A Typology of Social Roles in Usenet

[د. دنحا طوبيا كوركيس]

Dear All,
Below please find an interesting piece of research which I saved on my PC four years ago; it was found only this morning in a hidden file. Enjoy reading it.

Best,

Dinha




A Typology of Social Roles in Usenet

Scott A. Golder

Senior Honors Thesis
Harvard University
Department of Linguistics
March 23, 2003


Table of Contents

Acknowledgments. i

Table of Contents. iii

Abstract v

1. Introduction. 1

1.1 Usenet Technology. 2

1.2 Usenet as a Social Environment 3

1.3 Constructing Identity in Usenet 5

2. Methodology. 9

2.1 Data Collection: Ethics. 9

2.2 Data Collection: Methods. 11

3. Related Work. 12

3.1 Speech Communities. 14

3.2 Conversation and Community. 16

3.3 Participation Inequality and Common Ground. 19

3.4 Roles and Strategies. 22

3.5 Roles and Expectations. 25

3.6 Defining the Social Role. 27

4. A Typology of Roles. 28

4.1 Mapping the Social Landscape. 29

4.2 The Newbie. 30

4.3 The Celebrity. 36

4.4 The Elder 40

4.5 The Lurker 42

4.6 The Flamer 44

4.7 The Troll 48

4.8 The Ranter 50

4.9 Role Reputation. 52

5. The Effect of Roles on Community. 54

5.1 Gender as a Social Role. 54

5.2 Protecting the Borders. 57

5.3 Population Shift and Topic Shift 61

6. Conclusion. 65

Postscript 69

References. 71




Abstract

How does a social community function when most of its participants are silent? An intriguing observation about Usenet newsgroups is that individuals participating in these communities do not do so equally; a very small proportion of individuals participate very frequently while the vast majority participate very infrequently. One of the main goals of this thesis is to explore both the causes and effects of this inequality.

Sociolinguist Dell Hymes observed that sociolinguistic taxonomy is the first step to explanatory theories of sociolinguistic phenomena. In light of this observation, I use a combination of qualitative and quantitative methods to develop a typology of social roles in Usenet. These methods include ethnographic study of individual newsgroups as well as statistical analysis of rates of participation. The theoretical foundation for developing this typology is the ethnography of communication, a multidisciplinary approach that integrates ideas from linguistics, sociology, psychology and anthropology.

This typology serves two main functions. First, it explains participation inequality; users have different needs and goals, as well as different abilities and privileges, all of which require participants to behave differently from one another. Second, this typology explains changes in communities. A key premise of this thesis is that a community is defined by the roles that are being enacted within it. Change in the roles that the participants enact is shown to cause changes in the newsgroup community; these changes can come in the form of stronger or weaker social cohesion, as well as in shift in overall newsgroup topic or community standards.



1. Introduction
Usenet is a worldwide text-based conversational environment made up of a network of linked computers connected to the Internet. Conversations take place in groups called “newsgroups,” which are arranged hierarchically by topic. Each newsgroup is home to its own thriving social community. An intriguing observation about Usenet newsgroups is that the people participating in these communities do not do so equally. On the contrary, it has been observed that a very small proportion of the participants author a very large proportion of the messages (Whittaker et al. 1998). Many users will “post” a single message and never be heard from again. Others will read silently and never post even once, never making their presence known. One of the main goals of this thesis is to explore this inequality among participants. How does a social community work when most of its participants are completely or nearly silent?

The need for a general, yet qualitative study of conversational behavior in newsgroups is great. Linguists and other social scientists have studied individual newsgroups’ communities qualitatively, and large-scale quantitative studies have been performed on Usenet as a whole (e.g. Whittaker et al. 1998; Smith 2001), but no recent Usenet-wide qualitative study exists. This thesis attempts to fill that gap. Through empirical study of behavior in a large number of groups, I find commonalities in social interaction across many newsgroups and attempt to draw some general conclusions about a variety of behavior types and how they influence group conversation.

The result of this empirical observation is the development of a typology of social roles. Understanding that Usenet participants do in fact have unspoken roles in their communities is essential to understanding why participation inequality exists. Goffman (1959:16) believes that a social role is “the enactment of rights and duties attached to a given status.” Recognizing that users have different roles allows acknowledging that they have different rights and duties, as well as different goals, needs and responsibilities within their communities. Acknowledging these different traits makes it possible to understand why shifts in communities take place. If a newsgroup community were an undifferentiated collection of like-minded people, then one would expect all conversation to be strictly topical. This is not the case, however, and newsgroups’ conversations deviate from the main topic of the group within limits set by the group. Additionally, over months or years, even the general topic of a newsgroup can change considerably. Viewing this phenomenon within the context of roles, I attempt to explain why these changes take place.

1.1 Usenet Technology
Begun in 1979 at Duke University, Usenet was developed as a system to share posted messages across the computer network. Soon, it grew to include several research labs and universities nationwide (Hauben and Hauben 1997). Public use of the internet has grown rapidly in the past ten years, as it has been transformed into a large-scale commercial public space. As a result, the numbers of people and newsgroups have steadily increased. For 2002, over 14 million people were actively posting in over 70,000 newsgroups (Smith 2001).[1]

The top-level hierarchy is the first division among newsgroups. Though many of these top-level hierarchies exist, there are eight main top-level newsgroup hierarchies, called the big eight: rec for recreational topics like hobbies and sports, soc for social issues like religion, sexuality and ethnicity, comp for computer-related topics, sci for science-related topics, humanities for topics like art, literature and philosophy, news for discussion of Usenet itself, and talk and misc for other conversation. Another hierarchy, alt, is for any topic at all. While it is not one of the big eight, it contains a very large number of groups and is the site of much activity on Usenet. Therefore, it may be grouped with the big eight in importance to Usenet participants.

Within each of the top-level hierarchies, newsgroups are further divided. For example, within the rec hierarchy exists the rec.gambling hierarchy. It contains newsgroups such as rec.gambling.blackjack and rec.gambling.poker. These examples demonstrate the newsgroup naming system; each newsgroup is named for its location in the hierarchy, much like the naming of biological species in the Linnean taxonomy, e.g. homo sapiens.

1.2 Usenet as a Social Environment
Usenet is successful as a conversational environment because it is interactive. Interactivity, as defined by Rafaeli (1988; qtd. in Cherny 1999:149), is essentially a condition in which each subsequent message refers to the previous messages. Indeed, Usenet may be considered the most interactive of electronic environments because in Usenet, successive messages not only make up the stuff of communication; they also define the structure for it. Within newsgroups, conversations are “threaded.” Message threading is a system in which each message is linked to the message to which it refers, so that a message is read immediately after the message to which it is a reply. This allows a conversation to diverge and split, allowing the simultaneous development of many conversations that all originated from a single message. A user may skip a particular branch in a message thread, yet continue to participate in the remainder of the conversation. The threaded structure works in Usenet because it is an “asynchronous” medium (Chen and Gaines 1996; Herring 1999). An asynchronous medium is one that enables communication over longer periods of time, whereas a synchronous medium is one that is ephemeral, for participation in real-time only. Messages posted to newsgroups can be read and responded to by others at any point during the time the post remains on the news server, which can range from days to months, depending on the particular newsgroup and server.[2] This allows users to read and participate in each diverging conversation in the thread. The fact that posted messages remain available for a long time after the time of writing allows for the creation of a “persistent textual record,” which reinforces the “coherence” of Usenet as a conversational medium (Herring 1999).

All of Usenet employs the same underlying technology and all users participate through essentially the same interface.[3] Despite this fact, each newsgroup develops as a unique social environment, with its own community-defined norms of behavior. For example, a newsgroup that is accepting of all newcomers might respond favorably to a question that is seen as elementary, while a newsgroup that places a high value on initiative and self-reliance in answering one’s own questions might make such a newcomer feel unwelcome. Similarly, “community standards” define what types of speech are acceptable. A newsgroup devoted to discussing Christianity might develop a reputation for fierce arguments that deteriorate into name-calling and other angry attacks called flamewars.[4] Another newsgroup with the same ostensible topic might have very different community standards, in which name-calling and flaming are low-prestige behaviors. The name of the group, however, does not give any indication of this difference; if a new user were to begin reading posts in alt.religion.christian and soc.religion.christian, he or she would have no initial understanding of how each community works. It would be impossible to grasp immediately that the former is connected to heated political discussions and is frequented by zealots, but the latter is focused on questions of theology. This can be overcome by investing time in reading a newsgroup’s conversation to learn the topics of discussion within the community.

1.3 Constructing Identity in Usenet
Social interaction requires more than knowledge of the topic of conversation; it is also important to know something about the people with whom one is interacting. Given Usenet’s text-only environment, the lack of social information that exists in day-to-day conversation makes developing an identity rather difficult. Both intentional and unintentional forces combine to make this so; not only does the Usenet interface make all users look alike (e.g. black text on a white background), the information that is provided – at minimum, just an email address – is, by design, not necessarily accurate. Like in day-to-day conversation, Usenet participants develop a reputation through repeated interaction. Unlike day-to-day social interaction, Usenet participants’ reputations are based entirely on the things they say and the information they provide.

The most obvious and basic pieces of information used to establish an identity are the email address and “real name.” Every message posted contains the email address and real name of the person posting it. This information is located in a line labeled “From:” at the top of each message:

(1) From: "Scott M. Kozel" <kozelsm@attbi.com>

The names and email addresses provided are not necessarily real. Many users are worried about spam (unsoliticed email), so they obfuscate their email addresses in some way so that automated email address harvesters, a common tool among internet advertisers, cannot read them:

(2) From: "Ed Kaulakis" <kaulakisNOSPAM@comcast.net>

For a large proportion of users, privacy is presumably a very high concern. Many users will use a pseudonym and a completely false email address, so that no trace of their offline identities remain:

(3) From: grey <not@available.com>

Variations on the pseudonym abound. Some, like “grey” above, don’t tell much about the person behind the pseudonym. Other people will construct an entire identity around a pseudonym:

(4) From: keeshah@my-deja.com (Mother Keesha Holloway)

“Mother Keesha Holloway” is the nom de plume of a frequent participant in misc.kids, whose posts are very strongly worded disapprobations against lenient child-rearing and lack of religious devotion. Over a long period of time, regular participants in the newsgroup have concluded that Keesha is not a real individual, but rather someone’s long-played, well-crafted game. Some participants have made it their personal project to unravel the well-guarded secrets of Mother Keesha, leading to a great deal of disruption for some and entertainment for others. This is a tale that is played out in other newsgroups, and figures prominently in Chapter ‎4.
In addition to email addresses that are constructed identities, there are some addresses that are no identity at all. Anonymous remailers are services that will strip the identity of the sender. Confusion can result, however, when multiple people use such a service to post messages to the same newsgroup:

(5) From: Nomen Nescio (Use-Author-Supplied-Address-Header@[127.1])
Newsgroups: rec.gambling.blackjack
Date: 2002-10-31 18:00:16 PST
Subject: Re: Where's the old crowd?

It is not Norman Nescio, it's Nomen Nescio.
It is Latin for "name unknown". You might remember a lot of posts which came from nobody@replay.com with the name Anonymous. A couple years ago the Dutch guy . . . resurrected the service . . . with the moniker "Nomen Nescio" instead of Anonymous.

It is for people to use who do not wish to post with their real names and email addresses. It is a free service which anyone can use. If you recall someone posting on 9/18 it was someone other than me, it was someone else using the remailer.

Someone besides the author of this post had been using the “Nomen Nescio” remailer. Because two real-world identities are associated with this fictive, anonymous one, the participants in the newsgroup will have trouble disambiguating the two (or more) authors behind it, due to the almost complete lack of identifying information. More likely, however, the authors behind Nomen Nescio will be conflated and treated as a single entity. If one author-nescio runs afoul of the group, the other(s) will be dragged down with him and all those writing under the Nomen Nescio name may suffer the fate of being ignored or attacked.

Since email addresses are not in a one-to-one relationship with individuals (users may have more than one address, or may share an address with another person), the email address and attached “real name” are clearly not ironclad markers of identity (Donath 1998). However, the email address is the tool most readily available by which to identify an author, and the lack of other cues makes the email address all the more crucial, right or wrong.

By and large, most Usenet participants will select and keep an email address and real name for a long time, whether these names represent their identity honestly or not. As a result, each participant develops a reputation over time. The aquarium enthusiast who provides helpful advice on caring for one’s fish, for example, may over time begin to be recognized as someone who can be trusted (Donath 1998). Hobby and computer groups, especially, tend to value the useful contributions of facts and opinions. Over time, individuals like the aquarium enthusiast, who demonstrate responsibility, gain the respect of their peers and their investment of time in the community yields a sense of ownership. Such a user may then feel as though he has the right and duty to speak for the community, perhaps to defend it against those who would disrupt it. Participants who provide verifiably incorrect facts, perhaps deliberately, become known for this and they risk not being taken seriously or being confronted as troublemakers. Sometimes oppositional people encounter vocal defenders of the group’s collective beliefs. The interaction of these two people becomes well known, and each in turn becomes known for his or her fighting with the other.

In this way, the totality of each person’s communication leads to the development of a very specific identity, one that is marked by patterns of behavior and language use. This can be considered their social role in the group. The emergence of roles in Usenet is especially interesting because they emerge ex nihilo; no roles are defined by Usenet. All the roles result directly from the interactions of the participants. Because demographics are continually changing, the social communities themselves are continually in flux. Communicative boundaries in Usenet communities are porous; any person is able to participate in a newsgroup. With a rapid influx of new participants, each with his own ideas of what the community is or should be, a newsgroup could undergo change in style and even topic over time. The conversational topic and style of each group are very much subject to the collective behavior of the people who are participating in the group at the time. In light of this, groups undergo shifts in the community’s standards of etiquette or rituals of behavior over the course of months or years.

This thesis attempts to systematize the changes in communities in Usenet, framing them as a result of the roles enacted within them. Some common roles occur repeatedly; prolific community leaders emerge, as to people who challenge them. New users everywhere must be socialized within their newfound communities. It is important to observe both universal roles and roles that are enacted in only certain environments. For example, perhaps some roles exist only in hobby-related groups and others only in politically charged discussion groups. Others still are nearly universal, enacted in groups of all types. The typology of roles in Chapter ‎4 explores such roles, defining them within the context of metrics derived from research in the various social science, supporting them with empirical data in the form of ethnographic observations. Chapter ‎2 discusses the methods of data collection and theoretical concerns regarding data analysis. Chapter ‎3 explores supporting work in the social sciences
Chapter ‎5, The Effect of Roles on Community, examines these roles in context, exploring how the several roles affect the structure and type of conversation in their respective newsgroups. More traditional sociolinguistic concepts are discussed in terms f these roles. Like traditional communities, electronic communities’ speech registers vary. Gendered speech forms exist, as do ritualized speech acts including ritual insult. Finally, an analysis of topic shift amidst population shift will reinforce the idea of social roles as a determinant of conversational behavior in the speech community.

2. Methodology
In every neighborhood, you need to know the people who are the central figures so that you can understand how society works and who influences who.

William Labov (1997)

Researching social roles required devoting a great deal of time to observing the people who enact them. A key assumption in this thesis is that the “central figures” of a newsgroup are those users who have some sort of strong impact on the community, good or bad. To begin, I primarily observed the most active participants. They are the easiest people to see, and have the most impact on the community, at least, where impact is measured in volume. There are arguably central figures in communities who do not participate frequently – this sort of person is an important role, too – but those who participate most frequently undertake the most speech events and are therefore most likely to have a social and linguistic impact on those around them.

