Conversational Structure and Personality Correlates of Electronic Communication

Jill Serpentelli

Haverford College



Abstract
	Differences in communicative structure was studied on three electronic
communication systems, two interactive (LambdaMOO and Internet-Relay Chat) and
one non-interactive (VaxNotes, an electronic bulletin board), in two different
subsettings (on each system, a topic focused on computing and a more general
topic) by coding transcripts of conversations according to type of utterance.
Significant results were found that the LambdaMOO setting had more greeting
statements, statements indicating interaction with system code, statements
relating to computing, and a trend toward more affectionate statements.  Notes,
however, contained the most biographical information statements, more statments
coded "other", and a trend toward more humorous statements.  The experimenters
concluded that this study provides some basic data that can delineate the
differences between different systems of electronic communication, and that can
be generalized to speculate on some personality correlates of people who use
these systems.  However, the researchers also concluded that much more work
needs to be done on both communicative and personality aspects of electronic
communication.

Conversational Structure and Personality Correlates of Electronic Communication

	Computers from their advent have led to the creation of a subculture
set apart by the passion for a certain form of computer activity or
interaction.  This subculture has existed since the arrival of computing into
today, from the earliest people involved in the most technological aspects of
creating and programming the first computers, to those who devote careers and
hobbies to programming, to those who rely on electronic mail for conducting
business and maintaining friendships, even to those who devote time and effort
in mastering computer games.  Any serious computer-based interest affects the
way a person thinks about and interacts with both the machine and the world at
large.  With the emergence of increasingly complex ways in which to interact
both with and through the computer, computer-based social groups and
interaction are becoming a rich medium in which to study the various
personality correlates of those who choose to take part in such groups.

In the Beginning:  The Hackers

	One of the first such groups formed almost simultaneously with the
integration of the computer into our language and society.  From the beginning
of computers to the present, the "hacker" has been the subject of many an
anecdote in many writings.  Hackers then and now are people who, though often
defying the stereotypes that have been afforded to them, "like nothing better
than to fiddle with computers."  (Amy Goldschlager, personal communication) The
hacker develops a certain way of living and thinking which allows him to devote
much of his time to programming and ever improving his prowess at computing.
Despite the many descriptions of the hacker "type," there has been little
actual research studying the phenomenon of this culture or the psychological
makeup of its members.  However, the anecdotal literature is rich with
description.  Jennings (1990) in her writings notes a description of the
college hacker coined by Joseph Weizenbaum, which summarizes well the "hacker
persona":

    "Bright young men of disheveled appearance, often with sunken glowing eyes,
    can be seen sitting at computer consoles, their arms tensed and waiting to
    fire...When not so transfixed, they often sit at tables strewn with
    computer printouts over which they pore like possessed students of a
    cabalistic text.

    They work until they drop, twenty, thirty hours at a time.  Their food, if
    they arrange it, is brought to them: coffee, Cokes, sandwiches.  If
    possible, they sleep on cots near the computer.  But only for a few hours -
    then back to the console or the printouts.  Their rumpled clothes, their
    unwashed and unshaven faces, and their uncombed hair all testify that they
    are oblivious to their bodies and to the world in which they move.  They
    exist, at least when so engaged, only through and for the computers.  These
    are computer bums, compulsive programmers.  They are an international
    phenomenon (p. 74)."

To complement this intense, solitary image of the hacker, Jennings (1990) in
her writings also describes the travails of the hacker culture:

    "The college hacker rises like a vampire when the sun goes down in order to
    invade the computer room during off-hours, between 10:00 P.M. and 5:00 A.M,
    when computer time is cheaper and the computer works faster because fewer
    people are using it.  His alertness peaks during a cusp of the night when
    most people are deeply asleep.  The electronic alchemy makes time fall
    away...both his diet and his sleeping patterns go to hell.  So does his
    body.  He's sucked into a multihour computer confrontation that hackers
    call sportdeath, as he pushes his body to the limits in his mind duel with
    the machine (p. 71)."

The world of the hacker involves a near-obsessional relationship with the world
of computer programming, which lends a unique lifestyle and identity to those
who pursue these interests.  Jennings describes a world characterized by
solitude, discipline, and control, attributes which are most often ingrained,
to a greater or lesser extent, in the personalities of those who affiliate
themselves with the hacker culture.

	Turkle (1984), in her writings about the expression of personality
attributes through computers, describes more of the personality aspects that
form and are formed by an association with this kind of computer culture.
Issues of separation and exclusiveness figure prominently in her analysis,
especially in her descriptions of conversations with students involved in the
hacker culture of MIT.  She describes the two different life pathways these
students can take:

    "One path leads to what many MIT students call the 'real world.'...Those
    who take the second path flaunt their rejection of 'normal' society by
    declaring, 'We are the ugly men.  You can keep your hypocrisy, your
    superficial values, your empty sense of achievement.  We have something
    better and purer.' (p.198)"

The hacker culture seems in some senses to comprise "the elite of the outcast,"
a group of people who accept their dissimilarity with much of their social
group and reclaim it as a badge of honor.  Turkle suggests that the roots of
this elitism not only rests in the computer culture itself, but in the
backgrounds of those who are drawn to the computer culture.  She cites the
experience of another MIT student:

    "Most of these young men grew up as loners.  Many of them describe a sense,
    as long as they can remember, of a difference between themselves and other
    people.  Finally, they feel that they belong.  Alex is very clear about
    this: "I always knew I was weird.  I mean I didn't know why I was weird,
    but you could see from how other kids treated me that I must have had a big
    sign on me saying: 'Weird One - Fold, Bend, Spindle, and Mutilate this
    One.' (pp. 212-213)"

The hacker culture, thus, was and is in many ways defined as a group set apart
from "them"; both set apart from a group that can not understand their
single-minded focus on computers and pushed into that single-minded focus by
being "different" from their peers even before the advent of their interest in
computers.

	Turkle suggests that the hacker personality is formed partially by the
needs for control, safety, and perfection which are very prevalent for many
young people, especially for the young men who make up much of the hacker
population.  She explains the obsessional nature of hacking in these terms:

    People are not "addicted" to test piloting or race-car driving or computer
    programming.  They are addicted to playing with the issue of control.  And
    playing with it means constantly walking that narrow line between having it
    and losing it.  (Turkle, 210)

Despite this narrow boundary within the hacker culture of having control and
being out of control, maintenance of that boundary can provide the hacker with
a purity of control not found in other cultures.  The computer, despite its
inscrutability and often frustrating linguistic and logical constructions, can
be mastered and understood, unlike social interaction which is a maze of often
unpredictable reactions.  This aspect of computing, as well as the atmosphere
of the hacker culture itself, can make computing a safe haven for the introvert
and the perfectionist.  Turkle elaborates:

    It (the hacker culture) is a culture of people who leave each other a great
    deal of psychological space.  It is a culture of people who have grown up
    thinking of themselves as different, apart, and who have a commitment to
    what one hacker described as "an ethic of total toleration for anything
    that in the real world would be considered strange."  Dress, personal
    appearance, personal hygiene, when you sleep and when you wake, what you
    eat, where you live, whom you frequent - there are no rules.  But there is
    company. (Turkle, 213)

Thus, control, safety, and acceptance become intrinsic elements of the hacker
culture, influencing both the type of members the culture attracts as well as
the way it shapes its members once assimilated into the culture.

Gender Differences in Computer Use

	The movement of the computer culture from the very elite realm of
computer scientists and the hacker community to the more public, easily
accessible realm, made possible by the greater availability of personal
computers and the general progression toward a more computerized society, made
even more apparent the divisions between people who became assimilated into the
computer culture and those who stood outside it.  The increased visibility of
computing in society also increased the non-visibility of those who seemed to
be absent from this new phenomenon.  One of the most apparent and
widely-researched differences in computer use was the gender split in computer
interest.  Males have had, and still have, a greater interest in computers and
use them more.  This is a phenomenon that started in the hacker community,
according to Turkle:

    There are few women hackers.  This is a male world.  Though hackers would
    deny that theirs is a macho culture, the preoccupation with winning and of
    subjecting oneself to increasingly violent tests makes their world
    peculiarly male in spirit, unfriendly to women (Turkle, 210).

This view of the computer culture as technological, competitive, and generally
unfriendly to women who have been largely viewed as more oriented towards the
humanities and cooperation rather than competition, as well as this view's
actual effect on women, has been borne out in empirical study as well as
anecdotal literature.

	Many studies have shown that women do tend to avoid the areas of
computing which are the most technologically and programming oriented.  Temple
and Lips (1989) found that although women in their sample used computers as
much as the men, they took less formal computer science courses and did not as
often choose to major in the computer sciences.  The researchers note that "the
difference between women and men in this sample is not that women are avoiding
computers more than men are; rather, it is that women, more than men, are
avoiding the formal pursuit of specialized training and careers in computer
science."  (pp. 222-223) This study further suggests that this difference does
not constitute a lack of interest on the part of women, but rather a
male-dominated atmosphere in computer science that puts women at a disadvantage
when trying to enter the field.  The researchers assert that according to their
findings, women are not disinterested but are, rather, "scared off by
uncertainty about their own abilities - and uncertainty that is apparently
reinforced by, among other things, the attitudes of their male peers (p. 223).
These attitudes appear to create a vicious circle for women, in which the
computer culture started out male-dominated due to the greater percentage of
men in science and technology in general, and remains male-dominated through
negative attitudes toward women even though women's own conception of their
interests and abilities now more allows them to pursue an interest in science
and math related fields.

	Though studies constantly reiterate the finding that boys and men are
more confident and involved with computing in general, it has also been
consistently found that these differences lie somewhere else than lack of
innate ability on the part of girls and women.  Arndt, Clevinger, and Meiskey
(1985) found no correlation between gender and scores on a questionnaire
measuring attitudes toward computers.  Rather, one of the crucial factors
affecting attitude toward computers was amount of prior computer experience.
In the same way, Levin and Gordon's (1989) study showed a much greater effect
of prior computer experience on positive attitudes than of gender.  These
researchers also found that boys have much more extracurricular exposure to
computers, and show more positive attitudes toward computers as an important
part of their lives and future, two findings that mirror the general overall
greater effect of prior experience on attitudes toward computer use.

	Studies such as these seem to indicate several factors that contribute
to women's lesser involvement with computers and the computer culture.  First,
the programming aspect of computers put women at an early disadvantage, for
women are discouraged from participating in math/science related activities in
general.  The idea that computer programming is a "language", a field in which
women are thought to be proficient, seems to be downplayed in favor of
emphasizing computers as a "science", a male domain.  Second, the competitive
and solitary nature of a programmer's dedication to the computer world -
exemplified in the hacker culture but true to some extent of many
computer-related activities, majors, and careers - is at odds with the reality
of many women's experiences in relating to the world in a more connected,
cooperation-oriented fashion, as has been theorized by feminist writers such as
Gilligan (1982).  It appears that this initial exclusion of women from the
computer culture perpetuates itself by discouraging them from joining the field
and tailoring it more to women's ideas and interests.  Thus, the products of
the computer field most often remain those which would appeal to male users
(violent video games are a good example, as noted later), the culture remains
one of competition, and each aspect feeds into one another to further
discourage women from crossing this gender gap.

