Home / Articles / Volume 7 (2010) / Hanging by a thread: Topic development and death in an online discussion of breaking news
Document Actions



For more than 20 years, cyberspace promoters have advanced a dream of the Internet as an always emerging, egalitarian community. Iterations of this dream appear during media coverage of breaking news or technological developments, when journalists discuss blogging, crowdsourcing, social media, and other online trends tied to grassroots democracy and free speech. As an example of how hierarchies between major news media and their audiences have been leveled, a bystander’s cell phone video and student-generated online discussions rivaled major news media for attention during the Virginia Tech shootings of April 16, 2007 (Palen, Vieweg, Sutton, Liu, & Hughes, 2007). During the past decade, these egalitarian dreams have sometimes been called Web 2.0, a term that connects advancements in user-generated content capabilities with digital democracy (although the term Web 2.0 and its implications have been challenged by World Wide Web developer Tim Berners-Lee and others, e.g., Scholz, 2008). Yet as early as 1990, Poster claimed that technological advancements enabled “the ideal of participatory democracy of the Greek agora and the Colonial New England town meeting” (1990, p. 123). Similarly, Lanham (1993) theorized that in a technologized culture, electronic expression would create a newly democratized society.

At the same time, much expression in virtual discussion spaces revolves around special interests rather than community issues at large (Grossman, 1996; Kollock & Smith, 1996). Rather than supporting democracy through egalitarian public discussions, electronic enclave mentalities have been seen as a threat to community-based politics, in as much as “people with computers are creating interest-based communities in place of geographically-based communities” (Grossman, p. 14). Other scholars warn that selective engagement in Web activities may cause digital fragmentation, permitting “people to pursue their interests while ignoring much of what is happening around them” (Gattiker, 2001, p. 96). Finally, while software developments and Web codes may have made both sharing opinion and challenging authority easier over the last decade, the networks that enable these online connections have their roots in decades-old technology (Scholz, 2008).

As the technology developed, theory builders such as Poster (1990), Lanham (1993), Negroponte (1995), Bolter (1991), and Landow (1993) focused on its potential to enable egalitarianism, forming a canon1 around Marshall McLuhan’s visions of a global village. These theorists praised technology and hypertextual discourse as revolutionary forces that would affect humankind’s relationship with politics and culture. Ess (1994) endorsed the democratization claim made on behalf of hypertext and other computer-mediated discourse, based on Juergen Habermas’s theory of communicative action and Habermas’s formulation of the rules of reason in his Discourse Ethics.2 Making a similar theoretical claim, Poster described computer conferences as able to

upset the power relations, both economic and gendered, that govern synchronous speech. Factors such as institutional status, personal charisma, rhetorical skills, gender, and race—all of which may deeply influence the way an utterance is received—have little effect in computer conferences. Equality of participation is thereby encouraged. (1990, pp. 122-123)

Ess qualifies his claim for democracy in cyberspace, stating that discourse must be free from social coercion, including the “subtle but powerful cues of hierarchy, status, gender, and so on” (1994, p. 252). Poster, in contrast, dismisses these influences by stating that “new, serendipitous considerations, like typing speed, determine who ‘speaks’ most often” (p. 123).

Such optimism masks other experiences of computer-mediated communication (CMC), as revealed, for example, in studies of gendered interactions in Internet discussion groups (Carstarphen & Lambiase, 1998; Herring, 1993; Hobman, Bordia, Bernd & Chang, 2002; Lambiase, 2001a). Herring (1993), also using Habermas’s rules of reason, found that a few male members often dominate electronic discussions “in terms of amount of talk, and rhetorically, through self-promotional and adversarial strategies” (1993, p. 10), and that women are effectively censored when their contributions are ignored or belittled. In other ways, powerful and persistent language users—both men and women—control discussion groups and their topics of conversation, lessening the chances that all opinions are expressed (Selfe & Meyer, 1991). As this research suggests, topic control or interactional control theory is important to the understanding of relationships among conversation participants, of how topics are “introduced, developed, and established,” and of whether topics are controlled in symmetrical or asymmetrical ways (Fairclough, 1992, pp. 234-235).