2.1 Data Collection: Ethics
The set of data used as examples in this thesis was compiled through reading a number of weeks’ worth of conversation within the newsgroups from which the examples are drawn. The newsgroups from which the data in this study came were selected arbitrarily in order represent communities with a wide variety of topics. To this end, newsgroups in this study come from many of the big eight top-level hierarchies, including comp, misc, rec, sci and soc, as well as from alt.

An important benefit of ethnographic research in Usenet is that readership data is non-existent; the observer is himself unobserved. In this form of electronic research, there is no elicitation of data, only the silent recording of it. Thus, the “Observer’s Paradox” (Labov 1971), which describes the problem of elicited data never being completely “natural,” is avoided. Additionally, the problem of maintaining social relations with the informant is alleviated. For traditional fieldwork, it is important to build a trust relationship and to treat the informant respectfully, possibly providing compensation for his time (Vaux and Cooper 1999). Since the speaker is not aware that he is an informant and he is not going out of his way to provide the data, the linguist does not have to maintain a social relationship with the informant, and such considerations are no longer a concern.

The drawback of these benefits is the ethical problem that is presented; what responsibility does the researcher have to the people he is observing? Traditional fieldwork makes the relationship between the linguist and the informant overt; the latter is generally aware that he or she is providing data and that the data are for explicitly public, scholarly use. The newsgroup participant has no such awareness. Two contradictory goals arise, protecting the privacy of the informant and providing academically rigorous data. Susan Herring (1996) observes that two contradictory solutions present themselves. Legal scholars, Herring observes, suggest publishing names of and quotations from informants in full; they regard each message as a published piece of text. Others suggest that there is a “perceived privacy” in such electronic forums, and so pseudonym and paraphrase use is in order. Linguistic fieldwork requires attention to not only the ideas that informants express (which is what would be gained by using paraphrases), but the form of expression as well, so strict verbatim quotations are necessary.

The case is less clear for pseudonym use versus use of real names. The question then becomes, was the conversation taking place in a public or private environment? Though Usenet is undoubtedly public, conversations suggest that participants take ownership in the communities they construct, thereby creating what may be a false sense of privacy, or “perceived privacy.” That is, a private conversation in a public place like a coffee shop would undoubtedly be different if the participants were to have the same conversation in front of a live TV camera, though the first scenario is clearly in a public setting. It can be argued that, especially in the early days of Usenet, newsgroup conversations had a much greater sense of ephemerality; the network was much smaller, and many of the participants knew one another offline. There was no sense that archiving was taking place. In many ways, Usenet felt more like the coffee shop than the TV studio. Today, however, with widespread public participation on the internet and services like Google Groups maintaining archives of all Usenet posts, the “perceived privacy” has changed; people now expect their Usenet conversations to be available to anybody in the world.[5] Supporting this conclusion is the fact that many or most Usenet participants now use pseudonyms themselves. To create pseudo-pseudonyms would be superfluous. In light of these observations, all Usenet posts quoted in this thesis are quoted verbatim, with the author’s given email address and name.

2.2 Data Collection: Methods
In order to develop a comprehensive understanding of social interaction in Usenet, I have collected both qualitative and quantitative data about a variety of newsgroups. Quantitative and qualitative data provide different kinds of facts about behavior. Quantitative measurements of participation rates and response rates provide a framework for what is taking place within a newsgroup. Qualitative study of the messages provides insight into what the numerical trends mean and what behaviors are driving them.

Qualitative information about users’ interactions, collected over time, yielded significant trends in behavior. For each distinct user, a small biographical log was created, chronicling the user’s posts. [6] This log includes notes about the subject matter of the posts as well as others’ reactions to them. Users who participated more frequently had longer, more detailed biographies; users stood out whose posts were idiosyncratically about only one topic or particularly aggressive. This qualitative information made possible not only analyses of the participants, but also macro-level analyses of the behaviors commonly seen within newsgroups as a whole. Each user, taken alone, could appear to have some behavioral trait. Viewed collectively, however, it became clear when those sorts of behaviors were not personal traits, but were part of a group norm or community standard.

In addition to qualitative biographical data, quantitative data was also collected. To augment the qualitative data, statistical data were compiled for each participant and each newsgroup. These data included frequency of participation, occurrence of conversation initiation and average volume of responses per post. Written for the Loom2 project (cf. Boyd et al. 2002), the supporting software calculated these statistics over a cached set of newsgroup posts, generating profiles for each newsgroup and user. The group profiles were calculated relative to other groups. A user profile was generated for each participant in each group; if a person participated in three groups, for example, he would have three profiles, one for each group, calculated relative to the other participants in those groups. In each case, a user’s statistics compare him only to those with whom he interacts.[7] Together, the qualitative and quantitative data allow a micro/macro analysis of Usenet; the qualitative data present statistics for large numbers of newsgroups and participants and the qualitative analysis attempts to understand the meaning behind those numbers.

3. Related Work
The theoretical framework for this thesis is the ethnography of communication, the approach to the study of language whose foundations lie in linguistics, psychology and sociology. Hymes observes that, as a field of study, ethnography of communication may have a variety of names, but its distinctive feature is ethnographic analysis of conduct, regardless of the terminology used:

For many, the name of this perspective will be social anthropology or sociology, or psychology, or some other disciplinary category. The work required does fall somewhere within the purview of each such discipline, and there can be no quarrel with any, except to say that the division of the study of man into departmentalized disciplines seems itself often arbitrary and an obstacle. What is essential, in any case, is that the distinctive focus of concern advanced here be recognized and cultivated, whatever the disciplinary label.

(Hymes 1964:10)

In studying the conduct of Usenet groups, I adopt Hymes’ perspective that the speech community is the fundamental unit of study. At times, both Usenet as a whole and individual newsgroups will be studied as speech communities, with the understanding that speech communities, like all communities, have fluid boundaries and varying levels of granularity.

In developing a theory of social roles within the framework of ethnography of communication, I posit that the role one assumes in a community at once determines and is determined by the way one talks about things within the community. Because one is a leader within a social community, for example, he may speak more authoritatively. This authoritative speech in turn reifies his role as a leader. A mismatch of role and speech variety will result in discord; a leader speaking ineffectively may lose his status as a leader. By the same token, authoritative speech from a person not vested with such a role may be seen as presumptuous and certainly out of place.

Leadership, authority, popularity and power are important concepts in the examples drawn from the literature; Cherny’s ethnography of a social community online focuses on these concepts, and the inequality of participation level demonstrated in Whittaker et al.’s analysis is, in part, a result of such disparities. Holland and Skinner’s study of roles among college women explores the expectations people have for one another, given their perceived roles. Lastly, Erving Goffman’s theories of self-presentation are built upon in Jones and Pittman’s exploration of strategies for behavior as they relate to role acquisition. A synthesis of these works demonstrates the reflexive relationship between roles and those speech acts from which they result.

3.1 Speech Communities
Hymes (1972; 1974) provides a collection of concepts and categorizations useful in the description of linguistic communities. This includes an approach to segmenting speech into socially meaningful units, speech acts and speech events. He also explores methods of placing boundaries on communities. Hymes’ concept of “communicative competence” explains why simply being able to produce a syntactically correct sentence in a language is necessary but not sufficient for communicating in that language. “A child,” Hymes observes, “from whom any and all of the grammatical sentences of a language might come with equal likelihood would be of course a social monster” (1974:75). Instead, the child must understand the societal beliefs and attitudes about the things about which he speaks. This includes learning about standard ritual behavior. Saville-Troike (1982:21) extends this to include children who don’t “know better” than to use a “taboo expression in public,” as well as turn-taking behavior, and what to say to whom, and when. Newcomers to a newsgroup might go through the childlike ritual of being reprimanded for saying something that a seasoned participant would know to avoid. Additionally, it is possible to understand all the words in an utterance but not understand the meaning of the whole utterance in context. The bewilderment of a newcomer in a newsgroup is the outcome of such an occurrence, because the he is unfamiliar with either the style or the “in-jokes” of the group, or perhaps both. The new user may even misunderstand, interpreting at face value a comment that is, unbeknownst to him, loaded with in-group references. Hymes observes that linguistic form and meaning are inherently separable and offers the case of a child saying “I’m hungry” compared to a beggar saying the same. The intent of the first may be to delay his bedtime, while the intent of the second is more likely to receive charity (1974:5). Understanding the way in which a social group uses language is vital in order to participate fully within that group. Such a group may be considered a “speech community.”

Hymes (1974) defines a speech community as a group that “shar[es] rules for the conduct and interpretation of speech, and rules for the interpretation of at least one linguistic variety.” Because the vast majority of Usenet participants speak English, and because there is no phonetic information transmitted in the text-based communication of Usenet, it might appear that all users are speaking with the same speech variety.[8],[9] However, Usenet’s hierarchy of newsgroups lends itself to an analogous speech community hierarchy. A useful analogy can be found in speakers of English in the United States. Just as most or all native speakers of American English can understand one another with little or no difficulty, phonological and lexical differences notwithstanding, so too can most or all Usenet participants understand the content of most Usenet posts. In this way, all of Usenet can be considered a speech community. Even still, just as geographic regions and cities exhibit phonological and vocabulary differences as well as politeness, turn-taking and other discourse-level differences, linguistic features may develop in Usenet that are unique to a group or set of groups. Such features may be stylistic, involving standards for capitalization, punctuation or spelling, or lexical, involving acronyms or specialized meanings for common words. For example, “BOYC” is one lexical item that is limited in distribution to the alt.callahans newsgroup and is an acronym for “Beverage Of Your Choice.” The setting is a virtual bar called Callahan’s, where this term is used to offer a drink to another participant. Indeed, using this acronym in any other newsgroup would be as deviant a speech event as speaking a completely different language, because the term is not part of that newsgroup’s “linguistic repertoire” (Gumperz 1972:20), which comprises the speech forms, lexical items and discourse styles of that speech community.

3.2 Conversation and Community
In her 1999 book Conversation and Community, Lynn Cherny examined the linguistic register and communication patterns found in MUDs, or “multi-user dungeons.” The MUD is a text-based electronic community in which users on a computer network interact with one another in real time, bounding their conversation within a “virtual geography” of rooms. A subset of MUDs have themes, which give rise to role-playing within that theme. Cherny’s observations indicate a tendency toward science fiction and fantasy themes, like Star Trek, medieval towns and space stations. Each MUD’s text-entry system will support a variety of commands in addition to conversational text, so that the participant may interact with the MUD’s environment and objects within it. These commands also may facilitate gesture and other non-linguistic communication, such as hugging or hitting other users.

Though Usenet and MUDs are very different, they share a variety of features in their linguistic expression, which is unsurprising, given that both are text-only systems connected to the internet and their user communities have some degree of overlap (Raymond 2002). In both systems, users employ what have become common conventions in electronic communication. For example, absence of prosodic information is compensated for with the use of all capital letters, while absence of facial expression is compensated for with the use of emoticons.[10],[11] All these features have gained widespread recognition among participants in many types of electronic community as standard conventions (Crystal 2001), and are among the first aspects of electronic speech variety that new participants learn. Cherny’s study examines the features specific to MUDs, including turn-taking behavior, ritualized greetings and simple present tense verb usage.

The powerful message of Cherny’s linguistic-ethnographic study is that MUDs are homes to vibrant social communities with shared speech registers and norms of interaction and communication. Among other things, the study chronicles the social stratification of participants in one particular MUD called ElseMOO.[12] Before exploring the ElseMOO case, an overview of the control structure of MUDs in general is useful.

A MUD server is a software application run on a computer server. The central server software receives messages from the participants’ computers (“clients”) and distributes the messages to the other participants. Because a MUD is run from a central location, a certain amount of institutional control is necessary. It is the role of the “wizards,” or MUD administrators, to provide that control. Because the wizards are at once administrators and participants, a measure of social prestige and authority comes with the role. Cherny observes that Wizards themselves refer to their Wizardhood in terms of responsibility, calling the role “janitorial,” while non-wizard users view the Wizard role as prestigious and powerful. MUD Wizards also have the ability to grant others status as “programmer,” which gives those users the power to alter the geographic landscape of the MUD by creating and modifying the rooms therein. In ElseMOO, this privilege led to social unrest.

The “Power Elite” users in ElseMOO were a group of participants who were regarded variously by the general population as people who have been around a long time, people with programming ability, and people who are friends of Wizards. Because the definition of the term varied, there was no way to define who was part of the Power Elite group and who was not, but for many users there would be consensus about their Power Elite status; however, some users might be borderline cases. Additionally, the opinion one held of the Power Elite generally varied with one’s relationship toward them. For example, the Wizards cited by Cherny generally felt sympathetic toward the loosely-defined Power Elite group, claiming they have been active, friendly members and have served the MUD community by contributing useful programming projects. In contrast, some general users viewed Power Elite-ness as a privileged status and an exclusive social clique, and believed that those in the clique were “rude,” “standoffish” and “condescending” (1999:54).

The difference between a Wizard and a person who is Power Elite is important; a Wizard has a technologically vested, clearly defined role, albeit one that has a variety of social implications. A Power Elite, on the other hand, has no technological reason to have any more power or superiority than any other, regular MUD participant.[13] The lack of official status may be behind the animosity toward the Power Elite that is held by some participants. That is, because the status is unofficial, it may also be felt to be undeserved or unjust.

Every user quoted by Cherny seems to agree that the people who were Power Elite were active in the MUD community in some form, whether that form was in programming or in being very social. Thus, while the Wizard’s social role is granted to them by virtue of the technology of the MUD, the social role of the Power Elite is defined by the user’s behavior within the MUD.

By chronicling the Power Elite controversy in ElseMOO, Cherny demonstrates that electronic communities not only form speech communities, but “show characteristics of other types of community as well” (1999:293). Like MUDs, Usenet newsgroups develop characteristics of social communities, including differentiation into social roles. Unlike a MUD, Usenet newsgroups have no central server and thus have no administrative roles like wizards built into its architecture. People who become central figures in Usenet and in MUDs, including the Power Elite, do so by contributing to the social community around them. Cherny’s observation that the MUD has a linguistic register of its own implies that those who are active, prestigious participants conduct their participation in the register of the community. Their knowledge of how to communicate in the style that the community deems appropriate is an example of communicative competence (Hymes 1974), which is a useful concept in explaining why some users’ behavior is appropriate and other seemingly appropriate behavior is not.

3.3 Participation Inequality and Common Ground
Hymes observes that the rules regarding interaction and interpretation are normative (1972:63). When a participant in a speech event receives positive feedback after a speech act, his behavior is rewarded. On the contrary, when the feedback is negative or nonexistent, his behavior is punished.

When an individual’s behavior is not in line with the linguistic norms of the community, the response he receives may chastise him for his inappropriate behavior. Additionally, a total absence of response is also a kind of negative feedback (Zurcher 1983:140). For example, there is in many groups a prescription against messages that provide too little new content; such a message is likely to receive no response at all. Hackman (1992:204) observes that people are adept at determining whether or not they can find satisfaction in a group; if a participant feels that the feedback to his participation is negative, then he may not want to continue participating. Another alternative is that he may find it worth investing time in learning how to interact in the group properly.

Positive feedback, however, may be encouraging to the speaker, so that he continues to engage in social interaction and as a result grows further acquainted with the norms of the speech community. A positive feedback cycle develops, and the speaker participates in the social community with greater frequency. This positive feedback cycle suggests an explanation for participation inequality in Usenet.