	However, other research indicates that this gap can be narrowed
somewhat in order to allow women more access to computers.  Though the computer
culture may be discouraging women entry, some studies indicate that the
stereotype of women as less capable are less ingrained than might be expected.
A 1988 study done by Siann, Durndell, Macleod, and Glissov at the University of
Edinburgh compares the reactions of subjects to stories about one of two
computer scientists, "Kevin" or "Karen", and rate the scientist on a series of
personality attributes.  The researchers found that contrary to their
hypothesis that a female computer scientist would be stereotyped negatively,
subjects rated her just as positively and even more positively on many
attributes.  As the researchers note, their results suggest that negative
stereotypes of women are not the largest contributing factor to the gender gap
in computer science.  This suggests that the gender gap is more often caused by
the way computers are experienced and used by men and women than by outright
prejudice on the part of the computer culture against women's participation in
the field.  The implication is that if computers are made more accessible to
women's experience, if the computer culture became less solitary and more
affiliative, women would find easier access to computing in general.

Video Games - Computing for the Masses

	One of the first widespread computer phenomena, which led to a culture
of sorts, was the release and popularity of video games, both in arcades and in
the home.  Unlike the mostly underground and highly technological hacking
culture, the culture inspired by this phenomenon prompted more attention, media
and otherwise, and more research on the psychological effects of these
miniature virtual realities.  Studies seemed to reveal the same sorts of
profiles that were revealed in other aspects of computer use, along such lines
as gender, introversion, and nonconformity.  McClure and Mears, in a 1984
study, found that the main subculture of videogaming was made up of people they
described as young, male, bright, and competitive, much like the group that
seemed to be attracted to computers at large (McClure & Mears, 1984).  Melancon
and Thompson, in addition, found that arcade-based samples of both genders,
though they had sex-role preferences that were not as clearly defined as the
control population, tended toward more masculine and less feminine sex-role
preferences.  These researchers also noted that cognitive variables are of
limited use in researching computer game play, and that personality variables
should be considered when researching such phenomena (Melancon & Thompson,
1985).  This focus was considered in McClure and Mears' 1986 research, in which
they researched videogames and their effects on psychopathology.  Despite the
many media warnings of the negative effects of video games on young people,
these researchers found no significant correlation between a high rate of play
and conduct disorders or neurotic pathology.  The researchers concluded that
"video game playing is chiefly for casual enjoyment and not a reflection of
anti-social trends, thrill-seeking, or neurotic manifestations."  (McClure &
Mears, 1986)

The Electronic Bulletin Board

	Another widespread, communicative aspect of computing began to develop
soon after the video game culture became entrenched.  The use of electronic
mail systems and electronic bulletin boards became a major phenomenon across
the country, especially in academic settings.  The advent of the Internet, a
system linking the individual computer systems of colleges, businesses, and
government organizations across the country and world, expanded the use of
electronic mail, allowing people to send messages and ideas far beyond the
realm of those connected to one specific computer system.

	Much research on the effects of electronic mail and computer
conferencing on social interaction has been undertaken by Kiesler (1984) and
her colleagues, focusing on academic settings but also generalizing to office
and other non-academic based computer communication.  Her findings, especially
in academic settings, have fully supported the notion that computer
communication changes, often radically, the way people think and interact.  In
one 1984 study, she and her colleagues define the groups of college students
who heavily use computer interaction as an "alien culture" sharing several
common traits.  Since computers are used for so many different uses, note the
researchers, this culture attracts people with a wide variety of interests and
fields.  Further, people who use electronic communication are afforded
relatively direct access to the computer with personal computer accounts for
mail and other communicative activities.  Finally, the culture tends to be made
up mainly of students who are as a group young, smart, have few
responsibilities other than academia and thus have a flexible time schedule,
and have the stamina to stay up late into the night.  Kiesler et al stress the
importance of looking at computing as a culture as well as a tool, implying
that the nature of computing attracts a group of people with certain common
attributes and goals.

	Other researchers have studied and noticed this social aspect of
electronic communication among its users.  Hellerstein, in her 1985 study of a
group of users utilizing a bulletin board at the University of Massachusets,
notes that the users do form a subculture of sorts, in which users greet each
other verbally by username, can be found on the computer system all hours of
the day and night, and never tire of debating issues with one another.  She
further notes that such a bulletin board, as well as electronic mail, can
become a springboard for off-line relationships, both friendly and romantic.
Though these relationships are in themselves fairly normal to college
experience, Hellerstein observes, the members of this subculture are separated
from the college mainstream by their common experience with the computer.  She
notes that this culture can be pervasive in its members' lives to an extreme
sense, to the point where members shirk real-world responsibility in exchange
for spending time on the computer system.

	In analyzing the group of subjects Hellerstein termed "heavy users",
those who used mail and the bulletin board several times a week, she found that
heavy users use mail much more purely for social purposes, to initiate and
continue off-line friendships, and spend less time in phone or face-to-face
communication.  Heavy users also overall reported feeling too dependent on the
system, yet preferred to use it as a means of meeting others, and considered
the membership in the computer-based subculture a positive and freely-chosen
aspect of their involvement with electronic communication.  However,
Hellerstein also noted a common theme in her interviews with the heavy user
group.  Some subjects argued that electronic communication, because of its
attractiveness, ease, and unique capabilities, makes its users very prone to
addiction, or to being controlled by the medium instead of controlling it.

	Several common themes emerge from her findings.  First, the computer is
not, especially for this subculture, purely an anti-social medium.  Rather, it
becomes a new medium, and for many a better medium, of achieving new social
relationships.  Second, it appears that computer users see themselves as being
part of something set apart, something unique and special.  However, they also
acknowledge that this involvement can be overly seductive in its appeal, and
thus become problematic when it becomes intrusive on real-world concerns.

	Hellerstein concludes that "the idea that the computer mediates and
facilitates an individual's social life leads to many interesting
questions...are the rules for communicating over a computer different from
those of face-to-face communication or telephone-mediated communication?  When
people communicate over a computer is some quality missing, or does the
computer actually add something to the communication process?" (p. 196).  Other
researchers have attempted to answer this question by studying the nature of
electronic communication, as well as the nature of the type of communication
the environment affords.  Mihalo (1985) observes that one of the most salient
aspects of electronic communication, bulletin boards in his study, is the
relative anonymity of those communicating with one another:

    "In a face-to-face environment with a stranger, various social conditions
    inhibit communication.  Such characteristics as class, race, sex, age and
    dress can have a dramatic impact on the length and quality of a
    face-to-face interaction between strangers.  In a computerized bulletin
    board on the other hand, these barriers are absent.  Consequently, one must
    interact solely on the basis of what is written.  Without these barriers
    for interaction, there is a potential for developing more intimate
    relationships (p. 201)."

Mihalo argues that computer-based relationships, though they might seem limited
by distance, time, and anonymity, can be very stable relationships due to the
equalizing effect of the computer-based medium.  In this viewpoint, the
computer is not the anti-social or competitive medium of the hacker culture,
but rather a medium that can add a new dimension to human relationships.
Mihalo asserts:

    "One cannot predict that such relationships will emerge as frequent
    complements to other kinds of interactions, such as those in face-to-face
    encounters, but if they do, they will temper the bleak image, painted by
    futurists, of a completely impersonal society brought about by the computer
    (p. 205)."

Thus, the computer can be seen as less of a danger to human interaction than as
a compliment to this interaction.

	With this idea in mind, the most obvious question is how, concretely,
will computer-based communication affect communication?  There has been little
empirical study of this area, but a 1984 study by Kiesler et al provides some
striking insights.  In their study of electronic communication by both bulletin
board-like and more interactive means versus face-to-face communication, the
researchers found that the computer-mediated groups took longer to reach
consensus on issues, participated more equally in conversations, and were less
inhibited in behaviors like swearing and hostility toward one another.  These
findings occurred both with adult and non-student users and with undergraduate
student users.  Thus, it appears that this type of less-inhibited activity is
not just a function of the relatively young population that most often uses
electronic communication means.

The Multiple User Dimension

	In addition to more established mediums of computer communication such
as electronic mail and electronic bulletin board, one of the newer forms of
computer-based interaction is the advent of MUD systems.  A MUD, according to a
set of 1991 articles distributed through e-mail by Jennifer Smith, stands for
Multiple User Dimension, Multiple User Dungeon, or Multiple User Dialogue.
MUDs, originally created by Lars Penski (Smith, 1991), are text-based virtual
realities in which a created character can interact with its environment and
other players in such a way that it can have conversations, handle and create
objects and places, engage in combat, or any number of other options.  These
systems are run on servers throughout the country and world.  Users employ the
use of Telnet, a system that links Internet sites so that one site can "call"
another and log into a system long distance, to call these MUD systems.  New
players are usually allowed to log on as a guest character to explore the
system, and later are allowed to create their own characters and character
descriptions.  The Smith articles describe the MUD process after the player
connects to the system: "Each user takes control of a computerized
persona/avatar/ incarnation/character.  You can walk around, chat with other
characters, explore dangerous monster-infested areas, solve puzzles, and even
create your very own rooms, descriptions, and items." (Smith, 1991)

	The extent to which the user can employ these items and the way in
which s/he can use them is greatly determined by the type of MUD system that
s/he explores.  Smith further explains that TinyMUDs tend to be social MUDs, on
which players mainly meet to talk, joke, and have discussions (Smith, 1991).
However, the article goes on to note that LPMUDs (after Lars Penski the creator
of LPMUD), AberMUDs, and DikuMUDs (both named after universities in Sweden
where they were created are usually more oriented to the role-playing genre
(Smith, 1991) .  This is because TinyMUD and derivitaves of this programming
code is much more suited to room-building and description, while LPMUDs,
AberMUDs, and DikuMUDs are programmed such that the system can accommodate the
combat and character statistics needed to moderate an role-playing atmosphere.
These MUDs are very much like computerized Dungeons-and-Dragons or similar type
games, in which a character engages in combat, finds possessions, and gains
points in order to advance in the game's experiential hierarchy (Smith, 1991).

	No matter what kind of MUD system is being used, a user will have to
engage in some social interaction.  Despite the fantasy-based nature of the
environment itself, the social interactions that take place within it can be
strikingly realistic.  As Smith states, "The jury is still out on whether
MUDding is "just a game" or "an extension of real life with gamelike
qualities", but either way, treat it with care...certainly the hack-'n-slash
stuff is only a game, but the social aspects may well be less so (Smith)."  The
author's own personal experiences with MUD systems have confirmed this
analysis.  Social interaction on MOO seems far from taking on a fantasy-like
quality; though the environment may be somewhat fantastical, the conversations
and actions that take place are surprisingly sincere.  Players build
friendships, debate issues, and even have romances within the context of these
virtual realities, giving many of the more socially-based MUDs the feeling of
an actual community of people interacting in much the same way a real-life
group would interact.

Role-playing and Personality Correlates

	The atmosphere of MUDs, especially LPMuds but also to some extent more
socially-oriented MUDs, provides an interesting analogy to an older but no less
complex phenomenon of role-playing games (RPG's), such as the most famous
example, Dungeons and Dragons.  Role-playing games are social, rule-based
"adventures" in which players create characters with certain powers and
attributes, which in turn explore a fantasy realm created by the dungeon (or
game) master within the sphere of the game's rules.  Though role-playing games
are fantasy-based, like MUDs they can take very realistic social and
psychological attributes.  Very little actual study has been done on this
particular topic, but the anecdotal literature is revealing.