Messages generated for online discussion groups contain topics of conversation that guide responses and prompt participants to introduce new threads for discussion. So pervasive is the tendency for new topics to be introduced that the original purpose of an online discussion group may become buried under layers of unrelated topics, especially if discussion is unmoderated and the most active participants choose to pursue threads removed from the original topic. Despite some scholarship devoted to topic development in synchrononous chat (e.g., Goutsos, 2005; Herring, 2003), however, asynchronous discussion-group discourse has received little attention from the perspective of topic development, in terms of how topics are introduced, developed, and discarded in computer-mediated conversation. Nor has the effect of these topic changes been considered on participants who originally subscribed to the discussion for a specific purpose.

In this study of an email-based listserv discussion of the Oklahoma City bombing, I analyze the development and control of topics of discussion and their effects on participation. The results reveal that a few contributors gained control of the discussion group by posting inflammatory and emotional messages, by “misunderstanding” previous messages and construing them to conform to other topics, and by overloading the group with multiple postings about off-topic subjects. This last tactic of multiple messages inundates other subscribers’ online mailboxes, which is intimidating both for its sheer volume and for the often polemical language used, which discourages the presentation of other viewpoints. In this way, a handful of participants controls topics and eventually leads the group away from its original purpose. Their tactics force other members to retreat in silence, to unsubscribe from the group, or to refrain from posting on unpopular or trivialized topics.


This study of the OKLABOMB online discussion list, like other research into topic management, presupposes that participants have an “intuitive notion of topic and (are) sensitive to topic boundaries” (McLaughlin, 1984, p. 56). Topic shifts are initiated in face-to-face conversation through use of key words such as ‘like,’ ‘incidentally,’ ‘so,’ and ‘anyway,’ and markers such as ‘speaking of X’ and ‘not to change the topic, but . . . ,’ both of which show “sensitivity to conversational rules and concern for easing (the listener’s) processing task” (McLaughlin, 1984, p. 57). Sacks (1972), too, is concerned with topic transition in face-to-face conversation:

A general feature for topical organization in conversation is movement from topic to topic, not by a topic-close followed by a topic beginning but by a stepwise move, which involves linking up whatever is being introduced to what has just been talked about, such that, as far as anybody knows, a new topic has not been started, though we’re far from wherever we began. (Lecture 5, Spring 1972, pp. 15-16)

Sacks’ “stepwise move” is similar to the conversational device of semantic parallelism (Hobbs, 1990). Parallel associations work in conversation when, for example, “[f]irst we changed the argument and kept the predicate the same; then we changed the predicate and kept the argument the same” (Hobbs, 1990, p. 10).

Two other conversational devices contribute to topic drift, according to Hobbs: “chained explanations, or explanations that become topics in their own right,” and “metatalk used as a way of introducing a new topic” (p. 8). A chained explanation works in casual conversation when “the explanation sometimes turns out to be more interesting than the explained, and the conversation somehow never gets back to the original topic” (p. 11). It is possible, of course, for the conversation to return to its original topic. Metatalk or evaluation influences topic drift, Hobbs believes, when talk turns to “the goals of the conversation” (p. 9). Moreover, for scholars who study topic development, the topic can never be “a static feature of conversation but rather the outcome of an ongoing negotiation” (Dorval, 1990, p. 1). Conversational interactions work best when “participants are mutually engaged and their manifestations of this engagement are complementary” (Reichman, 1990, p. 25). The Dynamic Topic Analysis paradigm developed in the late 1990s by Herring (1999, 2003) operationalizes Sacks’ and Hobbs’ ideas about stepwise and parallel conversational moves. However, while designed for any multiparticipant CMC, it has been applied mainly to synchronous communication, rather than to the type of asynchronous conversation in the present study.