The findings of Whittaker et al. (1998) demonstrate that a very small portion of participants in Usenet (2.9%) write a disproportionately large percentage (25%) of the messages posted. Additionally, they found that 27% of messages are from one-time posters. Their study, based on data collected in late 1996, analyzed over two million messages posted in five 500 newsgroups over a six month period. They consider the proportion of repeat posters to be a measure of familiarity of the newsgroup participants, although they do not say who they believe to be familiar to whom, the repeat posters to the group as a whole, or the group to the repeat posters. They observe that participation inequality is not necessarily bad; some participants may be satisfied by participating infrequently, and only in regard to particular topics of interest. This observation underscores the particularly important notion that participation frequency not only varies widely, but also has strong impact on the role the participant plays in the group. The infrequent poster has presumably made a less significant contribution of time and energy to the group and has made less of an impact. The infrequent poster is therefore less likely to have gained any reputation at all, much less one of prestige, and can therefore have only a relatively less integral role. In large number, however, occasional posters may have a strong influence on their social landscape.

Instead of relying on a shared speech register, Whittaker et al. (1998) consider the “common ground” approach to be the most important lens through which to study communities online. The common ground approach relies on the notion that, if participants in a conversation hope to communicate with one another, then they must have some conversational topic about which they both have some knowledge. This idea is the foundation upon which theories of social differentiation and interaction rely – people are more likely to interact with people with similar knowledge than with those with dissimilar knowledge. This idea, long known by cocktail-partygoers, forms the basis for computational models of socialization, in which social interaction can be modeled solely by the simple exchange of facts (Mark 1998).

Supported by this theory, Whittaker et al. (1998) explore how individuals participating in Usenet are able to establish common ground. They cite the use of FAQs, or lists of Frequently Asked Questions, as one way of establishing common ground. A FAQ list is a tool some groups use to prevent redundant questions from disturbing the group. The intent is that newcomers will read the FAQ before asking a question, so that they will see the answers to the frequently-asked ones, and will not disrupt the group by asking one. This has limited success; a newcomer will likely have not read the FAQ (he may not even know it exists) and will ask a basic or redundant question anyway. The FAQ is useful because regular users can point the newcomer to it when such a circumstance arises.

A shortcoming of the analysis of Whittaker et al. is that it conflates common ground with institutional memory. For example, a newsgroup that focuses on highway transportation might have had a lengthy discussion on Boston’s Big Dig in the past year. Such a conversation might have allowed the community to exhaust its interest in that particular highway, such that nobody who participates with regularity has any desire to discuss it further. When a new user asks a related question a short time later, he may not understand why the group is reluctant to talk extensively about it. The newcomer’s post may appear redundant to the group, but to the poster himself, the message is innocuous and the response unfriendly. Listing this topic in the FAQ may prevent this situation, but it is not because common ground has been established. The poster already shares a great deal of common ground with the group (i.e. knowledge of highways), but what he lacks is the institutional knowledge necessary for positive social interaction.

The correlation between common ground and posting frequency does not necessarily apply to every individual poster. The infrequent poster may not be a marginal member of the community, and the verbose poster may in fact share very little with the community. These two factors – common ground and participation frequency – can be used to conceptually divide the social landscape into people who possess one, both or neither of these traits, each of the four groups defining a type or collection of types of participants. This method will be employed and these types described in Chapter ‎4.
Common ground and communicative competence are necessary conditions for sustained, successful social interaction, but neither alone is sufficient. The two methods of understanding social interaction are similar – both describe norms of how to talk about things, using successful transmission of information as the primary metric. The difference lies in the explanation; the common ground model explains the successful interaction as a result of the quality of the information, whereas communicative competence assigns responsibility to the quality of the transmission. A synthetic understanding of communication, successful or otherwise, should then rely on both metrics, with an understanding that not only must the information conveyed be appropriate, it must be conveyed in the appropriate way. While much of communicative behavior is not done self-consciously, occasionally an individual will find it useful to communicate in a self-conscious way in order to achieve a desired effect. This sort of communication, too, requires that the individual exhibit communicative competence.

Sociologist Erving Goffman’s (1959:2) work underscores the observation that self-presentation is a vital part of communication. He distinguishes between expressions given and expressions given off. Expressions given are those actions which are deliberate and intended to make a particular impression on others. Expressions given off, on the other hand, are the impressions that are actually made. From the perspective of the actor, the ideal social interaction is one in which he makes the impression he intends to. That is, the expressions given off are exactly those that are given. A person displaying high communicative competence will have the knowledge to speak and act in a way that achieves his optimal social result.

3.4 Roles and Strategies
Jones and Pittman (1982) suggest that individuals use a general set of social strategies in negotiating what Goffman (1969) would call a “strategic interaction.” They propose five strategies: ingratiation, intimidation, self-promotion, exemplification and supplication. These strategies are not exclusive; an individual might adopt any combination of these strategies in the process of defining his or her character. The intimidator stresses his ability to cause harm to his intended victim, whereas the supplicator pleads incompetence and relies on the good nature of others to get his way. The ingratiator seeks to acquire the trait of simple likeability. The exemplifier seeks to gain respect by setting himself up as a role model and self-promoter tries to earn admiration by possessing enviable qualities. Three of the strategies - ingratiation, self-promotion and exemplification - rely on the actor gaining a more positive image in the eyes of others. Being moral, smart and likeable, for example, are traits that bring others closer. Conversely, supplication and intimidation rely on the actor gaining a more negative image. Supplicators and intimidators attempt to get their way by proving themselves in need of help or by proving themselves a potent force that must be mollified.

The problem with intimidation is that it works best when the intimidator and intimidated share an involuntary relationship. Jones and Pittman use marriages, families and coworkers as examples in which, they explain, it is very difficult for one to break a relationship without great difficulty and personal cost. Ordinarily, social ties in Usenet communities are not difficult to break; one has simply to stop reading and participating in the group. Unlike fighting with your next-door neighbor, a case in which one must physically move away, controversy in Usenet can be settled by abrupt termination of the discourse, when participants abandon either the particular conversation thread or the newsgroup as a whole. Goffman (1969) observes that some interactions carry a problem of resolve; how determined is each participant to continue participating regardless of the cost to himself? Goffman's example is the board game; if one is losing at Monopoly or Trivial Pursuit, he can always find an excuse - a headache, the need to wake early the next day - to end the game. A problem with playing games, Goffman observes, is that the players do not always have a real commitment. The same could be said of Usenet; it is easy to abandon a newsgroup in which one feels he is 'losing.'

This state of affairs both benefits and foils the potential intimidator. Individuals seeking to intimidate, perhaps through inflammatory or aggressive speech, can do so at will. When challenged, they can move on to find other groups with participants more susceptible to intimidation. This may continue repeatedly until the intimidator finds someone to victimize. How can a Usenet intimidator cause harm? Diane Wilson (2001), a longtime participant in soc.support newsgroups, likens people who use intimidation and anger in newsgroups to abusers, observing that they are ``filled with self-hate'' and their self-hate is directed outwards. In a group of people seeking empathy and support, an intimidator can cause a great deal of psychological pain. Wilson, an observer-participant in newsgroups frequented by individuals in crisis and in need of support, is in a unique position. She observes that if you ignore such intimidators, they “are left with their own venom.” She advises her fellow participants in much the same way parents advise children teased at school: if you ignore them, they'll go away. This advice fits perfectly with Jones and Pittman's analysis of how to foil an intimidator: if a potential intimidator cannot lure others into a relationship, then he cannot intimidate. The Usenet “killfile” is a tool that enables each user to selectively ignore others' postings. By simply adding a user's email address to one's killfile, one will no longer encounter posts from that user. Donath (1998) remarks that this is like “the ostrich putting its head in the sand,” but it is also a way of forcing oneself to ignore posts that would otherwise encourage an angry response.

Intimidators abound in Usenet. Popular terms for various sorts of 'bad guys' in Usenet include “flamer,” “troll,” “net-loon” (or “net-kook”). The definitions vary as widely as the people they describe. It seems that they are not exclusive of one another; it is certainly possible to exhibit the behavior of a troll and a flamer, a net-loon and a flamer, etc. Though they vary in their approaches and styles, these four names describe individuals who share a common goal of insulting and hurting other Usenet users and disrupting community. Irritating and loathsome as these types may be, some of them can have long-term positive effects in the groups that they invade. Some of these types have become popularized in the electronic community. Most users know the existence of “trolls,” even if their understanding of what a “troll” is varies. As Holland and Skinner’s work demonstrates, overtly recognized cultural models can be useful in contextualizing social interaction.

3.5 Roles and Expectations
Expectation plays a large part in facilitating social interaction. Dorothy Holland and Debra Skinner (1987) observe that talk about gender types among college women makes significant use of categories when describing their fellow students, especially college men. These categories, which contain a great deal of implicit cultural knowledge, provide a framework for interaction by giving people an a priori sense of the personality of the people described by them. Their work is a key demonstration of the usefulness of social typologies.

Holland and Skinner’s (1987) “Prestige and Intimacy” uses the term cultural model to describe the categories that their informants, college women, use to describe college men, and vice-versa. Terms like “jock,” “nerd,” and “wimp” describe men who think they are physically powerful, are “terribly insecure,” or are otherwise “strange, socially unacceptable,” respectively (81). Likewise, men described women with terms like “princess,” “bitch,” and airhead.” Holland and Skinner explain that these types are not important in and of themselves, but rather are important for people who seek to know how to interact with others. Holland and Skinner’s argument is an explicit rejection of “dictionary-type definitions as a means of describing [cultural models]” in favor of a cognitive model that explains how classifications are organized within the mental lexicon according to social meaning. Holland and Skinner hypothesize that information about how to interact with others – what type of behavior to expect from them – in fact forms the basis for the womens’ categorization of the role types in their midst. They observe that the “social dramas” in which the men appear are more important than their individual traits when classifying them. For example, a girl pursuing a relationship with a boy must be able to understand his reputation as described by her female friends, so that she is able use this information to interact with him in the way most advantageous to her.

In addition to providing knowledge of how to interact with another person, as Holland and Skinner demonstrate, a social type also serves as a framework for understanding that person’s behavior. The way a person is perceived by others strongly affects they way his words are interpreted. If two very different people were to make the same controversial statement, it could be interpreted as ironic when made by one but derogatory and offensive by the other, if the first had a reputation of being reasonable and fair-minded and the second had a reputation marked by bigotry and ignorance. These hypothetical instances suggest the importance of reputation and context in interpretation of new behavior; each new action by an individual in a social interaction will be interpreted in light of interpretations of past interactions and knowledge of the individuals’ social roles.

Holland and Skinner analyzed the womens’ and mens’ descriptions of one another and distilled them into a small number of discrete dimensions, using a process called multidimensional scaling. The descriptions of men were categorized on two axes: how likeable the men are and how much they emphasize male/female differences. The women, in turn, were categorized according to their likeability and their status as possessions. Like Holland and Skinner, I use a variety of measures to classify social types. My analysis uses both communicative and strategic metrics to classify participants. Communicative metrics include participation frequency, communicative competence and common ground, while for strategic metrics I use Jones and Pittman’s five strategies.

Using information about cultural models or social types to interpret another’s behavior is a natural, common occurrence in day-to-day interaction. Even knowledge about non-overtly recognized social types becomes a framework for interacting with others. Indeed, Holland and Skinner observe that much of the knowledge about social types is taken for granted during social interaction. Like in the social world of Holland and Skinner’s college women, the social world of Usenet has a set of social roles that are neither formal nor overt.

3.6 Defining the Social Role
A variety of definitions for real-world social roles have been posited, but they are not adequate descriptions for social roles in electronic communities. Sociologist Louis Zurcher (1983), for example, follows Jerold Heiss’ definition of social role as a set of “behavioral expectations for what a person should do, when occupying a position in a specific social setting (223).” This is consonant with Holland and Skinner’s observations about role use and expectation. However, Zurcher’s example involves formal roles in well-defined environments, e.g. officer in the military or passenger on an airplane, and roles adopted by all members of a group, e.g. poker player in a poker game. Unlike these roles, those in electronic communities in Usenet refer neither to formal structures like the military, nor to group roles like poker game participant which apply equally to everyone.

The group role (e.g. Zurcher’s poker player) is somewhat more plausible than the structural role in the context of Usenet, to the extent that everyone is in the role of social participant, to some greater or lesser degree, and they are each part of an environment that has a set of social norms. However, these broad roles seem to define the role as “community member” or “participant.” I take such a status for granted and delve deeper into what makes members of the same community different from one another, rather than what makes them the same.

The prescriptive hierarchal role (e.g. military officer) is non-existent in Usenet; from a technological standpoint, there is a complete lack of formal authority.[14] A definition like Heiss’, which is based on what participants should do works well in a system in which cooperation is vital, like in Zurcher’s military or disaster-relief crew examples. A purely social community like a Usenet newsgroup, based on topical conversation, may seek cooperation and concord as an ideal, but it is understood that social conflict is occasional and inevitable and that it is perhaps annoying but certainly not life-threatening. Because of the lower stakes of the interaction, a definition of social role that includes prescribed behavior is a poor fit for Usenet. The fact that Zurcher focuses on the development of autonomy within role enactment further emphasizes that his concept of a role is something that is imposed on the actor rather than defined by him.

Instead, a definition that includes behavioral expectation based on both past behavior and tacit understanding of social position is necessary. Erving Goffman’s (1959:16) definition is useful; a social role in his conception is “the enactment of rights and duties attached to a given status.” This assessment has room for both formal and informal status, and acknowledges the relationship between status and behavior (i.e. rights and duties). That is, a role is the sum of what one does and what he can do.

The social roles proposed in Chapter ‎4 are almost completely non-overt. That is, Celebrities, Elders, Ranters, et al. are my own terms, formulated to describe categories of behavior. Other roles, like the Troll and Newbie, are recognized within the Useet community and are used in conversation, much like Holland and Skinner’s role types. Though there are important differences between the overt and non-overt types, all are discussed here within the context of communicative and strategic metrics.

4. A Typology of Roles
Before a typology of roles can be developed, two important ideas must be considered. First, it is important to develop a motivation for such a typology. That is, one must ask, what are the puzzles it will be used to solve? Second, it is important to explicate the theories driving the typology. Hymes’ observation about sociolinguistic taxonomy addresses the first of these considerations:

[S]ociolinguistic taxonomy is not an end in itself, any more than is language classification. A taxonomy is not in itself a theory or explanation, though it may conceal or suggest one. . . The work of taxonomy is a necessary part of progress toward models of sociolinguistic description, formulation of universal sets of features and relations, and explanatory theories.

(Hymes 1972:43)

The models of sociolinguistic description that this typology seeks to construct are, first and foremost, related to behavior in Usenet. The technical affordances and limitations entailed by Usenet’s interface alter the means of social interaction in a way that such a typology cannot apply directly to other environments, online or offline. This is not to say that social roles in Usenet have nothing to say about social behavior in everyday life. On the contrary, the study of linguistic behavior in one mediated context may shed light on linguistic behavior in some other medium. Whether through a computer screen or through our eyes, ears and mouth, communication takes place in some medium – never in a vacuum. While this typology outwardly identifies roles specific to Usenet, these roles are only manifestations of linguistic behavior passed through a particular medium, and the idea that linguistic universals are underlyingly at work should not escape the reader. With this typology as a tool, it will be possible to say something about them as well.