	In his essay on "Rational Coordination in the Dungeon," Terry
Toles-Patkin (1986) begins by explaining why any game necessarily takes on more
"real-world" connotations:

    "Games serve as extensions of social man, clarifying cultural forms that
    have become so familiar that their meaning is lost or obscured as we
    conduct the routine activities of every day life.  Play is unique in that
    it stands apart from ordinary life by virtue of its being "not serious"
    (i.e, noninstrumental in nature) but at the same time absorbing the player
    utterly and intensely.  No material interest or profit accompanies play,
    which proceeds within its own boundaries of time and space according to
    rules fixed in advance.  Play is a fragile activity; at any time reality
    may rudely reassert its rights either from outside the game context
    (through some interruption) or from within (by means of an offense against
    the rules or a collapse of the play spirit) (p. 1)".

Thus, Toles-Patkin suggests that any type of game is culturally, and one could
infer psychologically, revealing, because it engrosses the player fully in a
very intense environment of social rules.  Thus, games become both an
atmosphere of freedom from the details of everyday life and an opportunity to
look at aspects of oneself and culture in a purer and more secure context than
everyday life.

	This aspect of play is especially salient in an RPG context, according
to Toles-Patkin.  He notes the progression in gaming from a very rule-based
approach to the game to a more nuanced approach:

    "New campaigns composed of inexperienced players tend to mechanically kill
    monsters in what has been referred to as the "hack and slash" school of D&D
    playing, but when they become experienced in the basics of the game,
    communications among the players and the dungeon master take on new
    importance (p. 7)."

This progression is the crucial aspect of D&D that leads to its ability to be a
domain in which players can express and reconcile a variety of perspectives,
both social and psychological.  Toles-Patkin points out that the way in which
the games structure themselves, both in rules and plots, can be socially
revealing, while the dialogue and actions that take place within those
structures can be psychologically revealing:

    "...process and content serve very different functions: process symbolizes
    social structure while content symbolizes psychological functioning (p.
    8)."

Thus, Toles-Patkin implies that though the social structure of games like D&D
are on the surface "fantastical", the actual process mirrors social realities
that can draw out psychological facets of the players involved.

	In another anecdotal article, John Eric Holmes (1980) supports and
expands this notion of role-playing games as psychologically revealing:

    "The Dungeon Master's world is sort of a giant Rorschach test...Almost
    always, the personalities of the characters turn out to be combinations of
    people's idealized alter egos and their less-than-ideal impulses (p. 84)"

The many anecdotes related by Holmes in his essay focus on his players'
enactment of these "less than ideal impulses, and his interpretations of such
role-playing.  He notes the amount of violence that is most often present in
D&D campaigns, an aspect that has often brought criticism of RPGs, and argues
that RPGs are more likely to be an acceptable outlet for aggressive influences
rather, as critics have claimed, encouragement for real life aggression:

    "The level of violence in this make-believe world runs high.  There is
    hardly a game in which the players do not indulge in murder, arson,
    torture, rape, or highway robbery...

    I don't think this imaginary violence is any more likely to warp the minds
    of the participants than is the endless stream of violence in TV, movies,
    or literature.  Quite possibly it provides a healthy outlet for those
    people who are imaginative and inclined to enjoy the game.

    In order for the game to provide vicarious release for unacceptable
    behavior, the entire group of players must go along with the convention
    that game roles are independent of the actual players.  One teenager, who
    rarely complains, objected with untoward violence when his centaur
    character was robbed and abused by a character of his stepbrother's.  "It's
    the magic-user who did that to you," protested the other lad, "I didn't do
    it, he did.  He's a thoroughly despicable person!"  (p. 92)."

Holmes presents an interesting dichotomy here between the realistic and
fantastical elements of role-playing games.  After first suggesting that the
games can become poignantly realistic in terms of social structure and the
extent to which players use their characters to express aspects of their
personalities and desires, to the extent that the games become violent and
gross violation of real-world social norms, the fantasy-based aspect becomes a
protective sphere that allows safe airing and working out of such impulses.

	This aspect of role-playing, that of using it as a medium to work out
and better understand aspects of one's personality, seems to correlate in
Holmes' view with the fact that the most dedicated of RPGs tend to at least
begin their role-playing from their mid-teens to early adulthood, when they are
defining for themselves the salient aspects of their personalities.  Holmes
notes that in his experience, their characters tend to be, for the most part,
"role models" for some type of behavior they would like to be exhibiting but do
not in the real-world realm:

    "My earlier game companions consisted largely of teenage players, and these
    young people, caught in the awkward adjustment to the adult world, produced
    game characters who were suave, cool, deadly, and superbly adjusted to
    their world - samurai, elven magicians, and clever hobbits.  For these
    characters, there were few problems that could not be quickly solved by
    blowing somebody up with a fireball spell or slashing them to pieces with a
    shining katana (p. 87)."

From this and the many other anecdotes on RPG's that have been cited, it
becomes obvious that something else besides sheer acting and rule-following
occurs in the context of these games.  In the often-confused realm of growing
from childhood to adulthood, young people can find in RPGs an escape from the
stress of everyday complexities which can not be so easily eradicated with a
magic spell, and, more subtly, can also find an outlet for conflicts and
questions about identity, self-control, goal achievement, relationships,
intimacy, and many other aspects of personal growth.  It can be inferred that
similar needs are met by a computer-simulated role-playing environment such as
a MUD, since the MUDs involve the same type of character development and
abilities for self-control as the non-computerized role-playing environment.

Role-playing, Computing, and Fantasy-Proneness

	Despite the capacity of role-playing environments to provide a sphere
for a particularly revealing form of self-expression and growth, not all
people, even those who could greatly benefit from such a medium, are interested
in such environments.  It appears that the fantasy realm attracts a certain
group of people; how these people are similar is a yet unanswered question.
However, the way in which people react to and use such mediums should be highly
contingent on their reaction to and use of imagination and fantasy in general.
Thus, understanding the personality correlates of fantasy use could provide
some insights into the type of person who tends to be interested in
fantasy-based alternate realities, computer-based or not.

	Rhue and Lynn (1987) cite a 1981 study by Wilson and Barber of
personality types they term "fantasy addicts" or "fantasy-prone personalities".
They found several types of childhood experiences that correlated with
fantasy-proneness; among them were encouragement to fantasize from adults,
early creative situations such as piano or dramatics classes, and experiences
of loneliness, isolation, and need to escape some kind of aversive stimulus.  A
somewhat disturbing attribute that is related to the last attribute is Wilson
and Barber's finding that, at least in the context of their study, children who
had suffered some form of abuse, mostly physical or emotional, were more likely
to be fantasy-prone adults.  This finding further raises the question of the
role of fantasy as a coping mechanism.

	In a study of college students, Rhue and Lynn (1987) found that
subjects, for them a group of young adult college students, who tested high on
several tests related theoretically to fantasy-proneness also tended to report
childhood experiences of greater loneliness, a marked enjoyment of imaginary
games, and a good amount of time spent playing alone.  Fantasy-prone subjects
also produced MMPI profiles that were suggestive of highly unusual early
experiences, as well as experiences of conflict and alienation.  Several
subjects in the fantasy-prone group reported that they had been physically
abused as children.  This inconclusive finding suggests not that all
fantasy-prone individuals have been abused, but that abuse or negative
childhood experiences can contribute to an individual's tendency toward fantasy
proneness.  Thus, Rhue and Lynn's findings agree with the earlier Wilson and
Barber conclusions that there seem to be developmental experiences that
correlate with fantasy-proneness in adults.

	Given this finding, the criticisms that are often leveled at
role-playing games - that they can be an unhealthy escape mechanism, that they
can cause their players to become too emotionally invested in the
game/characters, thus putting them at risk for emotional instability - could
seem to be somewhat justified, since it seems that many of the people who would
be naturally attracted to such fantasy realms may have suffered adverse
experiences in the past which did affect or still affect their emotional
health.  However, Rhue and Lynn suggest an alternate viewpoint.  They suggest
that fantasy may be an adaptive function for many people, an escape mechanism
that is used so that the real world will not become unmanageable, a medium used
to ultimately function more effectively in the real world by preserving their
emotional health.  They conclude that fantasy-prones may be unusually strong
people rather than weak people who use fantasy as a "crutch", as critics have
suggested:

    "Fantasy-prone college students may represent a particularly well-adapted
    group of individuals who manifest a deep and extensive history of fantasy
    involvements (p. 135)."

Thus, despite the potential for fantasy to become a pathological escape
mechanism, results such as those of Rhue and Lynn suggest that it can just as
easily be a medium through which one can expand or improve an already rich
reality.

	Along with these studies suggesting that a certain type of background
affects the extent to which people use fantasy mediums, another study suggests
there is another crucial factor affecting whether a person is likely to explore
fantasy realms.  Maddi, Hoover, and Kobasa (1983) found that in a group of
subjects high on the scale of customary activation, or arousal level, their
level of external or internal orientation affected how they used this energy.
Those subjects with an external orientation, when placed in a waiting room
scattered with various interesting objects, would tend to touch and examine
more objects than subjects judged to have a more internal orientation.  The
experimenters concluded that both types of people both were showing an
imaginatively-directed type of behavior, but that those with an external
orientation have more of a motivation toward curiosity, and those with an
internal orientation, toward creativity.  This finding is especially
interesting when related to a MUD or role-playing game type of environment, in
which both aspects, both an internalization of a fantasy realm and actually
exploring objects and places in this realm, are equally salient.

	Fantasy-proneness and creativity appear to also be important aspects
affecting how people interact with mediums of computer communication, given the
other-worldly nature that such mediums can take on.  However, the extent to
which interactive computer mediums can simulate another reality in a
role-playing/fantasy capacity is very dependent on the medium used.  Bulletin
boards are already limited in their capacity to create an environment by the
fact that users do not communicate on these systems in real time.  However,
even in real-time mediums of computer-based communication, the apparent
potential of such systems to generate a "virtual reality" varies greatly.
Unlike MUDs, not all interactive communication systems provide a detailed
"environment" for description and exploration.  An example of such a system is
Internet-Relay Chat, or IRC.

Internet-Relay Chat:  A Gathering of Voices

	IRC is a network of conversational channels that can be accessed by
connecting, through the Internet, to one of many networked servers across the
country and world.  The internal structure of IRC bears some resemblance to a
MUD system; players log in under pseudonyms (though the /whois command makes it
possible for players to easily find out other players actual email addresses,
unlike MUDs which are more anonymous), and conversations take place on channels
that somewhat resemble MUD rooms.  However, unlike MUDs, neither characters nor
channels can have descriptions, and there are no interactive objects.  The
setting is much more like being in a conferenced phone call than actually being
in a room-like environment.  IRC also has many more players at one time than
almost any MUD system; hundreds of players could be logged in altogether at a
busy time of the evening; 25 to 40 could occupy a busy channel.

