, and tone of voice convey overtones and references that are clearly understood within a formal and well-understood set of contextual references. Accents and dialects are particularly important--one phenomenon uncovered by the researchers was that Japanese users who are fluent in the Osaka dialect are underrepresented on Habitat in comparison to their numbers on NIFTY-serve; something about the Habitat online culture inhibits the use of Osaka dialect expressions--itself a matter of some online discussion. This invisible and perhaps partially ineffable component of Japanese spoken communication is sometimes called kansei, which can be only loosely translated as an intuitive, partially aesthetic, sense of rightness about the contextual elements in a conversation.

Yoshida and Kakuta emphasize tha