The theoretical grounding for the typology is explicated in Chapter ‎3, in the work of Hymes, Whittaker et al., Holland and Skinner, and Jones and Pittman. The communicative and strategic metrics formulated therein are the means by which distinctions amog groups of individuals are drawn.

4.1 Mapping the Social Landscape
As a working definition, I proposed in Chapter ‎3 that a social role is the sum of what a person does and what he can do. This is an attempt to integrate the concept of the “social drama” that Holland and Skinner (1987:87) employ in identifying types of en and Goffman’s (1959:16) focus on the “rights and duties attached to a given status.” Like the “social drama,” a Usenet conversation is an environment where social interaction is on display. If male types are categorized according to the types of social interactions women observe them in, then Usenet participants can be categorized according to the conversation threads they participate in and how they behave in them. Their behavior reflects both what they do and what they are allowed to do. Celebrities, for example, may feel similar in the reader’s eyes by virtue of the fact that they share the trait of frequent participation; their status is dependent on this trait. Accompanying this status, they have the privilege of speaking on behalf of the group.

The communicative and strategic metrics proposed in Chapter ‎3 are not meant to be used for binary distinctions. Participants should not be thought of as competent or incompetent, but rather should be considered to possess a relative degree of competencecompared to other members of the community. The argument is the same for all other communicative and strategic metrics employed here. In keeping with the intentional relativity of these traits, while the role categories are determined by having or lacking various traits, this is not a solely feature-based typology. Indeed, Komatsu (1992) notes that, since the 1950’s, relatively few cognitive psychologists have supported such a model and effectively none support this “classical” model today. Instead, these metrics are used to explain the “social dramas” that are used to classify the participants.

No roles here should be taken as necessarily exclusive, except perhaps those pairs that would require a participant to have both high and low values for a particular trait simultaneously. For example, an Elder may be, and often is, considered a Celebrity. In the right community, a Flamer can be both an Elder and a Celebrity. It is very easy, in this framework, to behave in a way indicative of two or more roles simultaneously. Additionally, it is possible to enact different roles at different times or in different contexts. A longtime participant in a newsgroup may be an Elder there. Lacking a similar reputation, however, he will be just another participant in different newsgroup. Lastly, the roles a user enacts may change over time. Even Elders were Newbies once.

4.2 The Newbie
The Newbie, or new user, is one of the most widely-recognized overt roles in Usenet. The prototypical Newbie generally has little communicative competence and may have little common ground with the group. Common ground, measured by knowledge in a particular area, is often lacking most in Newbies in technology-related groups in the comp hierarchy, where technological know-how is highly prestigious. Communicative competence is often lacking most in Newbies in more socially oriented groups, where ritualized yet subtle behavior may be in greater use but invisible to the new participant. To make up for these shortcomings, many Newbies enact a role of supplication, pleading ignorance and relying on the kindness of others to ease his socialization process.

To address the perceived problem of Newbies’ lack of communicative competence, many newsgroups with FAQs will advise new participants to “lurk” for some time before participating. Lurking is the practice of reading the newsgroup’s conversation without participating oneself. For the new participant, time spent reading before participating serves as a socialization period, intended to teach him or her, by example, the expectations of the newsgroup. For some groups, this acculturation period is very important, because posts lacking competence are highly stigmatized. Posts that are not about a sometimes rigidly defined topic are grounds for public chastising. For example, in the FAQ for rec.arts.bodyart, it is stated that only certain types of body decoration may be discussed:

(6) If you would like to discuss bodypainting, please post to alt.art.bodypainting. The exception to this is the discussion about henna, which stays on your skin for a long time. Some people assume we discuss bodypainting because "bodyart" *can* include bodypainting by some . . . Since rec.arts.permanent-bodyart is an overlong and clumsy title, the founders of this group chose the simpler newsgroup title.

However, for some groups, like alt.support.stop-smoking, this is not the case:

(7) In most newsgroups, you are requested to "lurk" (read but not post) around a newsgroup for a few days before joining the conversation. By lurking, you get a feel for a group's character, its conventions, and the people who post to it. You don't have to do this in AS3![15] Jump right in and post as often and as much as you need to, BUT if you're thinking of posting to an ongoing argument or an angry thread which seems to focus on personalities and not on smoking/quitting issues, it might be a good idea to hold that post aside for a day or two, before you join the fray.

The FAQ in alt.support.stop-smoking advises Newbies to “jump right in” to the conversation. In a group that highly values topical conversation, a suggestion like this could be a dangerous invitation for annoyance. However, alt.support.stop-smoking routinely has conversations that are off-topic (i.e. the topic of quitting smoking) by design. For this group, topicality is less important than posting material that is interesting to the participants. Arguably, however, this requires a more sophisticated understanding of the group’s tastes. If alt.support.stop-smoking were simply about stopping smoking, then it would be relatively easy for participants to decide what is and is not acceptable. On the other hand, when all possible topics are up for consideration, a deeper understanding of the newsgroup’s collective likes and dislikes is necessary. This may not be a serious problem in practice, because new participants may also be likely to be cautious about the material they post.

Sometimes Newbies can be cautious to the point of excess. To make up for a potential lack of competence or common ground, a Newbie may resort to excessive supplication. This may include unnecessary apology or employing hedges:

(8) From: "just me" <No@ThankYou.com>
Newsgroups: misc.kids
Subject: Re: The problem of Mother Keesha solved

"Douglas and Jennie Jackson" <dandj_spamblowsit@alphalink.com.au> wrote
> Pardon my intrusion in the discussions about and from Mother Keesha but it seems to this
> newbie that the best way to get rid of her is just to have everyone ignore her posts and then
> basicallly, if we are lucky, she'll just go as no one will be listening and responding to her asinine
> posts.
> Jennie Jackson

You're no newbie! Newbies don't know that stuff! You are totally right: ignore the troll and it goes away.
-Aula

This pair of messages exemplifies a Newbie’s post and positive reinforcement from a more seasoned participant.[16] In response to a disruptive participant, Jennie Jackson says to ignore the disruption. She does this in a way that is rather self-effacing; she apologizes for participating (“Pardon my intrusion”) as though she did not belong and employs a hedge (“but it seems to this newbie”) while calling herself a Newbie, which is generally a low-prestige role. Aula acknowledges that Jennie is demonstrating knowledge of a common piece of advice on Usenet: if people are being disruptive, the best policy is to ignore them. She is paying Jennie a compliment and giving her positive reinforcement by saying she’s “no newbie” (though Aula likely does know that Jennie is, in fact, new). Zurcher (1983:139) suggests that seasoned participants in social environments will often offer positive reinforcement in response to a new person learning and enacting acceptable group behavior.

Not all Newbies are created equal. Being new to Usenet in general is very different from being new to a particular group. Like the rec.arts.bodyart FAQ, many FAQs will warn Newbies not to post empty messages or post the same message repeatedly due to unfamiliarity with the technology. This is a mistake only a Newbie new to Usenet would make. Though it is an effect of the technology, such a mistake is a competence issue as well – being able to communicate competently requires being fluent in the medium as well as the content. Content-level deviant behavior, like posting a classified ad or personals ad style message, are mistakes that either type of Newbie could commit. Yet another sort of social violation, like failing to quote material to which one is responding or typing in all capital letters, is a hybrid of the two – the Newbie might not have the technical knowledge to perform the socially accepted behavior.

The use of acronyms is yet another social barrier for Newbies. Recall from Chapter ‎3.1 that the acronym “BOYC” (“Beverage Of Your Choice) is used in alt.callahans, a virtual bar. For any person who is new to that newsgroup, an acronym such as this willbe opaque at first. Since it may not be possible to learn the meaning of an acronym from context while lurking, some new participants may post a message asking that it be explained.

(9) From: "box412" <nospam-troy_lea@hotmail.com>
Newsgroups: aus.motorcycles
Subject: What does squid mean

Might be a silly question but what does squid mean, is it an acronymn or is it as simple as it sounds, squished all over the road looking like a squid? I saw lots of them out on Sunday, quite a few of the thong variety I must add.

This person is undoubtedly new to the aus.motorcycles newsgroup, though perhaps not to Usenet.[17] The fact that acronyms arise in newsgroups is further evidence that social communities are in fact thriving within them, and underscores the difference between Newbies to particular groups and Newbies to Usenet in general.

The single most prevalent behavior among Newbies is question asking. New users have not been part of the community for very long and are often therefore less knowledgeable about the topic of the group. That is, they have less common ground. This is especially common in comp groups. It is so common that conventional forms of asking have arisen.

(10) From: Cayd Meier <cayd@attbi.com>
Subject: Beginner's Question
Newsgroup: comp.lang.c++

I'm going through the book “Teach Yourself C++”. There's some code listed in the book that will not compile without errors on Visual Studio. However, if I make a small change to the code it compiles fine.
Here's the first few lines of code as it is in the book: . . . [18]
No errors occur. Right now, it seems as if I can't trust the book. But I'm sure it's a little more complicated than that. Any help would be greatly appreciated.

(11) From: "Denis Gurchenkov" <dgurchenkov@excelsior-usa.com>
Newsgroup: comp.compilers[19]
Subject: looking for C pre-processor

Could somebody point me to a C pre-processor that is provided with source code? . . . I would be thankful for any information, including "have never seen such a thing" messages ;-)
P.S. I'm aware of GCC and lcc.

The first post, from Cayd, has a subject, “Beginner’s Question.” The author explicitly acknowledges his relatively low status in the community. His status is doubly low, because he is a new user (that is, he has never posted here before; whether and how long he had been lurking is of course unknown) and an inexperienced one (the book he is using is an introductory text). Subjects like “Beginner’s Question” are used with great frequency. Though it is a supplication, an attempt to take advantage of one’s low status, it belies a partial lack of competence. The experienced users in informational newsgroups, especially in the comp hierarchy, answer many questions from less experienced or knowledgeable participants; a subject like “Beginner’s Question” is uninformative and it may suggest that the post’s content is likewise uninformative, turning potential helpers away. Cayd begins his post by explaining the steps he has taken, showing that he has made an effort to fix his problem. Experienced users are unwilling and unable to offer advice if the poster does not do any work to resolve the issue himself first. The second post’s author includes information about what he already knows (i.e. GCC and lcc), showing that he too has done some background work. Not only does this show that the author has invested himself in finding a solution, it gives the potential helpers more information to work with. After an explanation of the problem, question/help posts virtually always end with pleas to the group for aid. Conventional phrases like “Any help would be greatly appreciated” and “I would be thankful for any information” are quite common and establish the poster as a supplicator; Indeed, the request for information, by its very nature, makes the poster a low-status participant, whose well-being depends on others’ adherence to the social norm of responsibility to help others (Jones and Pittman 1982:250).

The people who answer Newbies’ questions act as role models for them and participate in the group on a daily basis, providing useful information and conversation-stimulating ideas. These high prestige individuals are the Celebrities. In the example below, Peppe, a Celebrity in the comp.compilers newsgroup, responds to a post that demonstrates the poster’s extreme lack of competence.

(12) From: c928400@student.dtu.dk (Peppe)
Newsgroups: comp.editors
Subject: Re: Compilation of MS projects in in VIM

Thus wrote SigTom <thomas.wirsch@sig-group.com>
> In order to do this I suppose I have to call a tool like make + parameter.
> Is there anyboady who knows the exactly command ?

You leave us very little clue as to which program you are actually using. More info would be nice.
Peppe

The participant “SigTom” does not use any of the strategies that Cayd and Denis use above. SigTom fails to convey the idea that he is asking for assistance and that any assistance he is given is predicated on a community norm of helping. In short, he is presenting himself as ungrateful, as though he expects an answer to be provided. SigTom does not take Peppe’s advice and does not provide more information. It appears that, like many Newbies, SigTom was there only for answers, not for the benefits of a like-minded community. Many Newbies will participate in a newsgroup, initially seeking only answers to a question, but will find that being part of a community of people with a similar interest provides social benefits and will therefore stay and grow to become part of the community.

Peppe, a Celebrity in the group, attempts to help SigTom despite all of the shortcomings of his initial message (a message similarly not following conventions may simply be ignored). Peppe’s attempt to not only help SigTom but to educate him on how to ask a question as well is generous. It also reinforces the notion that people seeking advice must first try to help themselves, which is beneficial to the Celebrities of the community.

4.3 The Celebrity
Recall Labov’s (1997) observation quoted in the beginning of Chapter ‎2: “In every neighborhood, you need to know the people who are the central figures so that you can understand how society works and who influences who.” The Celebrity is the prototypicl central figure. Prototypical Celebrities are prolific posters who spend a great deal of time and energy contributing to their newsgroup’s community. Because they post so often, it would be virtually impossible for the casual lurker to read the newsgroup without encountering posts by many Celebrities. By virtue of the volume of the Celebrities’ posts, the casual participants’ impression of them will be their impression of the newsgroup. To put it succinctly, it is the Celebrity who defines what the community is.

Celebrities put participation inequality into action; when Whittaker et al. (1998) talk about the small percentage of participants who generate the largest volume of posts, they are talking about the Celebrities. Consider the example of alt.computer.consultants. In approximately one month, 137 individuals posted messages to the newsgroup. The top twenty most frequent participants (14% of the total number of participants) authored 75% of the newsgroup’s messages. As the following table shows, even among the top twenty posters, there is an order of magnitude variation in posting frequency; the most frequent poster authored over twenty times the number of posts of the twentieth most frequent: insurancenj@aol.com posted 515 messages and CTOCIO@e-janco.com posted 24.


AUTHOR
POSTS



insurancenj@aol.com
515



nietszche@large.com
383



alexy@mindhelicalwire.com
255



lizs2000@swbell.net
205



arthures@magpage.com
174



NorForJunkEmail@directinternet11.com1
170



LESLIE@JRLVAX.HOUSTON.RR.COM
149



dontspam@nospam.com
96



cscharff@mail-resources.no.spam.pls.com
73



joec@aracnet.com
43



test01a@ziplip.com
43



David.M.Fabian@sbcglobal.net
41



antispam9@attbi.com
39



cantunes@cox.SPAMTRAP.net
36



nomoreh1b@yahoo.com
36



ElmerFudd@yahoo.com
30



clipper@prostart.net
28



tbartkus@hotmail.com
28



usenet@mpreston.demon.co.uk
27



CTOCIO@e-janco.com
24



Table 1. Top Twenty Posters in alt.computer.consultants

The three most prolific participants’ posts comprised almost 50% of the top twenty participants’ posts. More importantly, these three together posted over 35% of the newsgroup’s overall messages. With only three people authoring over a third of all posts in the newsgroup, these three have a profound impact on what conversation in the newsgroup is like.

Simply posting with great frequency does not make someone a Celebrity. Celebrities are known for more than just the volume of their messages; they typically display high communicative competence and share much common ground with the community. In both cases, it is because they are the ones who typically define what the prestige speech forms are and what knowledge and beliefs are important to the group. A poster who sends only news articles or advertisements to a newsgroup is one example of a high-frequency poster who serves no social role at all, just a social function that may be in support of or detrimental to community. In fact, such a poster may be a computer program.

Competence is a desirable trait; it is demonstrated through posting messages that are socially acceptable by community standards. Acceptability varies by context, however; rec.gambling.blackjack requires that arguments be backed up significantly by statistics. alt.folklore.urban, a newsgroup devoted to exploring and debunking urban legends, requires solid corroborating evidence in the form of names, dates, facts and figures and, where possible, primary source document references. In contrast to these two groups, where acceptability requires evidence, alt.anagrams participants consider only posts that are polite and praise giving to be acceptable. A typical message thread consists of an initial post in which the author anagrams on a news headline or similar piece of text. Each successive replier will compliment the previous anagrammer before posting his own anagram of the text. Though not all participants always praise the previous anagram, it would likely be socially unacceptable to routinely not provide praise to the other anagrammers.