Social Interactions on IRC

	A user of IRC, once having logged on and joined one of the busier, less
structured channels such as "#hottub", is most likely to find him or herself in
a sea of chaotic statements.  A busy channel may have as many as 25 people all
talking to one another, subjecting the observer to many conversational threads
all converging with one another.  Conversation on the busier channels is almost
impossible beyond a superficial level; it would be impossible to initiate or
keep track of a serious conversation with so many people talking.  The effect
is as if a crowd of 20 to 30 stood in a room and tried to all have a
conversation with one another, instead of breaking off into smaller groups as
would most likely happen normally in a crowd of this size (Appendix A).
Another more realistic comparison is the phenomenon of listening to a channel
on a CB radio, which is made up of many different people all trying to talk to
one another at once (Amy Goldschlager, personal communication).

	On the more structured channels, such as those discussing religion,
computing, or a number of other topics, tend to have less people; perhaps four
to ten a channel.  Thus, conversation on these channels tends to be more
focused and coherent.

	IRC could, in many ways, be seen as a forerunner to MUD-type systems,
since it does share the similar characteristics of character names, rooms, and
real-time communication.  However, its capacities to create an imaginative
"virtual" environment is severely limited by its simple structure.  It provides
little more than a text-based conference call between various users; its
function is only communicative unlike MUDs which can also become an outlet for
character description, room building, role-playing, and active interaction with
the system itself by way of exploring rooms and manipulating objects.

LambdaMOO: An "Object Oriented" MUD

  The History of LambdaMOO

	Quite recently, a new type of MUD has been implemented which allows an
even greater potential for interaction between the user and the environment,
and for greater development of virtual characters.  This new system has been
dubbed MOO, or MUD-Object-Oriented.  It is similar in format and function to
TinyMUDs, but allows a greater range of user activity in building, actually
creating interactive objects, and custom-tailoring characters.  The first and
largest MOO system, until recently the only fully-functioning public MOO, is
the LambdaMOO system, supported by Xerox PARC and run from Palo Alto,
California.

	The authors spoke with the creator and "archwizard" of LambdaMOO, Pavel
Curtis, known as "Haakon" on LambdaMOO.  In a discussion with Curtis which took
place on the MOO system, he described the history of this new type of MUD
system.  Since MUDs in general and MOO especially are such new mediums, his
insights into the history of LambdaMOO as well as its social interactions were
invaluable in better understanding the medium.

	Curtis told us that the MOO project was started in early 1990 by
Stephen F. White, now known as ghond on the MOO.  Curtis explained that ghond
wanted to make this MUD "object oriented, a type of programming that allows a
user to program objects in such a way that one object can be the child of a
generic "parent" object, sharing most or all of the characteristics of that
parent.  (Curtis, personal communication).

	Curtis noted that ghond began testing the new MOO, set up on a server
at Berkely and dubbed AlphaMOO, in April of 1990.  Curtis became active in the
project at this point, as well as several other present "wizards" on LambdaMOO.
He took over the project from ghond (who, according to Curtis, was suffering
from "temporary burnout") in September of 1990.  Curtis explained to us that
AlphaMOO was the first MOO and did have some public access, but was not widely
used due to little advertisement and little documentation (Curtis, personal
communication).

	Curtis went on to explain how the transition occurred from AlphaMOO to
LambdaMOO.  After fixing bugs in the system, rewriting some of the code, adding
more programming capability, and writing documentation, he had created what he
termed "a truly separate entity" from the original AlphaMOO.  He dubbed this
new system LambdaMOO, after one of his names on the system and, according to
Curtis, "because it's a key word in some of the other non-mud research that I
do."  The new system was announced as open for public access on UseNet (a
world-wide bulletin board system) in February of 1991 (Curtis, personal
communication).

	Curtis noted that the original response to the announcement was
"lukewarm", but that the number of players increased slowly but steadily.
According to him, growth over the past few months (the conversation took place
in November of 1991) had been much more rapid, with more than 25 people logged
into MOO at almost any one time (Curtis, personal communication).  The
experimenters have seen this trend continue into May 1992, bringing a constant
stream of new characters, and a list of players that sometimes numbers as much
as 40.  The MOO community, according to Curtis, seems to be made up mainly of
undergraduate students using the MOO as a social and creative outlet (Curtis,
personal communication).

	When questioned by the authors about Xerox's part in supporting the MOO
server, Curtis explained their role:

    Xerox pays me to do more-or-less basic research.  They are not looking for
    products out of what I do.  They trust me to find interesting areas of
    research that might perhaps open up new opportunities either for them or
    for the CS community at large...Xerox has no commercial interest in
    LambdaMOO whatsoever (Curtis, personal communication).

However, Curtis' own paper concerning social interaction on LambdaMOO indicates
that the MOO and similar systems may indeed have other possibilities for use:

    The MUD model is also being extended in new ways for new audiences.  For
    example, I am currently involved in adapting the LambdaMOO server for use
    as an international teleconferencing and image database system for
    astronomers.  Our plans include allowing scientists to give online
    presentation to their colleagues around the world, complete with 'slides'
    and illustrations automatically displayed on the participants'
    workstations...I expect such specialized virtual realities to be
    commonplace, an accepted part of at least the academic community (Curtis,
    16, unpublished paper).

Virtual reality provides a new outlet for communication in a business and
educational realm, providing even more opportunities for nuances and
interaction than bulletin boards and e-mail.

Social Interaction on LambdaMOO

  The "Geography" of LambdaMOO

	Upon connecting to LambdaMOO, a user usually finds him or herself in
"The Coat Closet", a room that serves as home base for any character who has
not yet built his or her own home.  From here, the character (by typing "out")
can "walk" out into The Living Room, one of the major centers on MOO for group
conversation.  By typing direction commands (north, south) the player can
"walk" around the house and explore the many rooms and underground passageways
that are connected to it.  In addition, if a player wants to reach a room that
is faraway from where s/he is "standing", or if s/he wants to reach a room
which is not linked to the main structure of the house, the player can
"teleport" by using a @move me command.

	The effect of these many different rooms presents a mental image of the
house itself, sprawling in size with its many underground tunnels and attic
hideaways, but all interconnected in a realistic fashion.  However, the player
is also presented with the image of many other rooms and areas that form
"castles in the sky" in the MOO realm, areas that can only be reached by
fantastic means such as "teleporting".

  The Conversational Atmosphere of the MOO

	A player in a large group conversation on LambdaMOO will find several
noticeable differences between the atmosphere of MOO rooms and that of IRC
channels.  First, conversations take place in actual "rooms", whose
descriptions often affect a player's mental image of the atmosphere: the Living
Room's description consists of a cosy room with many chairs and an open
fireplace; another room, the "Makeshift Cafe", another room for large social
gatherings, has in its description many outdoor tables at which players can
"sit"; yet another room, Hacker's Heaven, presents the image of a busy computer
workshop filled with computers, printouts, and miscellaneous hardware.  These
different descriptions affect the player's image about what types of
conversation and behavior is appropriate to these settings (Appendix B).

	Second, players can create their characters in a much more detailed
fashion than on a medium like IRC (Appendix C).  Besides just being able to
write a "character description" outlining what a character looks like, what
he/she/it is wearing, and any other information that the player might like to
provide, players can also embellish their characters in a variety of other
ways.  Characters can be programmed to have a variety of different persona,
"morphing" between different names, genders, species, and descriptions.
Characters can also have elaborate "entrance" and "exit" messages to announce
when they have teleported into a room; these can range to arriving in a puff of
smoke, dropping in from the ceiling screaming AAAAIIIGH!, flying in on a magic
carpet, or any variation on such ideas.

	In addition, conversation itself can be richer on MOO than on an
IRC-like medium.  Besides just being able to type sentences and have them
appear on the screen, characters can choose to us the "say" command, which
places "(character name) says," in front of what a player types (BethAnne says,
"hi there!") or the "emote" command to place one's name before an action,
emotional gesture, or thought (Sylvan smiles).  These aspects add more nuances
to MOO conversation since characters can express thoughts and emotions in
greater detail.  Characters can also talk to other characters in a room
privately with the "whisper" command, or talk to characters in another room
with the "page" command.

	In a typical large-group conversation, a user of MOO is bound to find
less chaos than in such a conversation in IRC, partially because the average
amount of people in a large-group conversation on MOO tends to be five to ten
people.  Thus, conversational threads become less fragmented and easier to
follow.  At the same time, a user may have to keep track of many conversations
at once in such a setting, since two different threads might be in progress in
the room itself, while s/he might be whispering or paging with yet another
person in a private conversation.  Thus, though MOO conversation is more
coherent in terms of focus, it still often forces the user to learn how to
divide attention effectively between a few conversations at the same time, a
skill not often practiced in real-life conversation.

	The focus of conversation varies from room to room, though computers
and MUDding are always popular topics, especially in "Hacker's Heaven" which is
a room intended for such discussion.  In addition, characters often interact
with MOO objects in the middle of conversations (sitting on a chair, petting a
MOO dog, and so on).  Humor is also popular, due to the great potential of the
MOO environment for physical and slapstick humor, such as "bonking" people with
various objects, as well as verbal humor such as punning and general surreal
statements (Appendix D).  Finally, the extent to which people can manipulate
their own characters becomes more apparent in conversation; people are
constantly displaying their entrance and exit messages as they enter and leave
a room; players often comment on these messages, on other players'
descriptions, or on what other players might be "carrying" (Appendix E).  Thus,
on MOO there is a much greater sense of an actual conversation being had by
people that a user can "see", in the boundaries of an actual room and the
objects in that room with which the user can interact.  Perhaps because of
this, there is a good amount of discussion about the MOO itself, its different
rooms, its atmosphere, and its characters (Appendix E).

  Makeup of the Population

	The many and detailed social interactions that occur on the MOO provide
rich opportunities to further consider and study past findings on a variety of
topics mentioned earlier, such as fantasy, control, role-playing, gender
issues, and the unique psychological makeup of those who use computer
communication in general.  The LambdaMOO population is in itself revealing.
According to Curtis, over 90% of the population are students at colleges and
universities, and mostly undergraduates.  From what he has seen, he believes
that no more than half of the population is involved somehow in the computing
field.  Rather, he asserts that the increasing availability of the Internet to
more students at many colleges is bringing a much more diverse population to
MOO.  He also theorizes that the MOO population is most often above the norm in
both educational background and economic status.  Finally, he claims that the
MOO community seems to be almost 95 percent male (Curtis, 6, unpublished
paper).

	Several facets of this analysis present relevant topics for an analysis
of the MOO.  First, it is apparent that the participants in LambdaMOO are
people who have affiliated themselves with the culture of computer
communication to some extent, though not necessarily through a life commitment
to the computing field but rather through personal interests.  The LambdaMOO
population thus provides a rich environment for further study of the
personality correlates of computing, studying exactly what makes people of
different interests and disciplines similar enough that they are all attracted
to this particular form of communication.

Gender Differences on LambdaMOO

	The gender discrepancy that Curtis notes is also revealing in that,
even with such a socially-oriented medium as LambdaMOO, the MOO still shows the
same kind of gender bias as other social mediums as bulletin boards.  However,
the author in her personal experience has not found this skew to be as salient,
and wonders if the communicative aspect of a medium like MOO might contribute
to a breaking down of the gender gap by attracting more women.  Confirming the
existence or nonexistence of such a gender skew, and finding reasons for its
presence or absence is a topic for which MOO presents yet another environment
for further study.