Computer-mediated discourse is not as spontaneous as face-to-face conversation, and it requires “interactants (to) work to overcome limitations of the channel as they get to know one another” (Tidwell & Walther, 2002, p. 342). Neither is turn-taking in online discussions practiced in the same way as face-to-face interaction. Discussion group subscribers not only may spend more time preparing to communicate through their own posts to the listserv, but they also have more time to analyze the messages of others. However, participants of asynchronous electronic conversations may receive messages in an order that may seem chaotic. Making sense of a conversation that is received in bits and pieces is a little like reconstructing a printed message that has been torn into pieces.

Requests to remain on topic are one type of metatalk that occurs often in the OKLABOMB discussion studied here. Conversation strays from relevant to unrelated message content through gradual shifts, such as stepwise moves and parallel associations or chained explanations within one single message. Also common in CMC are abrupt shifts of topic—what Herring (Herring, 1999; Herring & Nix, 1997) terms “breaks.” This happens when isolated participants post messages without the visual clues of face-to-face conversation, without an obvious order for messages, with uneven access to media reports about a particular event, and/or with disparate language skills and emotional ties. Topic development and topic cohesion may be easier to maintain in electronic discussions and online newsgroups that focus on a global topic, but more difficult in chat rooms where users may exploit gaps to play with language (Herring, 1999; Herring & Nix, 1997). Sometimes in online chat, the topic is beside the point, because participants know that “the best way to get a response from a person in a chat room is to issue a provocative statement” (Rollman, Krug & Parente, 2000, p. 165). Topic maintenance, then, in both listserv discussions and chat rooms, may be controlled as a “joint accomplishment of participants,” or it may be asymmetrically controlled by a few (Fairclough, 1992, p. 234). When there is topic degeneration in online discussions, it may be due to the consensus of the group to allow degeneration or to the control of a few powerful participants who effectively force degeneration.

In terms of gender and topic maintenance, some researchers interpret male-female miscommunication in conversation as innocent (Tannen, 1990), while others (e.g., Henley & Kramarae, 1994) claim that when men ignore rules or do not establish rules they are “exercising a common prerogative of power. Those with lesser power do not have the option to ignore the other’s rules, or common rules” (p. 391). In terms of topic development, Henley and Kramarae assert that “men’s tendency to make abrupt topic shifts, that is, to ignore basic conversational rules, like their tendency not to link to the previous utterance (even when on the same topic), may likewise be seen as a prerogative of power, the power to define and control a situation” (p. 391).

It is useful to distinguish between two kinds of topics in this study: global and local (see Chafe, 1994, for discussion of global or “supertopics”). While the global topic of OKLABOMB was the bombing itself, many local topics or threads developed about the bombing, such as speculation about suspects and information about victims. For the purposes of this study, a thread or local topic is defined as two or more posts written by different participants who are engaged in conversing about similar information. These messages may or may not use the same subject (reference) line, and the subject line may not be applicable to the content of the post. A local topic or thread may be about the global topic of the Oklahoma City bombing, or it may not.

The Internet has been praised by many as a place where people may gather in a virtual space, despite differences that might inhibit or prohibit face-to-face communication. It is a virtue with a sharp edge, however, because in order to establish and negotiate meaning, disparate people must find common ground. Complete breaks with the main topic prove to be frequent in OKLABOMB, as group members struggle to define their conversational domain and their relationships with one another.

Data and Methodology

The data for this investigation are 332 messages posted during the first nine days of an unmoderated public electronic discussion group, OKLABOMB, that was established one day after the April 19, 1995, bombing of a federal building in Oklahoma City, Oklahoma.3 This asynchronous discussion was selected for topic analysis for two reasons, the first being that it focused on a well-defined global topic, as established by the listowner. One week after the bombing, he reiterates to participants the purpose of the list:

This list is designed as a place where those interested in the Oklahoma Bombing can discuss the issues surrounding it. Any conversation related to the bombing is welcome. This list is unmoderated; any post sent in will be sent out immediately to all list members upon receipt. Unacceptable uses of these lists: Discussions unrelated to the bombing. (Posted Wednesday, April 26, 1995)

The second reason for choosing this particular electronic discussion was because although it started out as a dynamic forum for the exchange of up-to-the-minute information, within a few days it had degenerated away from its original purpose into a forum dominated by a handful of participants posting on marginally related local topics or unrelated subjects. After a month’s operation, when postings had declined to a trickle, the listowner asked subscribers whether the discussion group should be discontinued or expanded to other topics. Thus the group provides an opportunity to study the dynamics of both topic control and topic degeneration.