Though a Celebrity shares a great deal of common ground with the community, this does not mean that he is always uncontroversial. A Celebrity’s views may be in the minority, so long as he presents them in a way that is deemed acceptable. In many newsgroups, high-prestige participants will sometimes engage in heated debate, though each shares much common ground with the community. Disagreement is acceptable, but the terms of the argument must be in keeping with the community’s standards. This could involve statistics or names and dates, as in rec.gambling.blackjack and alt.folklore.urban, respectively. When a participant cannot employ methods of discussion respected by the community, his competence is in question. Such a participant may be seen as a Ranter, a role described in Chapter ‎4.8.
The information shared by the Celebrity may be useful, thought provoking or entertaining; in any case, the community benefits from the Celebrity’s presence. The reputation that prolific, helpful posters develop is a high-status one, even if not all users recognize it immediately:

(13) From: dsew@packrat.aml.arizona.edu (David Sewell)
Subject: RFD: comp.software.sven
Newsgroups: comp.mail.elm,comp.editors

A year or so ago I engaged in a brief flameware with Sven Guckes over the issue of bouncing e-mail from spammers. Sven being the dominant voice on comp.mail.elm, I was a bit irritated that he was always so sure he was right, and I was even more irritated that he won his argument with me.
However, it has become increasingly clear to me that Sven really *is* right most of the time, at least about the Right Software To Use.

This poster is referring to Sven, a Celebrity in the comp.mail.elm newsgroup. As the maintainer of a variety of popular software programs, Sven is very knowledgeable in the areas of the newsgroups he frequents. By consistently providing useful advice, he has grown to be a recognized voice of authority in the comp.mail.elm community. A fixture of the community, a Celebrity is often mentioned by name in the posts of other participants, as Sven is here.

Why do people spend time participating in a community, devoting their energy to their respective newsgroups? Some participants spend years developing a reputation as an expert and take on de facto responsibility for the community. Chen and Gaines (1996) hypothesize that positive contribution to a social group has two results. The contributor gains a positive self-image and social power. A positive self-image may mean that the poster sees himself as nice, helpful or altruistic from helping others. He may also see himself as smarter or better than them because he possesses knowledge they do not. The second compelling result of contribution is social power, which is exhibited frequently in Usenet. By contributing frequently and in a positive way, one can establish oneself as a voice of authority, further leading to a development of positive self-image.

One hallmark of being a Celebrity is being “famous.” Celebrities are public figures that others know about and talk about. In the example above, Sven is not part of this conversation, yet his high standing in the community makes him a subject of conversation of others. One must be important and well-known to be a subject for others; it is impossible to publicly discuss someone who nobody knows.

The authority of the Celebrity extends to defining community standards, which is done largely as a by-product of enacting behavior that puts them into practice; by collectively behaving in a way they find acceptable, Celebrities set an example for others. They combine strategies of exemplification and self-promotion; not so much of themselves, but of their way of being. With an investment in the community made up over months or years of participation, Celebrities have an interest in seeing new participants behave in a way they find acceptable, lest they radically change the community or “steal” it from those who were there first. Participants who do not show respect for and deference to Celebrities will either learn to, as in the example above, or will encounter harsh rebukes from the other participants.

4.4 The Elder
The Elder is a special type of Celebrity, one who has been around for an exceptionally long time and is proud of it. He may participate less frequently than others, but the Elder remembers what the community “used to be like,” before many of the current participants arrived and even before the time of some present Celebrities.

(14) From: "Ralph Stricker" <bjpro21@ix.netcom.com>
Newsgroups: rec.gambling.blackjack
Subject: Re: Where's the old crowd?

I rarely post as the same questions are asked ad nauseam. I understand the problem as new people are logging on and have questions just as many of us years ago had, when we first became involved with BJ. I also had to take some time off from playing BJ due to my health. I have just been given the OK by my doctors to start playing again on a limited basis.

Ralph, along with a handful of other participants, used to be very active in rec.gambling.blackjack. Since that time, however, they have seen the community change before them. An influx of advertising and Newbies’ question-asking have made the group unpalatable for these Celebrity-Elders. In the rest of this conversation, Ralph’s compatriots likewise lament hearing the same questions repeatedly.

Like many societies, Usenet newsgroups tend to respect their elders. In the case of Usenet, though, the elderly may be in their thirties, or forties, or even twenties. The Usenet elders are defined not so much by their age, but by their relatively extensive knowledge and experience of the community history. In a geographically defined community, this may in fact require being seventy or eighty or ninety years old. In Usenet, where the communities have been around for, at most, twenty years, the standard for being an elder is far lower. Being an Elder is a prestigious role, sharing many boundary protection desires and responsibilities with the Celebrity.

Like Newbies, Elders have conventional speech styles. Only Elders, for example, can use words like “always” and “usually” because only they have the context to support such a claim:

(15) From: "Martina Livingston" <martina@mlhdesigns.com>
Newsgroups: alt.fairs.renaissance
Subject: Re: Escondido Ren Faire Questions.

"Rachal" <rachalbeene@hotmail.com> wrote:
> We are traveling down from above Reno to attend the Escondido Renaissance Faire as vendors.
> One thing that we are not clear on is the camping situation. Is there an onsite Actor's/Vendor
> Camp, we are only wandering vendors, so we won't have a booth to camp at. This is a long way,
> and a bit of a gamble, we won't be able to get a hotel. Can anyone help out here that has attended
> before?

Usually folks pitch a tent wherever and take it down in the day... there's a good chance you can get a place to stay … Count on a little rain.. It ALWAYS rains during this event, both spring and fall... not enough to stop the event, but enough to make it chilly and damp.

Martina Livingston is apparently a longtime member of the Renaissance fair community and an experienced participant in alt.fairs.renaissance, a newsgroup supporting the “Ren faire” community. Usenet Elders combine actual age and real-life experience with experience within the newsgroup, allowing them to be quite helpful to others.

Elders take pride in demonstrating their age, either through sharing information from the past or discussing things that require knowledge of events from long ago. In rec.motorcycles, motorcycle enthusiasts have popularized the faux-Old English term “olde pharte,” which they use to describe themselves in terms of experience: “I’m such an olde pharte, I’ve been riding Harleys since 1959.” For the rec.motorcycles crowd and others who have adopted it, this is a term of prestige. Likewise, one often expresses age through onomatopoeia; *creak* is an ironic play on the sound of joints creaking:[20]

(16) From: MG <janette_db@my-deja.com>
Newsgroups: alt.support.childfree
Subject: Re: OT - Hair Color

I've been afraid to use Punky Color again after that experience, although I've kinda gotten
too old *creak* to do Ronald McDonald hair anyway, sad to say.

4.5 The Lurker
As discussed in chapter ‎4.2, The Newbie, a FAQ will often suggest that new participants lurk for a time before participating. A feature of Usenet frequently observed as problematic is that it is impossible to know the size of the audience reading the nesgroup. Though it is trivial to calculate how many people are participating, no readership data is collected, thus making it impossible to know how many readers are lurking. Together, all these lurkers form an Invisible Audience that is often assumed to be there, but nobody really knows.

Goffman (1959:106) observes that “barriers of perception” bound social interactions. If the people engaging in social interaction in Usenet cannot see the lurkers, then it follows that the lurkers exist outside of these barriers, and their perception of the lurkers may be inaccurate. It is unknowable whether participants overestimate or underestimate the size of the Invisible Audience; the question that can be asked is, do participants feel there are relatively many or relatively few lurkers reading their posts? While the answer remains unclear, what is clear is that the expectations that participants have are founded on past interaction within the group; though participants arguably feel that some lurker may respond to their posts, they seem to be more concerned with reactions from other participants. For example, people in a soc.support newsgroup community may tend to respond helpfully and supportively to the problems of their fellow participants. Therefore, a participant in the group contemplating posting his own personal problems will have some expectation that the responses to his problems will be likewise supportive. Recall Hackman’s (1992:204) observation that individuals learn quickly to read the social cues of the group in order to form “behavior-outcome expectancies.” By observing other participants respond to one another’s behavior, the new poster can formulate beliefs about the potential results of certain actions. Since “behavior-outcome expectancies” are based on past community behavior, the activity of lurkers should, theoretically, play no part in the expectancy formulation. However, awareness of lurkers is generally universal. Therefore, whether their potential presence affects expectancy formulation or not remains an open question.

Not all lurkers are created equal. They do not all lack direct influence in the community. In fact, some influential participants may post only infrequently, yet their voices carry a great deal of weight in the community when they do. A side effect of the Invisible Audience is that a formerly frequent-posting, high-prestige participant who takes an extended leave from the group will be as invisible as a lurker who has never posted. At times, a prolific poster may decide that he is too busy to keep up with the newsgroup due to factors in his personal life, and will take a break from posting to the group but will continue to read the conversations taking place. Given a long enough absence, he may return to find himself unknown and misunderstood among the people who joined the community in the interim:

(17) From: Rusty Martin (rustyblkjk@aol.comBATSPAM)
Subject: Re: Where's the old crowd?
Newsgroups: rec.gambling.blackjack
Date: 2002-10-30 11:29:24 PST

Like Ralph said, answering the same questions over and over across the years has become a bit tedious. So I mostly lurk, and I'll throw in a smartass remark now and again.
I think anyone who's discovered this group in the last year probably thinks I'm only a smart aleck troll rather than someone who once lived and breathed this game and this newsgroup.

Like Ralph, Rusty is an Elder in the rec.gambling.blackjack community who has grown dissatisfied with the changes in the community and has ceased participating frequently. The important difference is that Rusty notes that new participants will understand him differently from how longtime participants will. Their expectations of group behavior will have been formed in a context that does not include large quantities of posts from Rusty, Ralph and their friends. Rusty’s delurking emphasizes how group populations and styles can change over time. How these changes are related is discussed in Chapter ‎5.3.
4.6 The Flamer
The Flamer is known primarily for his aggression and intimidation. Even having no reputation in a community, a Flamer is identifiable from his first post. He does not seek to become a legitimate part of the community, so establishing communicative competence is not an issue for him. His key behavioral strategy is intimidation through very aggressive language, yelling and controversial speech.

The term “flame” can be either a noun or a transitive verb. A flame (noun) is an extremely hostile message consisting of hateful speech and aggressive language. To flame someone is to send someone such a message. A Flamer, then, is a person who habitually engages in flaming behavior. The Jargon File[21] cites “flaming asshole” as an etymology for this family of terms, originating at MIT (Raymond 2002). Other fire-related terms have arisen in connection with “flame.” For example, before saying something legitimately controversial, a poster might say “donning asbestos underwear” or “putting on flame-retardant suit” as though expecting to receive flames in response.

Flamers tend to travel from group to group, seeking victims with whom to pick fights. While not attempting to construct an argument per se, a Flamer will identify the subject or subjects that are contentious for this community and will make controversial statements about them. For example, a Flamer visited misc.kids, a parenting newsgroup, and wrote about the ugliness of babies, a topic parents would hardly find tolerable.

(18) From: redtyrel@my-deja.com (Red Tyrel)
Newsgroups: misc.kids
Subject: Kindly keep your babies out of my face!

Look moms (and dads), you little babies ARE NOT cute. Sorry to break the news, they are UGLY. So stop bringing them into the office so that fat old bags (who couldn't if they tried) can get up off their fat arses and say things "oh isn't that cute". YUCK. Look here: Not everyone likes your shitty little babies ok?
And if you plan to flame me.. fine.. here I have something for you;
PHHHHT... That came out of my arse and it smells better than that baby you have!

This particular Flamer has insulted Linux users in Linux advocacy newsgroups, pro-abortion activists in abortion newsgroups, and Blacks in many newsgroups across Usenet. Though a Troll or Flamer may have some interest in the topic of conversation, the topic itself is relatively unimportant, so long as it is something over which someone could be hurt. A Troll who takes the argument too seriously or personally could be considered a Ranter instead.

There is a distinction that must be made separating true Flamers from others who simply engage in Flaming behavior. The example given above is a true Flamer, who shares many similarities with the Troll (see next section). He is the character described to this point. The true Flamer is not part of a community; whether he is successful in angering people or not, he will likely move on to another group seeking more victims. This contrasts strongly with the community member who occasionally engages in flaming behavior. When a conversation becomes particularly heated, it can “descend into a flamewar,” which is a protracted argument that has ceased to be about the topic at hand and instead has become a volleying back and forth of hateful, angry speech. It is particularly useful to note that the verb “descend” is paired with “flamewar.” This emphasizes the stigma attached to flaming behavior. Because there is a stigma associated with flaming, excessive flaming behavior by a participant in a conversation will often be met with criticism from the participants who are more level-headed and less emotionally involved in the conversation at hand.

Though in general it is true that, for most social communities, excessive flaming is stigmatized, there exist communities for which this is not the rule. The community in the alt.flame newsgroup, for example, is focused entirely on flaming. Indeed, being able to write quality flames is highly prestigious. The conversations in alt.flame do have topics, but the topic seems to exist only as a vehicle for hurling abuse. The topic, which is irrelevant here, is different from the “communicative goal” (Dundes et al. 1972), which in this case is to display one’s prowess at inventing creative and cruel insults. Even a conversation about video games becomes a flaming exhibition in the second message of the conversation thread.

(19) From: skippy@hell-flame-wars.net (Skippy)
Newsgroups: alt.flame
Subject: Re: What PC games are you currently playing??

JamesKirk@Enterprise.com (Fox Mulder) wrote:
>I am enjoying the free ride mode in Mafia very much.
>Splinter Cell is really cool too.

Splinter Cell is old news. Completed it ages ago on the X-Box. Get some new games, you tight-fisted ringworm.

Goffman (1969) observes that “verbal jousting” like this is cyclical; each round of insults must be bigger and more insulting than the previous. It is important never to let your opponent have the last word. Dundes et al. (1972) writes of Turkish boys’ “dueling rhymes,” in which a mediocre response is an implicit admission of defeat. The case is the same here; participants who have failed in the past to top their opponents develop reputations for ineffectiveness and as a result have lower social status:

(20) From: Hogarth <hogarth@address-pending.com>
Newsgroups: alt.flame
Subject: Re: Tonight's agenda.......

Diaper Boy! You finally summoned up the balls to post in alt.flame again. How's the incontinence these days, anyway?

The person Hogarth is flaming in the above example now is required to either respond with another insult or suffer the social consequences. The nickname “Diaper Boy” seems to have been attached to the person at whom this attack was aimed. The content of the message suggests that he has a history of not responding with a better insult and is now suffering further insults because of it.

The flaming taking place in alt.flame is a ritual insult, a kind of insult that Labov (1972:335) distinguishes from a personal insult. The former is about social play, whereas the latter is a serious attack. Competence in this community requires being able to engage in the former effectively, where effectiveness is measured in ability to outdo one’s opponent, leaving him or her unable to produce a better retort. This is most often in the form of long dyadic interactions that the participants call “duels.” Ritual insult is a kind of social performance; Bauman and Briggs (1990; qtd. in Baym 1995) observe that performance allows the community to evaluate each other’s skill. Ritual insult seems to be enjoyable because it is informative. It provides opportunity to perform for one’s friends and display one’s knowledge of socially prestigious behavior, which can lead to greater respect within the social group. This sort of flaming is done for the purposes of prestige within the alt.flame community. Its participants do not share the same goals as other Flamers, and so do not practice the same behavior of traveling from group to group seeking victims; they stay in their communities, performing for their intended audience.