	Curtis also notes that many players seem to play the opposite gender on
LambdaMOO, most often males playing females:

    As I've said before, it appears that the vast majority of players are male
    and the vast majority of them choose to present themselves as such.  Some
    males, however, taking advantages of the relative rarity of females in
    MUDs, present themselves as female and thus stand out to some degree.  Some
    use this distinction just for the fun of deceiving others, some of these
    going so far as to try to entice male-presenting characters into
    sexually-explicit discussion and interactions.

    Other males present themselves as female more out of curiosity than as
    simply an attempt at deception; to some degree, they are interested in
    seeing 'how the other half lives,' what it feels like to be perceived as
    female in a community.  From what I can tell, they can be quite successful
    at this (Curtis, 7, unpublished paper).

Curtis' speculations on the reasons for gender-switching on LambdaMOO are all
intriguing and merit further study.  In addition, this author, through
conversations and personal experience on the MOO, has found several other
concerns related to gender-switching on MOO.  In a conversation with another
MOO-using student, it was suggested by this student that people on MOO often
gender-switch because they are dealing with issues of sexuality, perhaps
questioning their own sexuality, and find that switching to the opposite gender
in virtual reality is a "safe" way of flirting with people who are (ostensibly)
the same gender as them in real life (Alis Marks, personal communication).  The
author has had one such conversation on LambdaMOO which supported this theory.
In addition, the author's thesis partner has suggested that gender-switching
can serve another social function, that of learning to communicate better with
the opposite gender.  A male who is shy about approaching women, or vice-versa
(though this author would speculate that it is more the former, judging by her
personal experience of LambdaMOO), might feel more comfortable doing so in the
guise of being "one of them" (Amy Goldschlager, personal communication).

	Curtis' remarks on female MOO characters raise more questions about the
effects of gender roles on MOO interactions:

    Female-presenting characters report a number of problems.  Many of them
    have told me that they are frequently subject both to harassment and to
    special treatment.  One reported seeing two newcomers arrive at the same
    time, one male-presenting and one female-presenting.  The other players in
    the room struck up conversations with the putative female and offered to
    show her around but completely ignored the putative male, who was left to
    his own devices (Curtis, 7, unpublished paper).

This apparent phenomenon of "virtual chivalry" has been confirmed in the
author's own experience as a female character in LambdaMOO; players, especially
male players, tend to be very helpful to female characters, attempting to gain
their friendship and often helping them extensively with learning the mechanics
of the MOO.  The author has discovered too the reality of Curtis' reference to
harassment; she has found that male characters tend to flirt aggressively and
sometimes to the point where it ceases to be enjoyable and resembles sexual
harassment.  The medium seems to free players to engage in this behavior with
more freedom than they would most likely do so in real life, often presenting
problems for the female characters who are left with the burden of dealing with
the unwanted attention which, though virtual, can still be disturbing.  Curtis
notes that these problems often dissuade characters from presenting themselves
as female, which in itself can lead to problems:

    Because of these problems, many players who are female in real life choose
    to present themselves otherwise, choosing either male, neuter, or
    gender-neutral pronouns.  As one might expect, the neuter and
    gender-neutral presenters are still subject to demands that they divulge
    their gender.

    Some players apparently find it difficult to interact with those whose true
    gender has been called into question; since this phenomenon is rarely
    manifest in real life, they have grown dependent on 'knowing where they
    stand', on knowing what gender roles are 'appropriate'.  Some players (and
    not only males) also feel that it is dishonest to present oneself as being
    a different gender than in real life; they report feeling 'mad' and 'used'
    when they discover the deception (Curtis, 7, unpublished paper).

These concerns about gender roles on LambdaMOO create some interesting
paradoxes that warrant study.  One the one hand, LambdaMOO provides a rich
opportunity to question and experiment with gender roles, both because its
structure provides the opportunity to do so, and because it seems to the author
through her own personal experience that the MOO community is made up of many
people who tend to be liberal as well as "quirky" in the sense that they often
do not accept more "mainstream" societal standards, including those of gender
roles.  Thus, LambdaMOO is an ideal situation in which to question the effects
that gender roles have on the way a person presents him/herself as well as the
way people react to her/him.  However, the tendency noted by Curtis of
characters' unease with neutral gender roles, the negative reactions that
characters who present themselves as other than their real-life sex can
provoke, and the often more stereotypical treatment of female-presenting
characters by treating them as more needy of help and sometimes as objects of
exploitation, reveals that MOO can also be an environment for playing out or
trying to maintain more traditional dichotomies of gender.  This suggests as
well that even within this non-mainstream computer subculture, members often
have trouble shedding more traditional assumptions about gender.  The extent to
which this correlates with the ostensibly male-biased population of MUDs is a
question for further study, as well as the more general question of how widely
the members of such a subgroup like MUDders tend to actively or subconsciously
contradict such conceptions.

Role-Playing and LambdaMOO

	The extent to which players descriptions in general, outside of just
gender, are a function of role-playing or are just mirrors of real life is
another fascinating aspect of LambdaMOO.  Curtis notes some of these aspects of
characters' descriptions:

    A large proportion of player descriptions contain a degree of wish
    fulfillment; I cannot count the number of 'mysterious but unmistakably
    powerful' figures I have seen wandering LambdaMOO.  Many players, it seems,
    are taking advantage of the MUD to emulate various attractive characters
    from fiction.

    Given the detail and content of so many player descriptions, one might
    expect to find a significant amount of role-playing, players who adopt a
    coherent character with features distinct from their real-life
    personalities.  Such is rarely the case, however.  Most players appear to
    tire of such an effort quickly and simply interact with the others
    more-or-less straightforwardly, at least to the degree one does in normal
    discourse.  One factor might be that the roles chosen by players are
    usually taken from a particular creative work and are not particularly
    viable as characters outside of the context of that work; in short; the
    roles don't make sense in the context of the MUD (Curtis, 8, unpublished
    paper).

Given the literature cited earlier on role-playing/fantasy games, and the
extent to which players use their RPG characters to express aspects, hidden or
not, of themselves, the extent to which players role-play on LambdaMOO or any
other MUD system is essential to study in an attempt to understand this social
medium.  The author would hypothesize that mediums like LPMuds would encourage
more role-playing than a medium like LambdaMOO, because the former is closer to
a more typical RPG format.  However, role-playing does occur to a greater or
lesser extent in more social MUDs like LambdaMOO, even if just to the extent
that the player has total anonymity and the opportunity to shape his/her
appearance and personality to his/her liking.  Thus, the same issues that
surround RPGs and fantasy in general surround the nature of communication in a
realm such as LambdaMOO.  The ways in which a fantasy-based, fabricated
character can be used to express very real aspects of a person's psychological
makeup, as well as becoming an exemplar of future goals, again become salient
issues in such a virtual reality.

	As the author found in her personal experience with LambdaMOO, virtual
characters can also differ from real-life characters with no conscious effort
on the part of the player, simply because the MOO's emphasis on almost purely
linguistic communication forces a type of communication different from the many
non-verbal nuances apparent in real-life conversation.  Thus, the MOO
environment can often force more bluntness in conversation, forcing a player to
communicate less shyly, or with less sarcasm, or without the verbal defenses
that s/he might use in real life.  This creates the interesting paradox that
the MOO can both force greater communicative honesty (if the player does indeed
want to be honest) while still creating a virtual personality that might differ
from a player's real life presentation, simply due to the radically different
nature of the two communicative media.

Communication, Intimacy, and Control

	Issues of communication on LambdaMOO lead to issues of the MOO's
potential to create intimacy between characters.  Despite the "virtual" nature
of MUDs, the "reality" element cannot be ignored in any social communications.
As noted earlier, MUDding can be seen as less of a game and more of an
extension of real life, since, though people play "characters," there are still
two real people communicating on either side of their respective computer
screens.  This often leads to the formation of very complex social and
emotional webs among MOO players.  Friendship and romance do occur, as well a
level of emotional and often physical intimacy bounded only by its confinement
to a linguistic level.  Though these communications are in one sense limited by
their linguistic nature, they can also encourage intimacy by encouraging people
to speak specifically and honestly.  Intimacy in a virtual realm is a subject
that has been studied little, but seems to be a crucial aspect in understanding
both electronic communication and fantasy role-playing in an electronic realm,
for it incorporates the same issues of control, introversion, isolation,
gender-roles, and other issues very relevant to understanding both mediums.
However, the exact ways in which these issues are expressed are still unclear;
though many are apparent on an intuitive level, this is an area that warrants
further study in order to better understand its dynamics.

Research Possibilities on Computer-Based Communication

	The author(s) have found a medium like LambdaMOO to be an ideal
environment to study the structure of conversation on an interactive electronic
medium, and its possible implications in determining the personality correlates
shared by users of such systems.  We would hypothesize that the programming
aspects of LambdaMOO incorporate and attract those types of people who share
some of that the personality correlates of the most general "computer culture"
that build up around programming and computer science fields.  Secondly, the
role-playing and fantasy aspects attract those who, as well as having some
level of comfort with computing, also share some of the personality correlates
of people who are attracted to fantasy realms in general as both a means of
enjoyment and of expressing more serious psychological issues.  Finally, MUDs
add a new and more nuanced dimension to the area of computer-based
communication, affording us a new realm in which social interaction will take
on unique and psychologically revealing forms.

	We would like to discover what, if any, traits are shared in common by
people who use media like MUDs.  However, we also believe that these
characteristics cannot be determined without understanding exactly how these
media are structured, and how different electronic media might produce
different types of discussion.  In analyzing the different ways in which
different electronic environments affect communication, we can then speculate
on exactly what types of people might be attracted to each of these systems, or
to systems like them in general.  We hypothesize that the issues raised in both
the anecdotal literature and the research we have cited on computing and the
computer culture in general, computing and gender, computer communication,
role-playing, and fantasy will figure strongly in this analysis by providing a
basis for such extrapolations.

The Pilot Study:  Determining MOO Demographics

	To get at least a sense of the types of people that tend to be drawn to
an interactive electronic communicative medium, we picked the most complex of
these media for a pilot study.  What follows is a description of this survey,
which we posted on LambdaMOO, and some of the initial data we have accumulated
from this survey.  The researchers created a joint character on LambdaMOO,
named "Dr.Sherry" after Sherry Turkle, created an "office", left a virtual
survey in the office, and also left notes in several other parts of the MOO
asking people to take the survey.

	The personality-oriented aspects of the survey did not produce
significant or revealing results, confirming the researchers' suspicions that
there are many more aspects of LambdaMOO and MUDS in general, such as
structural and conversational aspects, must be understood before we are able to
compile psychological data on the actual personality types of people that use
MUD systems or other similar electronic settings.  	Dr.Sherry also
expressed a willingness to talk to other characters interactively.  Though
response to this offer has been sporadic, it did provide good insight on and
experience with the MOO, which helped us better interpret MOO conversations and
create content codes.

	The LambdaMOO survey also provided data on the demographics of people
who use the MOO.  Of 55 people who answered the gender-related questions, 73%
are male in real-life and 27% are female.  Most play characters on MOO that are
the same as their real-life gender, though 7.5% of the males report female
characters, and 5% play neuter characters.  13% of the females play neuter
characters; none reported playing male characters.  In response to the question
about ownership of a computer and Internet access, 70.9% of the 55 people who
answered own a computer, 89.7 of these computer owners have modems, and 92.7%
of the total respondents have 24-hour Internet access.  When asked how they
most spend their time on MOO, 21.4% of the 56 people who responded to this
question spend the most time exploring the environment, 16.1% spend the most
time programming, 60.4% spend the most time interacting, and 1.8% could not
decide.  In response to the career/major question, 55.3% of the 57 people who
responded were somehow connected to the computer science field, 7.1% to the
other sciences, 17.9% to the liberal arts, 7.1% to the social sciences, 8.9% to
other disciplines, and 3.6% were as yet undecided.