I employed a method of ethnographic observation that included subscribing to the OKLABOMB discussion list and observing its interactions, as well as gathering media reports related to the content of its messages. This was combined with conversation analysis focused on topic development, to determine how topics developed and expanded, changed into other topics, or were dropped altogether. Topic cues available to OKLABOMB participants included the rules of the discussion group mailed to each member who subscribed. These rules prohibited discussions unrelated to the bombing and public attempts to squelch unrelated discussions “simply because (list members) believe they shouldn’t be brought up.” (Participants were directed to complain about off-topic messages in private messages to the listowner.) “Irrelevant chain letters and forwards of unrelated material” and “advertising” were also discouraged by the rules.

I classified each of the 332 messages as being either on or off the global topic of the Oklahoma bombing, as well as being on or off the local topic indicated in its subject line. Messages were also analyzed by the total number of messages posted each day and the total number of different subject lines used by participants each day, to reveal both the extent of local topic shifting and also complete breaks with both global and local topics. The numbers of postings by women and by men (or by unknown sex) were tabulated, and a tally was kept of the number of participants who posted multiple messages in one day. In text-based CMC, gender is most obviously constructed by use of a name associated with being female or male. These cues for determining gender were equally available to all participants and were taken at “face” value; that is, a participant with a typically female name was classified as a female, and one with a typically male name as male.


The results show that the original topic—or global topic—of the discussion group (defined for this study as the bombing itself, along with local topics about those injured and killed, rescue efforts, suspects, and government response) degenerated by many “stepwise moves” or “parallel associations” among local topics. In addition, many messages contributed to the OKLABOMB discussion constituted full breaks with the global topic altogether, including conversation concerning the Branch Davidian compound in Texas, militia groups in the United States, and news media bias. The bombing itself became a minor thread among more dominant topics—including a metadiscourse among some participants about the disintegration of the discussion and the domineering tactics and offensiveness of some group members.

Plotting the number of each day’s messages that were on the global topic of the Oklahoma bombing provides evidence of the degree to, and the speed with which, OKLABOMB participants strayed from the stated purpose of the discussion (see Figure 1). For the purposes of Figure 1, messages that contained any connection to the bombing through local topics—speculation about suspects, information about the condition of other buildings in Oklahoma City, the death toll, media accounts about the bombing, the bombing’s impact on international politics, and so on—were considered to be “on” the global topic. By the third day, on-topic messages had dropped to one-third, and during the next six days of this nine-day period, on-topic posts averaged just above one-third, at 36%.

Figure 1. Percentage of discussion group messages on global topic

Participants adhered to the group’s purpose most closely (65%; 17 out of 26 posts) when the OKLABOMB discussion began on April 20. Off-topic posts on the first day included a request from a reporter to use a participant’s message and her reply, as well as several messages about events that took place in Waco, Texas, a year earlier4 and about Israel’s expertise in combating terrorism. The first on-topic messages were offers of sympathy to Oklahomans, as well as speculation on possible suspects and up-to-date information about the rescue efforts. One eye-witness account of the bombing was forwarded to the group by OKLABOMB’s listowner and served as a kind of ideal message type for the list, since it was so clearly within its purpose.