Like the Flamer (i.e. the traveling sense), the Troll and the Ranter are outgroup roles, social deviants whose relation to the social community is, on the surface, generally negative. They are also known to engage in flame-like behavior.

4.7 The Troll
Jennie Jackson and Aula recognized the prescribed behavior for dealing with a Troll. Rusty Martin observed that others might mistakenly think he is a Troll. Who is this ubiquitously-recognized, yet reviled figure? The Troll is a master of “identity deception” (Donath 1998). He makes others believe he is someone he is not. A Troll attempts to pass as a valid member of the social community and begins to subtly provoke other members by writing messages that outwardly appear as honest attempts to start conversation but are really designed to “waste a group’s time by provoking a futile argument” (Herring et al. 2002). Herring et al. further posit that a Troll’s formula for success is to make a show of willingness to engage in legitimate discussion while “refusing to acknowledge” or “willfully misinterpreting another’s point” in order to perpetuate conversation.

The term ‘troll’ comes from a fishing activity, in which the fisherman puts out bait and waits for fish to bite. Likewise, the electronic Troll is a person who posts messages that are designed to “get a bite” – that is, stir someone to a predictably angry response. While “troll” is a verb in its original sense, it has been nominalized as a description for the person engaging in the act online, whereas fishing communities prefer “troller.” Herring et al. (2002) use “troller” to describe the online variety, but this form appears to be unattested in public use, except once in the alt.troll FAQ from 1997. Part of this may be due to the other sense of the noun “troll,” a mean, ugly character living under a bridge, which is often spuriously suggested as an etymology for the term.

For the Troll, communicative competence is a key trait. He must be adept at understanding and using the styles of speech a community deems acceptable, so as to not appear as an outsider. The Troll is dangerous precisely because his identity as a Troll and therefore his true motive is not known until it is too late. Donath (1998) considers the Troll’s actions to be part of a “game” he is constructing and forcing the others to play against their will and without their knowledge. His conversation begins innocently enough; this is how the trap is set. For many Trolls, this trap is for Newbies especially. If the Flamer attacks like a bomber, seeking to harm everyone, the Troll does so like a sniper, seeking the especially vulnerable Newbies. Because they are less familiar with the community’s standards and practices, Newbies are more likely to fall for a less-than-perfect performance by a would-be Troll. Goffman (1959) notes that, “a single note off-key can disrupt an entire performance.” For this reason, more experienced participants in the group are likely to “out” the Troll by noticing the “off-key notes” and posting a response to the Troll’s, directly accusing him of being a Troll. The more competence one has in the register of the community, the easier one will be able to identify a Troll’s deviant behavior. It is advantageous for a Troll to appear to be like any other new participant in the community. If a Troll is considered to be just another Newbie, then his lack of competence may be written off as inexperience instead of insincerity. While this can work to the advantage of a Troll, it can be to the detriment of a Newbie; on occasion, posts from previously unknown participants will be met with accusations of being a Troll.

(21) From: "Andersen Henrik" <Henrik.Andersen@iu.hio.no>
Newsgroups: alt.computer.consultants
Subject: Format C:

One day it came to my mind that I should get rid of that old Windows 95 OS and install a newer version of windows. So I formatted my hdd. So far so good. But when I booted the PC I got this awful error- message:
Invalid system-disk!
Try again, Abort etc....

No matter what I did I got the same irritating message. I expected to get to the command- line with the C: prompt. What have I done wrong?? Is there a simple way out of this misery?? Please help....

This post may or may not have been a legitimate request for assistance. “Is this a joke?” one participant responded. “Are you really asking why there isn't an OS after you formated the drive or are you pulling our legs?” asked another. Said a third, “If you are trying to be funny, it worked. I laughed.” They did not accuse him outright of being a Troll, but they were clearly suspicious of the poster’s sincerity. Donath (1998) observes that excessive trolling in a newsgroup can undermine the trust that the group gives to new participants. This has the effect of making new participants uncomfortable, leading to diminished growth in the community.

It remains unclear whether this post was an attempt at trolling or not. Since a Troll intentionally begins a conversation innocuously, his deceit takes time to build and execute, and an initial trolling post is, by design, indistinguishable from any other post. Over time, Trolls can grow from naïve to contentious and critical, or from “chilly to rude” (Donath 1998). Once this change takes place, however, a Troll has already baited one or more participants in the newsgroup. Once a participant is emotionally charged, it may be hard for him or her to let go.

In Chapter ‎3.4, I noted that intimidation works best when an involuntary relationship is at stake. Since Trolls and Flamers both rely on intimidation to achieve their desired result, they might be foiled by the voluntary nature of conversation in Usenet However, though the Troll or Flamer may not be linked in an inextricable relationship, the people they are intimidating often have strong emotional ties and investments in the community. Hijacking the group, the Troll or Flamer puts that relationship, not his own, on the line. Flamers, however, are easier to get rid of than Trolls; since their disruptions are much shorter, they travel around more frequently. Compared to the Troll or the Flamer, however, the Ranter’s disruptions can be quite long; once engaged by group participants, a protracted, frustrating relationship has begun that can last months or years.

4.8 The Ranter
Like the Troll, the Ranter seeks to stir up pointless debate. A prolific writer, the Ranter posts with great frequency on a particular issue or issues and is unique in his or her lengthy posts and single-mindedness. Ranters exhibit some Troll-like characteristics, in that they feign participation in a legitimate debate in order to goad others into responding. Unlike a Troll, however, a Ranter has an agenda. For the Ranter, the topic of discussion is of utmost importance. The FAQ of the alt.troll newsgroup, where Trolls gather to discuss their art, draws a distinction between Ranters, which they call “mission posters,” “nuts” or “net loons,” and “trolls” because the latter engage in what they are doing for fun, whereas the characters described by the first three terms actually believe what they are saying.

Because of their high posting frequency, Ranters may appear statistically similar to Celebrities. One Ranter, who goes by the name Chive Mynde, pushes a radical environmentalist agenda that includes distrust of science’s effects on global health in newsgroups like sci.chem, a community of people who make their livings as scientists. There are some statistical differences between this Ranter and the Celebrities in his midst. For example, typical of Ranters, he does not participate in conversation threads that he does not initiate himself. His messages are responded to at a low rate, and he has seven times as many orphan messages – messages to which nobody responds – than the average participant. Like most Ranters, Chive Mynde has a lot to say and says it at great length, but he is not well-received by the group as a whole.

Those who do respond to a Ranter are often trapped in a seemingly endless argument. Like the Troll, the Ranter fails to acknowledge other’s points. However, the Ranter’s behavioral strategies of choice do not always include intimidation, like the Troll or Flamer, but rather self-promotion and exemplification. The Ranter offers truth and, often, morality. Jones and Pittman note that “the exemplifier seeks to project integrity and moral worthiness” (1982:9). When their values are used a bludgeon with which they browbeat their opponents, the Ranter can be considered to be an intimidator. Often, intimidation is not used immediately; the Ranter’s interaction begins more peacefully. This is in keeping with the notion that intimidation requires a stronger relationship so that the intimidated does not just walk away. Building a relationship through conversation might give the Ranter enough leverage such that intimidation can be successful. Successful Ranters can dominate a group, like in the case of alt.computer.consultants, which is described in Chapter ‎5.3.
Some participants will take it upon themselves to fight with the Ranter to defend the honor of the group. As an exemplifier, the Ranter claims to present the truth; the foil that his opponents tend to employ, then, is discrediting him and his purported facts. The Ranter’s opponent has conventional vocabulary that fits well with his purpose. Words and concepts like “lies,” “truth” and “evil” are common. Because Ranters are ideologues, they become embroiled in the debate through attacks on their beliefs. Character attacks are not necessarily out of the question. Like Trolls, Ranters often use pseudonyms to protect their real identities. As a result, these figures are vulnerable to being attacked as “frauds” as well. Because social information is limited to that which is presented, it is difficult to establish credibility for oneself and fairly easy to discredit the reputation of others. While this is useful for combating Trolls and Ranters, it is dangerous to the legitimate participants as well.

4.9 Role Reputation
The reward a newsgroup participant earns for investing time and effort in the newsgroup’s social community is a reputation for a particular kind of behavior. A person’s reputation is a tool for contextualizing his future behavior. This is especially important in an environment like Usenet, where social cues that would ordinarily mark a speech event as deliberately unusual or ironic are absent. Reputation is perhaps more important electronically due to the potential for forgery and identity theft. A good reputation is a powerful tool against having one’s character misrepresented, as in the case of “Dorothy”[22] in misc.kids.

(22) From: toto <scarecrow@wicked.witch>
Newsgroups: alt.parenting.spanking,misc.kids,misc.kids.health, alt.activism.children
Subject: Re: What does well-behaved mean

Please note that someone is spoofing my headers also posting things I would never even think, much less say, so if anyone sees posts under my nic that seem unusual, check headers and realize that I have never posted from anywhere other then alt.net or my ISP - no anon remailers are in my repetoire [sic]... I don't nym shift, I don't troll, and I don't play the games these guys play .

Someone was “spoofing [her] headers,” which is Usenet-speak for forging messages so that they appear to have been authored by her. In a technical sense, it is not possible for a reader to see such a post and understand that it is a forgery. Socially, however, it would be very jarring. Dorothy is a Celebrity in misc.kids, and her reputation is a respectable one. The forged message was not consonant with that reputation; it contained long, graphic descriptions of acts of incest, something entirely out of place in a legitimate parenting community like misc.kids. She writes in this reply that these posts were “unusual,” that they contained things she “would never even think, much less say.” These claims cannot be made by someone whose reputation does not support them. If Dorothy were a Newbie with no reputation in the group at all, or were a participant who had engaged in disruptive behavior in the past, then these defenses would be unsupportable.

If this message had been posted to other newsgroups under her name, then the participants there could potentially think her to be a troll or troublemaker of some kind. Furthermore, in another community, her defense would not serve her as well because she would not have the reputation in those communities to give credence to her claim. Additionally, if this behavior continued in misc.kids – if the forger did not stop on his own – then Dorothy’s claims would eventually be unhelpful. One can make an appeal to one’s reputation only so many times before the reputation itself is tarnished. Initially, however, other regular participants in misc.kids do not just accept Dorothy’s explanation, they assume it. Since they are used to the posts that are part of Dorothy’s “repertoire,” they seem to assume that the dissonant posts are not legitimate; not a single participant responds angrily to the inappropriate forged messages, because Dorothy’s reputation in the misc.kids community prevents her from suffering undue blame.

5. The Effect of Roles on Community
A central point in this thesis has been that the kind of conversation that takes place in a newsgroup is highly dependent on the roles that are enacted by the participants. The preceding chapters laid out a variety theoretical tools for thinking about roles, as well as some roles derived with those tools. Now, these roles are viewed in the context of their effects on real communities.

The first part of this chapter discusses gender as a role; specifically, how exclusively male presence, or the perception of it, makes the environment hostile to women. The gendered language and assumptions that are acceptable by the community’s standards seem to be so only because the community is perceived to be exclusively male.

The remainder of this chapter deals with the effect roles have on change within communities. Protecting one’s social community, because one has made a great emotional investment in it and/or derives a great deal of satisfaction from it, is very important. The presence of strong community leaders ensures that the community’s cohesion remains intact in the face of disruptive influences. The opposite outcome is that an influx of disruptive characters in a community that does not have as much social cohesion can result in loss of the community to the intruders. By way of example cases, I argue that the interaction between the types of roles being enacted by participants in the newsgroup determines the community’s fate.

5.1 Gender as a Social Role
The Celebrities in rec.gambling.blackjack have names like “Big Ed,” “Harvey Cohen,” “Doug Grant,” “Rusty Martin” and “Seeker.” The first four of these individuals are likely all male; they have traditionally masculine first names and never offer any information suggesting that they are not male. The fifth name, “seeker,” likely belongs to a male as well. Again, he presents no information to the contrary. In an approximately month-long data set, no posters to rec.gambling.blackjack had traditionally feminine names, though many, like “Seeker,” had non-gendered names.

It is important to explore why there is the perception that rec.gambling.blackjack is male-only, regardless of whether or not it is true. Perhaps “male” is the default category for Usenet participants, and “female” is a category reserved for those who deliberately associate themselves with it. If this is the case, it could be for a variety of reasons. Perhaps more men than women participate on the internet; if so, then one could argue that this is a statistically valid, if sexist, assumption. Even if more men than women do participate in Usenet, a more compelling explanation is that communities’ topics of interest are socially gendered. For example, alt.sewing’s participants seem to be predominantly female, identifying themselves with names like “Sally,” “Elizabeth,” and “Julia.” Sewing may be considered, stereotypically and historically, an activity engaged in mostly by women. Since blackjack players are likely predominantly male, an activity-based gender assumption could be at play.[23] Aycock (1995) observes that a chess newsgroup’s participants always refer to Grand Masters as “he.” Aycock notes that the participants attributed this to “natural” reasons – that men are more naturally competitive. This is further evidence that, in electronic communities, stereotypes are reified and exaggerated rather than dispelled.[24]

(23) From: mhall@netcom.com (Abdul Jalib M'hall)
Subject: Helping Doug Grant WBBP to win - sky critter camouflage
newsgroups: rec.gambling.blackjack

The most important fashion accessory of all is a Hot Babe (tm). Counters are stereotypically alone. Break the stereotype, have a babe. As much as you might like girls with great legs, pretty faces, and nice butts, the sky will see none of that clearly. As Bryce Carlson suggests in his book, apparently, a girl with bodacious ta-tas is what will keep the sky critters entertained. They won't want to bar you... that would mean the girl goes too. If she is really incredible, they won't even be able to concentrate enough to count the deck down on you.

This author wrote a very long post about how to avoid being detected as a card-counter by casino management. The language in this post does more than stereotype women; it subtly demonstrates that the participants consider it to be an all-male group by implying that no women are present. If he argues that “the most important fashion accessory of all is a Hot Babe,” then it follows that he is writing this for people who are not “Hot Babes” themselves, i.e. men. Of the seven messages posted as replies to this one, four of them were simply to commend the author on “a great post,” “an excellent post” and “one of the best posts I’ve ever seen.”[25] Universal acclaim for sexist language breeds only more sexist language and further stereotype reification.

The advertising industry seems to have adoped the idea that “sex sells” as a primary guiding principle. Though it is noted above that there are no posters identifying as women in rec.gambling.blackjack, this is not exactly true. The only posts, however, that have traditionally feminine names attached to them are advertisements, none of which ever receive replies:

(24) From: "Doris Stevens" <dsteavens@yahoo.com>
Newsgroups: rec.gambling.blackjack
Subject: BLACKJACK and much more.... TRY and WIN <<<

Greetings from VORTEX CASINO, world's premier Internet casino. You can play casino games like in Las Vegas, win real money instantly, right on your computer over the Internet. LATEST SOFTWARE AND SECURITY

All new players receive US$30 gift with their first purchase of gaming chips just for joining Vortex Casino. You can also PLAY FOR FREE.