Implications of the Pilot Study for Research

	Even these preliminary percentages reveal some characteristics of
LambdaMOO and MUDs in general that would be salient in further study.  First,
the population of MOO-users, though definitely skewed toward males, does not
appear to be quite as skewed as Curtis noted in his paper.  In any case, the
causes and effects of this gender bias are central in any study of MUD users.
The majority of MUD users are from the still male-biased computer science
field, though the breakdown of career and major choices, as well as the gender
breakdown of subjects within each field, is another salient topic for further
study.  The large majority of users who emphasized the social aspect over the
programming or exploration aspects of MOO presents many topics for further
speculation and study, including exactly how social interaction on a MUD system
operates, how it differs from real-life conversation, and how it differs from
other forms of virtual communication.  In addition, the way in which people's
social styles differ or not from their styles on a MUD is an important issue;
do people use this as a substitute for or as an addition to real-life social
interaction.  Finally, it seems that the atmosphere of any particular MUD
system would affect the way in which social interaction occurs, depending on
whether the MUD was more or less role-playing oriented and whether it could be
programmed and tailored to its users.

	Before any large-scale psychological study of MUD systems could be
executed, it appears that it is first crucial to understand the actual
structure and content of conversations on MUDs.  From this data, further
studies can be modelled to see exactly how the different types of communication
that take place on these systems might be indicative of the different types of
people that might use any one medium, or how different types of electronic
media affect the way its users communicate.

Studying the Structure of Computer-Based Communication

	The experimenters decided to compare three mediums of electronic
communication to determine differences among their conversational structure.
Two of these mediums were LambdaMOO and IRC, since the great similarities yet
great differences between the two lend themselves well to discovering exactly
how the different atmospheres of the two affect electronic communication in
itself, as well as the personality issues surrounding such communication.

	The third medium employed by the researchers differs more from the
first two in structure.  The researchers decided to also focus attention on the
in-house bulletin board which runs on their own school's VAX/VMS system,
VaxNotes.  Notes is a bulletin board with an active community of users from the
researcher's own school, a co-educational liberal arts school of about 1,200
students, as well as from the institution's sister school, an all-female
liberal arts school of 1,200 students..  Notes will provide a good contrast to
MOO and IRC in several respects.  A very active community of students in a
common educational setting and social community uses this facility as a medium
for expression and communication.  Notes itself is a non-interactive medium,
unlike MOO and IRC.  In addition, Notes users post under their own usernames
and not pseudonyms, and are much more likely to know each other in a real-life
setting.  However, there are still "conferences" like MOO rooms or IRC
channels, which can be focused on a very specific topic or be more generally
oriented.  On Notes, there is a "Miscellaneous" topic which is for banter and
conversation, and also a "Computer Problems" topic which, like Hacker's Heaven
on MOO and #hack on IRC, is more focused on discussing computing.  Both
conferences can be used to discuss the same types of topics that are discussed
on MOO and IRC, but without the interactive medium.

Expectations of the Research

	In doing our observations of all three settings, we began with very few
hypotheses, both because we were doing the observations in a naturalistic
fashion and wanted to impose as few preconceptions as possible on the findings,
and also because there is little background information on the differences
between these three settings which would allow us to form hypotheses.  However,
we had some general sense of the settings which caused us to expect a few
effects.  We expected that MOO settings would contain more paralinguistic cues
in the form of facial expressions, nodding, and so on, due to the presence of
the emote function in the MOO setting and the lack of any such function in IRC
and Notes.  Though users of IRC and Notes can and do express emotion/facial
expressions/actions by utterances such as "*bonk*" or ":)", MOO is the only
system that has a preprogrammed function for expressing such emotions or
actions.  Thus, MOO better invites such behavior.

	We also expected an effect due to gender of observer, either on
flirtatious or hostile behavior toward a female observer, the latter due to the
still-prevalent notion of computing as a male domain, and the fact that the MOO
and IRC settings have more male than female players.  We expected that
conversation in a computer topic-oriented room, channel, or Notes topic would
indeed show more utterances about computing in general, the atmosphere of the
MOO and IRC, discussion of players and rooms, and perhaps be slightly more
intimidating/hostile to newcomers attempting to join the conversation, while a
less-focused setting would have a much more free-form pattern of conversation.
Otherwise, we tried to analyze the data with as little preconceptions as
possible as to the number of different types of utterances that would be found
in each setting and subsetting.  Instead, we focused on creating a set of codes
that would reflect accurately the many types of utterances we had seen in our
experiences with the three mediums.

	We believe that finding out more concrete attributes of the nature of
MUD conversation, as opposed to other forms of electronic communication, is
important for several reasons.  First, the increasing sophistication and
widespread nature of MUD systems is an aspect of computer communication that
can no longer being ignored, for it appears to be a medium which is creating an
entirely new subculture, with its own language, customs, and paralinguistic
means of communication.  Second, considering the possible uses of such media
for the scientific or business communities, as settings in which to generate
and debate ideas, a study of how this type of problem-solving communication
occurs would address the question of how this medium will affect the creative
and communicative process among such groups.  Finally, such a study would
address the even more general issue of the implications of virtual reality, as
it increases in complexity and its ability to simulate an intricate and
compelling environment and social medium.  As more people begin to explore
these environments and use them as mediums for self-expression, socialization,
and relationships, the nature and implications of such communication is a
crucial issue for psychology and all related disciplines.

Method

Subjects

	The subjects were participants in three communicative settings: MOO,
IRC, and Notes.  Sample sessions were recorded from each of the three mediums,
such that the number of subjects in any given session varied.  For our two MOO
logs, the number of people in the Living Room setting were five (all apparently
male) and eight (six apparently male, two guests with indeterminable gender),
respectively, and the number of people in the Hacker's Heaven setting were both
nine (in both cases, 8 apparently male, 1 apparently female) respectively.  For
our two IRC logs the number of people in the #hottub setting were 22 (18
apparently male, 4 apparently female) and 25 (23 apparently male, 2 apparently
female), and the number of people in the #hack setting were 9 (all apparently
male) and 13 (again, all apparently male).  For the Notes logs, the number of
people in the Computer Problems log were five (four male and one female), and
the number of people in the Miscellaneous log were seven (four male and three
female).

	Since all the players on MOO and IRC use pseudonyms, it is difficult to
determine the actual sex or age of the players.  However, through our pilot
survey on MOO and our own experience with MOO and IRC, we can make some
assumptions as to the demographics of the subjects.  Both MOO and IRC users
tend to be fairly young, ranging in age from about 16 to 25, though we have met
a range of people from pre-teen age to late thirties.  The survey on MOO
indicated that the MOO population tends to be skewed toward males; this also
appears to be true of IRC judging from the experimenters' interactions there.
Most of the players are students, since it is mainly students that have access
to the Internet through a school-based accounts.  Younger people may use their
parent's account or a hacked connection; older people may have an account as a
grad student or connected to their work in the computer field.  The MOO survey
also indicates that over half of MOO users work or plan to work in computer
related fields; experiences with IRC do not present a clear picture of whether
this is true of the IRC population.

	Notes users, on the other hand, can be more accurately described.  They
are students at the experimenters' school or its sister school, two liberal
arts colleges of about 1,200 students each, one of them an all-women school and
the other co-ed.  They are mostly all of college students' age, 18 through 22,
except for the occasional faculty member or alum that may post to or read the
bulletin board.  Despite the gender-skewed population that the association of a
co-ed and all female school produces, Notes appears to be about half male and
half female.

Procedure

	The MOO and IRC logs consisted of sessions of 20 minutes each, ten
minutes spent in each of two subsettings, the MOO Living Room/IRC #hottub
channel or the MOO Hacker's Heaven/IRC #hack channel.  Two logs from each
system were obtained, each with a female and male observer, named BethAnne and
Sylvan.  The experimenters would move the character into the room or channel,
say hello, and simply watch, only responding to direct questioning.  After 10
minutes were up in the first subsetting, the experimenters would move the
character to the other subsetting and repeat the process.

	The Notes logs were made up of a selection from the Miscellaneous topic
(corresponding roughly to the Living Room on MOO and #hottub on IRC) and a
selection from the Computer Problems topic (corresponding to Hacker's Heaven
and #hack).  A fairly general, conversational log was chosen from
Miscellaneous, rather than one with a very focused topic; the log from Computer
Problems was taken from the same time period as the Misc log was posted.

	The utterances in each log were coded according to a coding scheme
devised by the experimenters.  The codes (Appendix F), revised on the basis of
practice with earlier sample transcripts of LambdaMOO, encompassed the many
types of behavior the researchers had noticed in their experiences with MOO,
IRC, and Notes.  Throughout this refinement process, if it was found that there
were two or more codes that were ambiguous in their correct application to
similar types of statements, these codes were collapsed, though collapsing two
types of behavior into one code might have at first seemed counterintiuitive.
The codes encompassed greetings, biographical information, gestures that keep
conversation going ("nudges"), expressions of affection, humor, and hostility,
physical action, interaction with system code, talking about computers and
programming, discussing rooms, topics, or channels in general, discussing
individual's behavior and characters, and an "other" category for uncodable
statements.

	The experimenters decided to code only the first forty utterances of
each subsetting, since this was the size of the smallest subsetting transcript.
In order to get a better sense of how individual players contribute to such
conversations, the utterances were tallied by player, according to the number
of each type of utterance each player had made in the course of the first forty
statements.  Each player's setting, subsetting, and sex of observer present was
noted, and then the number of utterances of each code-type made by each player
were tallied.

	The experimenters derived the percentage of their reliability for their
coding of each setting/subsetting; the average of these reliability percentages
was 82.9 percent.  After coding separately and figuring out this reliability
percentage, the experimenters discussed their differences in coding and agreed
on which codes should be used for the utterances on which they had initial
coding disagreements.  Though the reliability percentage was less than the
experimenters had hoped, they also realize that it is almost impossible to
refine the codes to the extent that reliability would be greater, due to the
extremely subjective nature of interpreting MOO, IRC, and Notes conversations,
especially IRC due to its much less structured nature of conversation.

Results

	The utterances were tallied according to player, noting the player's
setting (MOO, IRC, or Notes) subsetting (Living Room/#hottub/Miscellaneous or
Hacker's Heaven/#hack/Computer Problems), gender of observer (neutral for
Notes), and the number of each type of utterance they made (codes A through M).
Multiple regression analyses were performed on the data.  We discovered no
significant effect of observer gender, so did not include this variable in
further analysis.  These results are summarized in Tables 1-16.

	A multiple regression analysis of the mean number of greetings in each
setting and subsetting reveal a significant main effect for both setting (p <
.001) and subsetting (p < .001), and a significant interaction effect (p <
.003).  Post-hoc t-tests of the means reveal that there are significantly more
greetings on MOO than on IRC (p < .009), and that there are more greetings in
the Living Room/#hottub subsettings than in the Hacker's Heaven/#hack
subsettings.  The interaction effect suggests that the difference between the
number of greetings in the two subsettings on MOO is much greater than that of
IRC.  There are no significant effects of VaxNotes for this type of utterance.