Sincerely,
Vortex Casino
http://www.vortexcasino.netfirms.com

Observe that the content of this post contains no gender-related information at all. Further, the post is signed by the Vortex Casino, not “Doris Stevens,”[26] who purports to be the author of the message. Similar posts from other online casinos are posted under names “Poker Ana,” “cheryl” and “carol.” Each post is a simple advertisement like the one above, signed by the casino or company name, often with no first-person text. The importance of the name in the “From:” line is great; it is the first thing that a reader sees when browsing the newsgroup’s messages. The reader decides whether or not to download and read the entire article based only on the “From:” and “Subject:” lines. The conclusion that can be drawn is that these posts are posted with deliberately female first names in order to attract the attention of the completely male community and lure them into reading the advertising message.

There exists a possibility that there are women present in rec.gambling.blackjack; that some of the non-gendered names belong to women who perceive it as an environment that is hostile to women. Bruckman (1993) observes that women online are likely to receive unwanted attention, usually with sexual overtones. Tying this in with Hackman’s (1992) observation that people are adept at determining whether a social group can provide them with satisfaction, it is reasonable to assume that women might find rec.gambling.blackjack to be an environment where they think they might be dissatisfied due to the responses they may receive. Since gender identification is flexible to some degree, women may feel pressure to conceal or lie about their own gender.

5.2 Protecting the Borders
Members of a community have an inherent interest in protecting the community from outsiders. This interest is especially strong for people who have spent a considerable amount of energy investing in community building, like Celebrities and other leader types. Usenet’s technology includes a practice called cross-posting, which allows an author to post a message to multiple groups at once. This has the effect of allowing readers of all newsgroups to which the message was posted to participate in the conversation together. This can have both positive and negative consequences; preventing the negative consequences is a task that is often taken up by group leaders.

When a post is cross-posted, two or more distinct social groups are thrown together, without their knowledge or consent. As a “region . . . bounded by barriers of perception” (Goffman 1959), a newsgroup that is part of a cross-posted conversation has its boundaries violated. The set of people with whom one could now interact has grown considerably, without the time or context necessary to assimilate these new people into the collective perception of the group. Consequently, a cross-posted conversation can be disconcerting to the newsgroup, just as any population explosion would be to any social group. For this reason, most groups’ participants are interested in keeping cross-posting to a minimum.

One method of bringing a cross-posted conversation thread to an end is to remove the newsgroups’ names from the list of groups to post subsequent messages to. This is called “trimming” the list of groups, and allows the conversation to continue only in one’s “home” group. In the next example, Aula guards the integrity of conversation in misc.kids by removing cross-posted headers and limiting the thread to misc.kids only.

(25) From: just me (No@ThankYou.com)
Subject: Re: Barbie shop with me cash register?
Newsgroups: misc.kids

/notes wild crossposting.
/cuts xposts except for those directly related to parenting
/figures that probably will mean the idiot who has shut himself in a small glass box will miss this post. oh well. trolls are trolls, after all!
-Aula

As a respected poster to the group, Aula has an unspoken license to act on the group’s behalf in such a way for everyone’s common benefit. She acknowledges that the person from whom the post originated is likely cut off from the conversation now and will likely not see her reply, but this is a small price to pay for ensuring the group’s social boundary remains secure. Though participants cannot build rigid fences around themselves, removing cross-posts is one way to minimally defend against invasion.

Cross-posting is not the only way a community’s borders can be violated. Unwelcome individuals, if not given a clear message that they are in fact unwelcome, may attract others. As observed in Chapter ‎4.2, new participants are notorious for asking quesions and seeking help and advice. While knowledgeable participants may like answering questions for all the reasons already described – self-image, public praise, etc. – these participants are a limited public good and so must be wary of free-riders (Kollock and Smith 1996:116). Free riders are people who take advantage of the resources offered by a community without contributing anything themselves. In this case, the resources are the answers that knowledgeable people provide. Though all new users are initially free riders – they don’t have anything to offer the community yet – it is important to identify which ones will never contribute back. One way community members can protect against the eternal free riders is to raise the standard for asking a question in the community. By requiring that answer seekers demonstrate that they have spent some time researching their problem themselves, a community ensures that fewer and worthier come seek their assistance.

One manifestation of this raising of the standard is sci.chem’s “no homework” policy. “Uncle Al” Schwartz, a Celebrity in that newsgroup, answers more questions than any other participant, but often does so in a terse way:

(26) From: UncleAl0@hate.spam.net
Newsgroups: sci.chem.
Subject: Re: avidin
(27) From: UncleAl0@hate.spam.net
Newsgroups: sci.chem.
Subject: Re: organic chemistry

*selah* wrote:
> At what temp. is avidin in raw egg white deactivated?
shirin wrote:
>
> what is the IUPAC name of cubane

Wei, R., and Wright, L.D. "Heat Stability of Avidin and Avidin-Biotin Complex and Influence of Ionic Strength on Affinity of Avidin for Biotin" Proc. Soc. Exp. Biol. Med. 117 341 (1964)
Now why would a person need to know that except for homework?

http://www.acdlabs.com/iupac/nomencl...93/r93_689.htm

Why couldn't you do that?
Or look up the C8H8 in a "Chemical Abstracts" Collective Formula Index.




Observe that, in both examples, he provides the answer the poster is seeking, but does so in a way that makes it clear that the way the question was posed was somehow inappropriate for the group. A closer look at the first post shows that, after providing a brief yet unmistakably helpful answer, Uncle Al rebukes the poster for obviously not putting any effort into seeking his own solution before asking publicly for help. The original poster is informed subtly that further supplication will not be looked upon kindly. The second post demonstrates sci.chem participants’ general aversion to homework questions. A simple nudge like, “Now why would a person need that except for homework?” is a kindly way to provide help and to get the point across that continued questions like this would be even more unwelcome. By providing double-edged answers that both assist and rebuke, members of this community can continue to provide the social support of a community while protecting themselves against those who would take advantage of them.

The individual who seeks to protect the community enjoys a personal benefit. Every time a participant successfully protects the social borders of the community, he reaffirms that he has the right to do so. Uncle Al, for example, demonstrates that he is indeed permitted to speak for the sci.chem community and say what sorts of questions are unacceptable and why. This is good for the newsgroup because it ensures that there are people willing and able to support its best interests, and it is good for the individual who gains social prestige by having this privilege.

Dealing with invasion requires coping mechanisms. Many cultural groups have sought to reclaim the vocabulary of hateful speech used against them, redefining it as an in-group term, rather than an insult imposed from outside. A neologism, “spamagram,” has been created by the participants in alt.anagrams to describe just this act, whereby a spam message posted to the group is turned into an insulting anagram:

(28) From: Paul Martin Howard[27]
Newsgroups: alt.anagrams

> A little while back, I was browsing through some newsgroups, just like you are now. I came
> across an article similar to this that said you could make thousands of dollars within weeks with
> only an initial investment of six dollars! I thought: "Damn, this must be a scam and illegal".
=
It was, but as that didn't stop you, you chose to vex us all with a new incarnation of the same limited story. Oh, I wish this jerk's cash claims were true, but I know such unmitigated trash is illegal and lacks kudos. So, we ignore mail or blame? O no! I'll find anagrams among all this swill.

By taking a spam posted to a newsgroup (i.e. the quoted text) and anagramming it (i.e. the text below the = sign), the author has been able to make a joke at the spammer’s expense and share it with the community.[28] By using spam as anagram fodder, the alt.anagrams participants’ community is disrupted less by spam. Their anagramming does not affect the volume of spam they receive; rather, by contextualizing the spam, it has a less damaging effect socially.

Though a newsgroups’ participants can employ a variety of strategies to combat those who would bring about change in the group, sometimes change is inevitable, especially given a significant change in population. Simply put, when the participants of a group change, the group changes as well.

5.3 Population Shift and Topic Shift
Groups change over time. An important question is, how does that change take place? We have seen that longer-term, more active posters who have invested in the community have a relatively greater stake in seeing the community succeed. In Chapter ‎4.3, uccess was implicitly defined as maintenance of the status quo. In this section, it becomes necessary to complicate that definition. Two seemingly contradictory hypotheses arise: First, longer-term, more active posters have more power in enforcing group boundaries. Let us call this Shift Hypothesis 1. Aula’s cross-post trimming and Uncle Al’s curt replies demonstrate this hypothesis. Second, longer-term, more active posters have greater privilege to push and extend boundaries of acceptable conversation. Let us call this Shift Hypothesis 2; it is demonstrated in the following example:

(29) From insurancenj@aol.com
Newsgroups: alt.computer.consultants
Subject: Re: BITCH!...BITCH!...BITCH Was.... "H-1Bs now can stay longer"

>From: Carey Gregory antispam9@attbi.com
>That's because medicine is broken into specialty now. The "GP" of yesteryear is, for the most
>part, a thing of the past. Most of the doctors you would call a "family doctor" today are certified
>in internal medicine or similar. You will find very, very few doctors in the US who don't fit into
>one of the categories listed in that URL.
>
>But I think we've drifted far off the subject matter of alt.computer.consultants......

True but I think we have had a pretty good conversation. Here is a question for you. Do you think that there would be a shortage today if 50 years ago congress instead of letting the corporations import cheap nurse made those same corporations improve working conditions and maybe put a buck or too into the education of Americans kids looking for a career?

In this example, the conversation has drifted in topic from health care and immigration to specialization in medicine. InsuranceNJ is the most prolific author in alt.computer.consultants. He participates in many conversations in the group, has posted more messages than almost any other participant, and is often controversial. Carey Gregory, a regular participant, is not so prolific a writer, and may be more concerned with rigid adherence to the topics of which the group has shown themselves to approve. In contrast, InsuranceNJ feels comfortable enough in this environment to have whatever conversation he likes, regardless of its relationship to the central topics.

If longer-term, more active posters can maintain group boundaries (SH1) and are the only ones who can drive change (SH2), then logically, no change could ever come about. However, since change does come about, these definitions must not be contradictory. Perhaps it is the case that these hypotheses apply to different levels within the community. It may in fact be the case that long-time participants do in fact want to maintain the status quo on a newsgroup level, but have the ability and perhaps desire to push boundaries on a conversation level. Precisely because of SH1 (Celebrities can maintain newsgroup-level boundaries), SH2 can be true (Celebrities can push conversation-level boundaries) without significant group-level changes taking place. That is, Celebrities are forces of short-term change and long-term stability.

How, then, does long-term change take place in a newsgroup? The newsgroup alt.computer.consultants is an example of a group that has undergone radical topic shift in the recent past. In this upheaval, a number of Ranters were able to take control of the group, establishing themselves as the dominant figures in the group and destroying the community existing there. In mid 2000, a number of now very active participants arrived in alt.computer.consultants and began to discuss the H-1b visa program and influx of Indian programmers in US technology jobs. Over the course of a few months, the newsgroup changed from being a resource for computer consultants (sharing technical support, how to budget a consulting business, etc.) to being a xenophobic, racist group, spouting hate toward Indians and anti-immigration propaganda about US economic policies. Charting the frequency of the terms “India” and “H-1b” shows the change that took place within the newsgroup:

Occurrences of "India" and "H-1b" in subject lines per 100 posts
in alt.computer.consultants [29]


• India
• H-1b






A change in population is responsible for the change in topic. InsuranceNJ and a number of other now very active participants in alt.computer.consultants began to participate in the newsgroup around this time and participated eagerly in the immigration discussions that continue to this day. In response to this change, many participants who were dissatisfied left the group to create a moderated version of the group, alt.computer.consultants.moderated, instead. The new group is moderated; a responsible volunteer moderator must approve all posts. In the case of accm, moderation is a community protection tool. When the members of alt.computer.consultants who were not interested in discussing immigration and H-1b visas left the group, the proportion of those people who did rose. This change in community also effected a change in roles for the participants. Those who began as Ranters – highly active participants with little common ground – became the de facto leaders of the group. The topics important to them, immigration, etc., became the prestige topics and thus their ideas and beliefs became common ground, and the takeover was complete.

The volume of unacceptable behavior relative to acceptable behavior grew too high for the long-time participants to combat or tolerate it; this is the cause of population and topic shift in alt.computer.consultants. However, there are underlying causes at work. There exist a variety of possible reasons for why the people interested in discussing computer consulting did not win out against those interested in H-1b. First, the influx may simply have been too great; if public verbal censure does not dissuade unwanted participants and there are too many of them to ignore, being overwhelmed is possible. Second, the nature of conversation may have played a part in the change. The participants in alt.computer.consultants actually discussing computers did not engage in very long conversations. The typical conversation involved a question being asked and a handful of answers being supplied. This made for very short conversations with frequent turnover. On the other hand, the conversations on immigration and H-1b could last for hundreds of messages. These longer, more populous conversations could have the effect of “drowning out” other conversation. Hackman (1992:244) observes that only a cohesive group can reject deviant behavior. Short conversation threads, though they may be a result of the kind of conversation (i.e. technical), could be a symptom of lack of cohesion. Whittaker et al. (1998) use thread depth (i.e. conversation length) as a proxy for interactivity, which is a measure of successful group communication. If this newsgroup was failing to have prolonged successful communication and was experiencing it in short bursts instead, then it is likely they were not experiencing the cohesion necessary to reject the deviant behavior.

6. Conclusion
Usenet newsgroups are homes to active, ever-changing social communities populated by people who carry out a variety of roles in their daily social interaction. I have demonstrated in this thesis the benefits of a role-based framework for interpreting social interaction in usenet communities and introduced a typology of roles with which to structure that framework.

Using a combination of ethnographic and statistical approaches, I developed a typology of social roles that is firmly grounded in the ethnography of communication. Communicative competence, the ability to express oneself in a socially meaningful and acceptable way, is an important metric in this typology, as is frequency of participation and common ground shared with one’s fellow community members. Additionally, the strategies employed in one’s social interaction are fundamental in understanding one’s social function in the community.

I demonstrated that the participation inequality that Whittaker et al. (1998) observed is due to participants’ different needs and goals, as well as different abilities and privileges. Some participants seek community and invest themselves in developing a cohesive social environment. Others are driven by information needs and seek answers, not community. Some seek to disrupt others’ community-building effort for nefarious or ideological reasons. With such divergent agendas, it is reasonable that users’ patterns of participation vary. More subtle is the notion that participants’ abilities and privileges vary as well, especially considering that Usenet is designed to be an egalitarian system with no overt, prescribed controls. However, we have seen examples of some community members who develop reputations as trustworthy, knowledgeable and helpful, and some who have yet to develop any reputation and are still lower status members of the community.

By defining a role as the sum of what one does and what one can do, I attempted to reconcile two approaches to understanding social roles. Holland and Skinner (1987) employed the “social drama” as a tool for categorizing individuals by the social scenarios in which they participate. Additionally, Goffman (1959:16) emphasized the “rights and duties” attached to social status. Combining these two theories, we find that it is important to consider both what one does (i.e. in the context of his “social dramas”) and what one can do (i.e. the rights and duties ascribed to him). One result of defining a role this way is that it becomes clear that roles are reflexive; by enacting the behaviors associated with a role, one solidifies his status as having that role. “Uncle Al” Schwartz is a good example; by defending his community’s social boundaries, he reaffirmed both his right to do so and his status as a leader who can speak for the community. Celebrities like him, Peppe, and Sven Guckes build community and social cohesion within their respective newsgroups. Strong communities are better protected against those individuals who would disrupt them, as Chapter ‎5 demonstrated.
The importance of roles in understanding communities cannot be overstated. Indeed, a key claim of this thesis is that the kind of conversation that takes place in a social community is dependent on the roles enacted within. For example, too many new participants asking questions can risk depleting the resource of knowledgeable help, in the cases of sci.chem and comp.compilers, or can make the Celebrities weary, as in rec.gambling.blackjack. Finally, the case of alt.computer.consultants demonstrated what can happen to a group when the balance of roles changes. Arrival of too many Ranters led to a complete change of focus for that newsgroup and drove many members of the community away; it is a completely different newsgroup today from what it was three years ago. Indeed, these are all examples of how roles can be seen as forming the structure of the community. Just as roles are the sum of what people do and can do, a community is the sum of the roles that are enacted within.

To conclude, I return to Hymes’ (1972:43) observation about the use of sociolinguistic taxonomy:

[S]ociolinguistic taxonomy is not an end in itself, any more than is language classification. A taxonomy is not in itself a theory or explanation, though it may conceal or suggest one. . . The work of taxonomy is a necessary part of progress toward models of sociolinguistic description, formulation of universal sets of features and relations, and explanatory theories.

(Hymes 1972:43)

I choose to reiterate Hymes’ belief that taxonomy is part of progress toward building explanatory theories, because ethnographic research is susceptible to what Paccagnella (1997) calls “interpretive flexibility.” He observes that computer-mediated communication, even more than face-to-face communication, is subject to being interpreted differently by different people. I agree with his claim; in an environment where fewer social and linguistic cues are presented, there is greater room for the researcher to “fill in the gaps,” so to speak. Ensuring that one’s analysis is as free from personal unconscious biases as possible is an ad hoc solution that each ethnographer must build on his or her own. Having a taxonomy in which to frame ethnographic research online is an important step in developing a systematic approach for rigorous fieldwork in electronic communities.





Postscript



There are some people who are so irritating that merely ignoring them is somehow unsatisfactory; every fibre in your body screams out at you to grab them by the throat with both hands, shake them until their teeth rattle, and shout in their face "We're not fucking interested, understand? And you're a waste of space!".

Sadly, my newsreader doesn't seem to implement this feature.

Tequila Rapide, alt.peeves, June 23, 1996



Usenet is like a herd of performing elephants with diarrhea -- massive, difficult to redirect, awe-inspiring, entertaining, and a source of mind-boggling amounts of excrement when you least expect it.

Gene Spafford, rec.humor.funny, March 9, 1993





References

Aycock, Alan. 1995. “Technologies of the Self: Foucault and Internet Discourse.” Journal of Computer Mediated Communication. 1(2).

Bauman, R., and Briggs, C. L. 1990. “Poetics and performance as critical perspectives on language and social life.” Annual Review of Anthropology. 19.

Baym, Nancy K. 1995. “The Performance of Humor in Computer-Mediated Communication.” Journal of Computer Mediated Communication. 1(2).

Boyd, Danah, Hyun-Yeul Lee, Daniel Ramage. and Judith Donath. 2002. “Developing Legible Visualizations for Online Social Spaces.” Proceedings of the Hawaii International Conference on System Sciences. Big Island, Hawaii. Jan. 7-10, 2002.

Bruckman, Amy. 1993. “Gender Swapping on the Internet.” The Internet Society (INET ’93). San Francisco, California. August 1993.

Chen, Lee Li-Jen and Brian R. Gaines. 1996. “Knowledge Acquisition Processes in Internet Communities.” Proceedings of the 10th Knowledge Acquisition Workshops. Banff, Canada. Nov. 9-14, 1996.

Cherny, Lynn. 1999. Conversation and Community: Chat in a Virtual World. Stanford, CA: CSLI Publications.

Crystal, David. 2001. Language and the Internet. Cambridge: Cambridge University Press.

Donath, Judith. 1998. “Identity and Deception in the Virtual Community.” Communities in Cyberspace., ed. by Marc Smith and Peter Kollock. Routledge.

Dundes, Alan, Jerry W. Leach, and Bora Ozkok. 1972. “The Strategy of Turkish Boys’ Dueling Rhymes.” Directions in Sociolinguistics: The Ethnography of Communication., ed. by J. Gumperz and D. Hymes. New York: Holt. 130-60.

Goffman, Erving. 1959. The Presentation of Self in Everyday Life. New York: Doubleday.

---. 1969. “Strategic Interaction.” Strategic Interaction. Philadelphia: U of Pennsylvania P.

Gumperz, John J. 1972. “Models of the Interaction of Language and Social Life.” Directions in Sociolinguistics: The Ethnography of Communication., ed. by J. Gumperz and D. Hymes. New York: Holt. 1-25.

Hackman, J. Richard. 1992. “Group Influences on Individuals in Social Organizations.” Handbook of Industrial and Organizational Psychology. Vol. 3., ed. by M.D. Dunnette and L.M. Hargh. Palo Alto, CA. Consulting Psychologists Press, Inc. 199-267.

Hall, Kira. 1996. “Cyberfeminism.” Computer-Mediated Communication: Linguistic, Social and Cross-Cultural Perspectives., ed. by Susan C. Herring. Philadelphia: John Benjamins Publishing Company.

Hauben, Michael and Ronda Hauben. 1997. Netizens: On the History and Impact of Usenet and the Internet. Los Alamitos, CA: The IEEE Computer Sociey. Also available online at http://www.columbia.edu/~hauben/netbook/

Herring, Susan. 1996. “Linguistic and Critical Analysis of Computer-Mediated Communication: Some Ethical and Scholarly Considerations.” The Information Society. 12(2). 153-168.

---. 1999. “Interactional Coherence in CMC.” Journal of Computer Mediated Communication. 4(4).

Herring, Susan, Kirk Job-Sluder, Rebecca Scheckler and Sasha Barab. 2002. “Searching for Safety Online: Managing Trolling in a Feminist Forum.” The Information Society. 18(5).

Holland, Dorothy and Debra Skinner. 1987. “Prestige and Intimacy”. Cultural Models in Language and Thought., ed by Dorothy Holland and Naomi Quinn. Cambridge, EG: Cambridge University Press.

Hymes, Dell. 1964. “Introduction: Toward Ethnographies of Communication.” The Ethnography of Communication., ed. by J.J. Gumperz and D. Hymes. (American Anthropologist Special Publication 66 (6), part 2.) Washington DC: American Anthropological Association. 1-34.

---. 1972. “Models of the Interaction of Language and Social Life.” Directions in Sociolinguistics: The Ethnography of Communication., ed. by J. Gumperz and D. Hymes. New York:Holt. 37-71.

---. 1974. Foundations in Sociolinguistics: An Ethnographic Approach. Philadelphia: U. of Pennsylvania P.

Jones, Edward E. and Thane S. Pittman. 1982. “Toward a General Theory of Strategic Self-Presentation.” Psychological Perspectives on the Self., ed. by J. Suls. Hillsdale, NJ: Lawrence Erlbaum. 231-62.

Kollock, Peter and Marc Smith. 1996. “Managing the Virtual Commons: Cooperation and Conflict in Computer Communities.” Computer-Mediated Communication: Linguistic, Social and Cross-Cultural Perspectives., ed. by Susan C. Herring. Philadelphia: John Benjamins Publishing Company. 109-128.

Komatsu, Lloyd. 1992. Recent Views of Conceptual Structure. Psychological Bulletin 112. 500-26.

Labov, William. 1971. "Methodology." A Survey of Linguistics Science., ed. by W.O. Dingwall. College Park: U of Maryland Press. 412-97.

---. 1972. Language in the Inner City. University of Pennsylvania Press.

---. 1997. "In Search of the First Language." Nova. WGBH. March 18, 1997.

Landfield, Kent. 2001. “NetNews Moderator’s Handbook.” Available online at http://www.landfield.com/usenet/moderators/handbook/

Mark, Noah. 1998. "Beyond Individual Differences: Social Differentiation from First Principles." American Sociological Review. 63. 309-330.

Paccagnella, Luciano. 1997. "Getting the Seat of Your Pants Dirty: Strategies for Ethnographic Research on Virtual Communities." Journal of Computer Mediated Communication. 3(1).

Potenza, Marc N. M.D., Ph.D.; Steinberg, Marvin A. Ph.D.; McLaughlin, Susan D. M.P.A.; Wu, Ran M.S.; Rounsaville, Bruce J. M.D.; O'Malley, Stephanie S. Ph.D. 2001. “Gender-Related Differences in the Characteristics of Problem Gamblers Using a Gambling Helpline.” American Journal of Psychiatry. 158(9). 1500-1505.

Rafaeli, S. 1988. “Interactivity: From New Media to Communication.” Advancing Communication Science: Merging Mass and Interpersonal Processes. 16., ed. by R.B. Pawkins, J.M. Weimann, S. Pingree. Sage Annual Reviews of Communication Research. Beverly Hills, CA: Sage.

Raymond, Eric, ed. 2002. “The Jargon File 4.3.3.” http://www.catb.org/~esr/jargon/

Saville-Troike, Muriel. 1982. The Ethnography of Communication: An Introduction. Baltimore: University Park Press.

Smith, Marc. 2001 Netscan: A Tool for Measuring and Mapping Social Cyberspaces. http://netscan.research.microsoft.com.

Whittaker, Steve, Loren Terveen, Will Hill, Lynn Cherny. 1998. “The Dynamics of Mass Interaction.” Proceedings of Conference on Computer Supported Cooperative Work. 257-264.

Wilson, Diane. 2001. “About Trolls and Flamers.” Available online at
http://www.firelily.com/support/depression/trolls.html

Vaux, Bert and Justin Cooper. 1999. Introduction to Linguistic Field Methods. Munich: Lincom Europa.

Zurcher, Louis A. 1983. Social Roles. Beverly Hills, CA: Sage Publications, Inc.



--------------------------------------------------------------------------------

[1] The preferred citation for the Netscan study is from 2001, because that is its year of publication. The data are updated continually.

[2] The “news server” is a computer server connected to the internet which transmits Usenet messages among other servers.

[3] This technology is called the Network News Transportation Protocol, or NNTP. It is the protocol by which Usenet messages are propagated among news servers.

[4] In Usenet, flame may be either a noun or a transitive verb. A flame is a message that is extremely hostile, often using angry speech and aggressive language. To flame another person is to send him or her such an angry message. A flamewar constitutes a series of exchanges of flames in a single conversation. See Chapter ‎4.6 The Flamer.
[5] Google Groups is a service provided by Google, Inc. which maintains a public, searchable archive of Usenet. It is available online at http://groups.google.com. Many of the examples in this thesis were found in the Google Groups archive.

[6] A human reader, unlike a computer, can read messages and understand the case of multiple posts coming from a single person using two or more email addresses. This can be innocuous. For example, an individual may have email addresses for work and home and will post to the newsgroup with whichever is more convenient at the time. This highlights the difficulty of some forms of computational analysis of social communication.

[7] Though it is beyond the scope of this thesis, this dataset would provide an excellent information source for a study on register switching in newsgroups. Since each newsgroup is its own speech community, reason dictates that users who participate in multiple newsgroups must be adept in using the speech varieties of all the groups in which they participate.

[8] The vast majority of Usenet newsgroups use English. Exceptions to this include newsgroups dedicated to specific languages or to non-Anglophone countries.

[9] Following Hymes’ (1972:45) use of the term, I use the term “speech” to refer to uses of language that are textual as well as vocal, and explicitly include written communication on the internet in this group. Hymes takes speech “as a surrogate for all forms of language, including writing, song and speech-derived whistling, drumming, horn calling, and the like.”

[10] For example, use of all capital letters implies yelling. This feature has become so ingrained in Usenet and beyond that all-capitals is called yelling even when it is clear that the speaker did not mean to yell, but rather did not know better than to use all capitals. Other text forms that compensate for prosody include surrounding words in asterisks or underscores or, where available, using boldface.

[11] “emoticon” is short for “emotion icon” and is a facial expression constructed of punctuation marks. For example, a colon and left parenthesis forms a smiley face when one turns one’s head 90 degrees to the left, and indicates humor or lack of seriousness.

[12] A MOO (“MUD, Object Oriented”) is a near-synonym for MUD. Other related acronyms include MUSH and MUCK, and are so named based on the variations in the underlying server software (Cherny 5). The difference to the user is negligible or nonexistent.

[13] This excludes the case of general users who have been granted programmer status by the Wizards. As mentioned earlier, the programmer status does in fact grant the user some technical control over the design of rooms and objects within the MUD. While such users may be Power Elite, the group is loosely defined; there are programmers who are not Power Elite and there are Power Elite who are not programmers.

[14] There exist a subset of newsgroups that have moderators, who must approve each post before it is distributed to news servers worldwide. The moderator does, in fact, have an official, administrative role. Kent Landfield observes that moderation is seen in groups that are news or information sources and are not meant for conversation, or groups that act as lower-volume substitutes for high-volume groups in which most readers could not or would not want to keep up. In general, most Usenet groups are not moderated, and there is no formal control. Moderated groups are therefore explicitly excluded in this thesis, except where otherwise noted.

[15] AS3 is shorthand for alt.support.stop-smoking. Many newsgroups will abbreviate the newsgroup name in a similar way. This applies even to participants; rec.arts.bodyart is called RAB, and its participants RABbits.

[16] This example contains two posts. The first of which is from “Jennie Jackson.” The second is from “Aula.” Note that Aula is quoting Jennie Jackson’s post (quoted material begins with the ‘>’ character); Jennie’s post comes first, and Aula’s below. This is the standard Usenet quoting convention.

[17] The aus hierarchy is for Australia-based newsgroups. aus.motorcycles is in many ways similar to rec.motorcycles; in fact, they share this acronym.

[18] Source code sample removed.

[19] Comp.compilers is a moderated newsgroup. This message, however, is typical of new-user questions in both moderated and unmoderated groups.

[20] Crystal (2001:90) contends that angle brackets are more common than asterisks for denoting gestures and facial expressions, arguing for forms like <creak> rather than *creak* because, in his estimation, asterisk use is idiosyncratic and has a variety of other functions. I would argue that angle bracket use is an older form and is declining in use. Since older members of a community are more likely to preserve old language forms (Vaux 1999:150), one might hypothesize that <creak> would in fact be more frequent than *creak*. However, this does not appear to be the case. Google Groups data reveal that the asterisk form, *creak*, is more common with a ratio of 3 to 1.

[21] The Jargon File is the online name of this resource. It was published as a book under the name The New Hacker’s Dictionary.

[22] It is clear from her email address (toto@wicked.witch) that “Dorothy” is a pseudonym. However, it is one that seems to be used for privacy, not for the anonymity often used as a shield for deviant behavior.

[23] A Yale University study (Potenza et al. 2001) found that more men than women sought help for gambling addiction and that men were more likely to be addicted to blackjack or poker, whereas women were more likely to be addicted to slot machines or bingo.

[24] For further reading, see Hall (1996) in References.

[25] The remaining three replies were not gender-related.

[26] Note the typographical error in the “From:” line. “Stevens” in the real name is spelled differently from “steavens” in the email address. This is common among pseudonyms used for nefarious purposes. For example, “Chive Mynde,” the Ranter from Chapter ‎4, has an email address “chyveminde@my-deja.com” – note the transposed y and i. It is unclear whether these spelling changes are intentional. However, this does reinforce the idea that these are in fact pseudonyms; no real person would misspell his r her own name.

[27] This example is reconstructed from a subsequent post that replied to the one shown here. No text has been altered, except that one level of leading ‘>’ has been removed.

[28] In case it is unclear, the quoted text is the body of the spam message. The unquoted text below the = sign is an elaborate anagram of the spam text.

[29] It remains unclear why “India” references rise and fall rapidly around 1995.
o

st
.
w
a mnn
o.
s

[سمير الشناوي]

اشكرك د دنحا
ساخذ نسخة واقرئه عند عودتي

تقديري
سمير