	The analysis of the mean numbers of biographical information statements
reveal a significant main effect for setting (p < .017).  Post-hoc t-tests show
a significantly greater number of biographical information utterances in IRC
than in MOO (p < .032).  There are more bioinfo utterances on Notes than on
MOO, though this difference only approaches significance (p < .054), and there
is not a significant difference between the number of bioinfo statements in
Notes and IRC.

	The analysis of the mean numbers of code statements (statements showing
a player's interaction with system code) show a significant main effect for
setting (p < .002), subsetting (p < .041) and a significant interaction effect
(p < .045).  Post-hoc t-tests show a significantly greater amount of code
statements in the MOO setting than the IRC setting (p < .034), and that there
are more code statements in the Living Room/#hack setting than in the Hacker's
Heaven/#hack setting.  The interaction effects that there is a much greater
difference in the number of code statements between the two settings on MOO
than there is between the two settings on IRC.  There was no such significant
effect for VaxNotes.

	The analysis of the mean number of humor statements per person show a
significant main effect for both setting (p < .006) and subsetting (p < .06),
and a significant interaction effect (p < .02).  However, no two pairs of means
differed significantly, according to t-tests, though the cell means do indicate
that Notes has the most expression of humor.

	The analysis of the mean number of "technobabble" (talking about
computers) statements per person reveals that there is a significant main
effect for setting (p < .001), subsetting (p < .003) and a significant
interaction effect (p < .001).  Post-hoc t-tests show that there is a
significant difference between MOO and IRC (p < .027), with MOO having a
greater number of technobabble statements per person, and that there are more
technobabble statements made in hacking settings than living-room type
settings.  No such significant effect was found for VaxNotes, but the
interaction effect suggests that there is a great difference in Notes between
the amount of technobabble statements made in Miscellaneous and Computer
Problems, while there is little difference between the amount of such
statements made in the analogous settings in MOO and IRC.

	Analysis of the mean number of affectionate statements per person shows
a significant main effect for subsetting (p < .037).  This is due to the fact
that there are no expressions of affection that took place in hacking settings
in our transcripts.  MOO had more such expressions, but only to an extent that
approaches significance as a main effect for setting (p < .053).

	The analysis of the mean number of physical action statements per
person shows a significant main effect for subsetting (p < .034).  The cell
means show no physical action in the MOO or IRC living room/#hottub
subsettings, no physical action in either of the notes subsettings, and none in
the IRC #hack subsettings.  The only subsetting that shows any physical action
at all in our data is MOO Hacker's Heaven.

	The analysis of the mean number of statements coded "other" reveal a
significant main effect for setting (p < .004).  The cell means for statements
show that many more of the Notes statements were coded as "other" than the MOO
or IRC statements.

	A second analysis was performed on the data, this time to simply tally
the mean number of each type of utterance made in each setting and subsetting
overall, instead of per person (Figures 1 and 2).  Few meaningful significant
effects were found for this analysis.  This is most likely due to the fact that
this analysis is less suited to elucidate the differences in conversational
structure, because of its insensitivity to the number of statements made by
each individual person.

	The meaningful significant finding illustrates how the difference
between tallying types of statements per person produces more specific results
than tallying statements overall.  There was a significant main effect for
greeting statements for setting (p < .009) and subsetting (p < .009).  Looking
at the cell means, it can be found that there are more greetings overall on IRC
than MOO, which seems to contradict the finding that there are more greetings
per person on MOO than IRC.  However, in the context of the two settings this
finding makes sense.  In IRC, there are more greetings overall due to the
larger number of people, but less per person, apparently since MOO is an
environment in which people tend to be more inclusive of new people in
conversations.  Thus, there is less greeting going on overall in MOO settings,
but each individual is much more likely to make more greeting statements than
an individual player on IRC.

Discussion

	According to the data, MOO had significantly greater amounts of
greetings, interactions with code, technobabble, and physical action, as well
as a trend toward more affectionate statements.  Notes had more biographical
information, "other" statements, and a trend towards more humorous statements.
The Living Room/#hottub setting (though not the Notes Miscellaneous setting)
had significantly more greetings, while Hacker's Heaven and #hack (though not
the Notes Computer Problems setting) had more statements indicating interaction
with code as well as more technobabble.

	The absence of any effect of observer gender does not support the
anecdotal evidence that females at least in the MOO setting encounter more
flirtatious, and often more hostile, behavior toward them.  However, the lack
of a significant result for this factor does not mean that this effect does not
exist, for the experimenter's own experiences on MOO and IRC fully support the
anecdotal evidence.  The lack of effect of observer gender could be partially
caused in the MOO setting by the fact that BethAnne was a new character, and
this author found in her own experience with her regular female character
(TamLin) that she did not encounter much flirtatious behavior until she was a
few weeks old on MOO and thus more comfortable interacting with the community,
talking to more players in a more comfortable fashion, and known as a female
name.  On both IRC, the lack of effect could have also been due to the fact
that we did not code private messages to BethAnne, or utterances made by one
player to another that only the two of them can hear.  In the IRC log, we had
two (male) people page us acting in an extremely friendly, forward fashion,
even though nobody in the room did so publicly.  We suspect that if we had
gathered several more logs of each kind, and coded private messages in MOO and
IRC, we would have encountered more flirtatious behavior toward BethAnne in the
MOO and IRC settings.

	The results for greetings support the experimenter's experiences on MOO
and IRC.  On MOO, the conversations in Hacker's Heaven tend to be the more
focused, with people actively discussing the MOO, players, programming, trying
out objects, and making jokes, than those of the Living Room, which often just
seem to be more players "hanging out" with no real topic of conversation in
mind (though intense and active conversations do occur; they just seem rarer in
this setting according to the experimenter's experience).  Thus, there is much
more attention focused on greetings in the Living Room, since new arrivals do
not disrupt the flow of conversation quite as much, than in Hacker's Heaven,
where people might be too focused on their topic of conversation or figuring
out their programming work to greet each new arrival.  In IRC, both the #hottub
and #hack settings are very active conversations, though #hack like Hacker's
Heaven is more structurally focused on one topic.  IRC is composed of more
greetings per conversation, but only due to the large crowd of people; on MOO a
player is much more likely to get a greeting from any individual person.  This
supports the experimenter's experience that less attention overall is paid to
new arrivals on an IRC channel than new arrivals to a MOO room, and that the
MOO is a much friendlier atmosphere than IRC.  In addition, the greater number
of greetings in the Living Room/#hottub subsettings supports the notion that
both of these subsettings are less focused than the Hacker's Heaven/#hottub
settings, which are more oriented toward discussing topics specifically related
to computing rather than just small talk.

	The greater amount of biographical information on IRC supports the
hypothesis that IRC is less of a fantasy realm than MOO; that people are less
playing characters and more just there to talk as "themselves" simply under
pseudonyms.  This is not to imply that MOO characters are simply playing a
role, but that there might be more subtle or unsubtle role-playing and actual
development of a character personality (which may or may not be related to the
player's real life personality) occurring on MOO than IRC.  The fact that the
Notes data does not have significantly more biographical information statements
than MOO is difficult to understand, since it would seem that Notes is somewhat
closer to the non-virtual realm, with users using their real life names and
often talking about college life and one another.  However, the lack of
significance could just be a function of the small data pool; this conclusion
is especially supported by the fact that the trend of more bioinfo statements
on Notes than MOO does approach significance.

	The findings for the number of "code statements" made in each setting
and subsetting for MOO and IRC bears out well the experimenter's experience
with the mediums.  MOO should have more interactions with the system since it
is highly programmable, and thus invites more interaction with code than does
IRC.  Though it initially seems that the hacking settings should have more code
interactions than the living room/hottub settings, this is not true, since our
code statements included people's entrance and exit messages to a room or
channel, and many more people tend to pop in and out of the living room
settings than the hacking settings, which are less chaotic in this manner.

	The experimenters are unsure how to interpret the humor data.  Though
there is an overall effect for humor, there is no significant difference
between the pairs of cell means.  There seems to be no obvious reason why any
of the three settings would have more humor statements than any other one,
except perhaps that MOO has a greater capacity for surreal behavior due to the
fact that it is the one closest to a "virtual reality" of the three mediums.
Why Notes would have the most humor, whether significantly so or not, is
unclear to this experimenter.  A possible explanation is the non-interactive
quality of Notes, which requires that participants make more novel and
attention-getting statements to keep other participants' attention (Doug Davis,
personal communication).  This would not be as necessary on MOO or IRC, due to
the fact that the interactive nature of the medium forces players to keep
attention regardless of the type of statement.

	The effects for technobabble are consistent with MOO and IRC just as
the effects for code statements are consistent.  MOO is a more programmable
system, which would account for more technobabble in all subsettings as
characters learn how to interact with the system, while IRC technobabble seems
to be more confined to the channels set aside for talking about computing.  Of
course, subsettings focusing on computing on either setting should have more
computer-related statements, which is reflected in the data.  The lack of
significant effects for subsetting in Notes may have been due to the small data
pool, since the cell means indicated that there was a good deal of technobabble
happening in Computer Problems while none was happening in Miscellaneous.
However, for setting, the data most likely indicates that Notes users spend
much less time talking about computing than IRC and MOO users.

	It appears from the data on affection that hacking settings have much
less affection than living room/hottub type settings, which supports the
experimenter's insight that hacking settings are often more aloof,
intimidating, or hostile in atmosphere.

	It would seem that MOO is much more conducive to physical action than
IRC, due to the presence of the emote key in MOO, and that Notes is not
applicable to this code at all.  However, the lack of a significant main effect
for setting may just be due to the small data pool.  There seems to be no
apparent reason why there would be more physical action in MOO Hacker's Heaven.

	The significantly greater number of "other" statements in MOO simply
indicates the experimenter's assumption that MOO is a less comparable medium to
MOO than is IRC, and this cannot be coded as well according to the codes
originally constructed in accordance with the MOO setting.

	From this data, it can be concluded that MOO is a friendlier
atmosphere, more conducive to interaction with the system, and more conducive
to physical action in the virtual sense.  It can also be concluded that hacking
settings in all cases tend to be less friendly toward newcomers, as well as
lending themselves more to both talking about computing, and, in the MOO's
case, actual programming in the midst of conversation.

Implications for Personality Research

	The findings of the study suggest avenues of further research both in
the area of conversational structure as well as of personality analysis.  It is
obvious that another study of conversational structure is necessary to better
understand electronic communication, and to confirm that the nonsignificant
trends in the data are actually significant differences between the systems.  A
study with differently structured or more generalized codes, or with more
detailed ways of coding each utterance (such as coding for tone and content, or
for line length) would most likely lead to a greater number of significant
results.

	Despite the need for more detailed data on conversational structure in
electronic mediums, the initial data and personal experience we have does
present some indications as to the type of people that might be attracted to
such systems, thus presenting suggestions for further research in the area of
personality psychology.

	The issues presented by the first formation of a "computer culture" are
still relevant to those who choose to communicate through electronic media,
even though such media are more widely available and used.  Bulletin boards,
IRC, and MUDs (the latter especially) are fairly new media; though their use is
growing, they are still not in use by a large segment of the population.  Thus,
those who use these media do form a somewhat exclusive group, just as the
original devotees of the computer culture formed their own exclusive group.
Like the latter group, the electronic communication subculture should be
expected to form its own set of jargon, etiquette, expectations, and other such
rules.  This trend is already occurring, evident in such phenomenon as new
abbreviations (for example, IMHO, meaning "In my humble opinion," is common on
bulletin boards; the terms brb, meaning "be right back," and IRL, meaning "in
real life," are common terminologies on interactive systems) and new, often
unclear rules of etiquette (is it appropriate to make harsh personal attacks on
a bulletin board?  is it appropriate to use MOO programming skills to "spy" on
a person in another room?  is the use of obscenity on IRC justification for
being "kicked" off a channel?).  Since this new language and set of rules is so
new, they are constantly in flux; even the experienced user can come across new
jargon, and there are constant disagreements over appropriate behavior on all
of these systems.  However, this phenomenon contributes to the exclusivity of
the electronic communication world, since it constitutes the existence of a
subculture which can only be understood from within.

	Judging by the greater number of code statements, or on MOO, it can be
inferred that to fully appreciate the MOO system it is necessary to know more
about its workings and how to interact with it.  This indicates that even less
people will discover and use this new type of system, since it is more complex
and harder to learn.  Thus, a topic for further research is whether the MUDding
community is indeed the most exclusive, and if so, how this affects both the
type of person that uses MUD systems.  In addition, the extent to which this
exclusivity affects the development of the subculture itself can be researched.
For instance, do MUD systems tend to have a greater amount of jargon and more
disputes over appropriate behavior than the other two mediums?  Finding such an
effect would suggest that MUD culture is more exclusive and separate from the
real-life realm, and thus richer and more detailed as a culture in itself.

	The issue of computing being a realm in which a participant can have
strict control over self and actions is also very relevant both to our findings
and to further study.  Since we found that more biographical information tends
to be discussed on Notes, it makes sense that a person who is shy and wanted to
have the control of remaining anonymous would be even more attracted to an IRC
or MUD system than s/he would be to a school-based bulletin board where his/her
identity could be easily discovered.  In addition, in the MOO setting, a
participant could have the most control over the creation of a character, even
one of very different looks, personality, or even gender than the participant's
real-life identity.  It appears that MOO, both because of its anonymity and
potential for detailed character description/role-playing, is the best medium
in which to study issues of control.  However, all three mediums seem relevant
to these issues.

	Further research must be done on the effect of observer gender on
electronic communication.  Though the researchers obtained no significant
effect of gender on the communicative structure of conversations, it seems
unlikely that the gender of a player would have little impact on his or her
experiences with these media.  Based on prior studies of gender and computing,
it appears that women might be less likely to become involved in such media.
This is supported both by the gender skew of MOO noted by both our pilot study
and Curtis', the skew toward males in our IRC and MOO subject pools, and the
fact that Notes is only half-female in a community made up of three times as
many females as males.

	Even when becoming involved in such systems, the communicative
structures and norms of the systems many often be affected by their presence,
or have an effect on their comfort level with the use of such systems.  On IRC,
though no significant observer effect was noted in the data, the researchers'
BethAnne character received some hostile remarks after joining the #hack
channel, suggesting that many of the players felt a female character had no
business participating in a discussion of computing.  No such effects were
noted on MOO, though it could be argued that the greater amount of technobabble
in the Hackers' Heaven setting (as well at the #hack setting on IRC) is more
likely to become intimidating to women, who tend to generally have less
experience with computers and programming.  However, electronic communication
is likely to bring more women into computing in general, since it defies the
stereotype of computing as a solitary, competitive activity.  In any case, both
the effect of observer gender on the type of communication that takes place on
electronic systems, as well as the effect of the systems on what gender is most
likely to use them, must be further studied.

	Our data also indicates that MOO is the most detailed of these
environments, both in its capacity for character and room description and,
judging by the number of code statements in MOO, its capacity for programming
and manipulation of the system.  This aspect of MOO brings into question an
aspect of electronic communication noted by Hellerstein in her study, that of
addictiveness.  As noted earlier, she quotes students in her study who warn
against the addictiveness of the bulletin board environment, sometimes to the
point of causing its users to ignore other real-life concerns.  Considering
that our data indicates that IRC and MOO have a much greater potential for
creating a "realistic" atmosphere, both because of their interactive natures
and, in MOO's case, the ability to create a detailed and interactive virtual
environment, the issue of addiction becomes an even greater concern.  MOO
addiction was a topic discussed with us at many points in our conversations
with MOO participants, but was not an issue we had the capacity to look at in
our study.  However, the reasons for the seemingly addictive nature of
electronic communication, as well as the extent of addiction that different
environments can produce, is a crucial topic for further study, especially as
more people use such systems and become prone to becoming affected by this
issue.

	Many other aspects of computing and its personality correlates are
discussed in the literature, but the data does not present enough information
to make assumptions about their relevance to the systems we studied.  For
instance, the issue of intimacy in virtual settings cannot be addressed well by
the data.  The greater number of biographical statements made by people in
Notes may lead a reader to think that participants in Notes are the most
self-revelatory, but this negates the fact that, from the researchers'
experience, most personal conversations on MOO and IRC take place on private
channels/rooms, not in the common channels or rooms.

	It might seem that the role-playing aspect of MOO or even IRC, which
allow the capacity for gender-switching or for presenting a different
personality, might detract from the potential for intimacy in such systems.
However, some of the data and the inherent characteristics of both MOO and IRC
contradict this notion.  The greeting data indicates that MOO is a fairly
friendly environment, with participants who are very willing to draw people
into conversations.  It can be inferred that this atmosphere would tend to make
people very comfortable with the MOO environment, and thus encourage them to
talk more with people.  Also, due simply to the anonymity of MOO and IRC,
people often feel safe to say things they would not say to a person
face-to-face, or even virtually if their real name was being used.  Finally,
the text-based nature of all three mediums force people to be very detailed and
specific in their self-expressions.  When discussion emotional or biographical
issues, this form of communication can prove to be very self-revealing.  In any
case, the ways in which the conversational structure of electronic
communication might affect issues of intimacy between participants deserves
further study.

	Other issues in the literature, such as the relationship of Rhue and
Lynn's concept of fantasy-proneness to involvement in computing, or the aspects
of electronic communication contributing to Kiesler's finding that it takes
longer to reach consensus by electronic means, cannot be addressed by our
findings.  However, these are issues that can be addressed once the structure
of electronic communication is better understood, since this knowledge will
allow researchers to better determine what types of people become involved in
these structures.

	Through our research, we have gained some insight into the ways that
different electronic mediums can affect conversations taking place on them.
Through analyzing this data, it is obvious that this is a new and broadening
field which deserves much more detailed research.  More information on this
subject will provide insight into the way participants interact with these
systems, allowing psychologists to better understand the personality correlates
of those attracted to electronic media, the psychological effects of such
systems, and how the interaction of the human mind with computer-based systems
can produce new and revealing psychological phenomena.

References

Arndt, S., Clavenger, J., and Meiskey, L.  (1985).  
Students' Attitudes Toward Computers.  Computers and the Social Sciences, 1,
181-190.

Curtis, P.  (1991).  Mudding:  Social Phenomena in
Text-Based Virtual Realities.  A draft of an unpublished paper (dated November
11, 1991), submitted to the Conference on Directions and Implications of
Advanced Computing.  Available at [parcftp.xerox.com]/pub/MOO/papers/DIAC92.ps.

Curtis, P.  Personal communication (via LambdaMOO),
November 18, 1991

Davis, D.  Personal Communication.

Gilligan, Carol.  In a Different Voice:  Psychological
Theory and Women's Development.  Cambridge, Massachusetts:  Harvard University
Press, 1982.

Hellerstein, L. N.  (1985).  The Social Use of
Electronic Communication at a Major University.  Computers and the Social
Sciences, 1, 191-197.

Holmes, J. E.  (1980).  Confessions of a Dungeon
 Master.  Psychology Today, November, 84-94.

Jennings, K.  (1990).  The Devouring Fungus:  Tales of
the Computer Age.  New York:  W. W. Norton and Company, Inc.

Kiesler, S., Siegel, J., and McGuire, T. W..  (1984).  
Social Psychological Aspects of Computer Communication.  American Psychologist,
39(10), 1123-1134.

Levin, T. and Gordon, C.  (1989).  Effect of Gender and
Computer Experience on Attitudes Toward Computers.  Journal of Educational
Computing Research, 5(1), 69-88.

McClure, R. F. and Mears, F. G.  (1984).  Video Game
Players:  Personality Characteristics and Demographic Variables.  Psychological
Reports, 55, 271-276.

McClure, R. F. and Mears, F. G.  (1986).  Videogame
Playing and Psychopathology.  Psychological Reports, 59 59-62.

Maddi, S. R., Hoover, M., and Kobasa, S. C.  (1983).  
High Activation and Internal Orientation as Factors in Creativity.  Journal of
Creative Behavior, 16(14), 250-255.

Marks, A.  Personal communication.

Melancon, J. G. and Thompson, B.  (1985).  Selected
Correlates of Computer Arcade Game Play.  Perceptual and Motor Skills,
61(3, pt. 2), 1123-1129.

Mihalo, W.  (1985).  The Microcomputer and Social
Relationships.  Computers and the Social Sciences, 1, 199-205.

Rhue, J., and Lynn, S. J.  (1987).  Fantasy Proneness:  
Developmental Antecedents.  Journal of Personality.  55(1), 121-137.

Siann, G., Durndell, A., Macleod, H, and Gilssov, P.  
Stereotyping in Relation to the Gender Gap in Participation in Computing.
Educational Research, 30(2), 98-103.

Smith, J.  (jds@math.okstate.edu).  (1991).  FREQUENTLY
ASKED QUESTIONS:  Basic information about MUDs and MUDding (3 part series).
Public communication posted to rec.games.mud on Usenet on November 27, 1991.
The updated versions of Frequently Asked Questions can also be found on
spamalot.ecn.uoknor.eud (129.15.24.15) in pub/local and on
hardy.math.okstate.edu (139.78.1.13) in pub/uploads.

Smith, J.  Personal communication.

Sproull, L., Kiesler, S., and Zubrow, D.  (1984).  
Encountering an Alien Culture.  Journal of Social Issues, 40(3), 31-48.

Temple, L., and Lips, H.  (1989).  Gender Differences
and Similarities in Attitudes Toward Computers.  Computers in Human Behavior,
5, 215-226.

Toles-Patkin, T. (1986).  Rational Coordination in the
Dungeon.  Journal of Popular Culture, 20, 1-14.

Turkle, S.  (1984).  The Second Self:  Computers and
the Human Spirit.  New York:  Simon and Schuster, Inc.

(Credit should be given to several MOO players in helping with our background
information, notably Pavel Curtis (Haakon/Lambda) and Judy Anderson
(yduJ/Nosredna) of Xerox PARC, as well as the programming and implementation
of the survey, notably Judy Anderson as well as Dan Burford (Gilmore), West
Virginia University '95.  Thanks is also due to all the participants in
LambdaMOO who took the survey and/or spoke to us about the issues involved in
this project.)