A decrease in participant adherence to the main topic of the bombing was often preceded by breaking news. Many OKLABOMB participants used the discussion forum as an alternative “grassroots” news medium, or as a supplement to major news media, much like others have utilized the Internet during crisis situations, including China’s Tianammen Square stand-off and the Kobe, Japan, earthquake in the 1990s, the attacks on New York City’s World Trade Center in 2001, the Virginia Tech shootings in 2007, and the Haitian earthquake in 2010. Just after 4 p.m. EDT April 20, 1995, on OKLABOMB’s first day, the listowner posted a message about the FBI issuing composite sketches of two white suspects (U.S. Attorney General Janet Reno had released the sketches at 1 p.m. EDT), and by 1 p.m. EDT on April 21, Reno announced that the FBI was holding Timothy McVeigh as a suspect in the bombing. With these two announcements, the attention of list participants shifted from the bombing itself to the motive, and then in stepwise moves to discussion of militias, civil liberties, government conspiracies, military disenchantment, journalistic malaise, gun control, and the future of American-based terrorism. April 22, the day after American suspects were arrested, shows a substantial drop in topic adherence; just 33% of posts (8 of 24) were on the global topic of the bombing or on its many related local topics.

One example of the stepwise moves among messages about possible suspects began with a general discussion on April 20. On this first day of the discussion list’s creation, many posts were related to the bombing suspects and whether they were international terrorists, Branch Davidians, militia members, or part of a government or anti-government conspiracy. Two examples from this thread follow:

In response to this local discussion about suspects, other participants begin to express uneasiness concerning generalizations being made about possible bombing suspects, in stepwise moves that begin to shift the topic away from specific, official information about suspects. Two of these responses are reproduced in examples (3) and (4):

While these two posts stay loosely on the topic of the Oklahoma bombing, they also project future threads and development of topics unrelated to the discussion group’s purpose, such as anti-government threads with no references at all to Oklahoma City’s bombing. Single oppositional exchanges, such as is illustrated in the quoting and response in example (4), “can encompass talk about a range of different subjects. Argument begun about one issue can escalate to include many points of contention between the parties,” as Goodwin and Goodwin (1990, p. 97) note. These researchers describe an intersection between topic and the organization of argument, where difference is expressed but still resides within “a coherent whole.” In fact, example (3) is an illustration of a participant using a local topic to address the global topic, an implicit reference to a coherent whole.

However, the conversation on OKLABOMB soon explodes the notion of a coherent whole, especially when fewer than half of all posts after April 21 attend to the discussion’s official purpose. Whereas the electronic discussion started basically as an information exchange, once a suspect was arrested and charged on April 21, the context of the discussion changed dramatically, as did the content of many of the messages, which became more argumentative and/or broader in scope. Another message illustrating the changing scope of messages on April 22 (the day that the first significant decrease in on-topic messages occurred) follows:

This message’s author, Jeremy, seemingly makes a connection to the bombing in his first sentence, but then he announces that he wants “to retract a bit (and) look at the big picture”—in other words, he chooses purposefully through metatalk to change topics. He moves from the threads in examples (3) and (4) about suspects and conspiracies to a new topic and asks “Does our future hold worse?” This is far from where the suspect threads began in examples (1) and (2). In response to Jeremy’s message, another participant never mentions the Oklahoma bombing at all, moving the thread even further away from its source:

Nonetheless, the total context of the Oklahoma bombing event, including news media reports and commentary, provides a somewhat logical frame for understanding example (6). Goodwin and Goodwin argue “that the organization of the talk is constrained and organized by its ethnographic context” (1990, p. 98), which includes setting, speech genres, the type of participants present, and the speech events themselves. In fact, all of the stepwise moves from example (1) to example (6) seem coherent within the context of the event. Yet, undeniably, a new thread about militia groups has arisen, through stepwise moves and parallel associations. Future messages concerning this thread and another anti-government thread about the Branch Davidians in Waco, Texas challenge and ultimately draw attention away from conversation about the bombing. When adherence to the original discussion topic drops again on April 26 and 27 (38% and 21%, respectively, of messages were on the main bombing thread), participants again use tactics to change the discussion’s direction while maintaining a token relationship to the discussion topic. Through phrases such as “given the incident in Oklahoma City ...” and “in the wake of the tragedy in Oklahoma City”—used by Bill in two messages on April 26—one participant goes on to break completely with the global topic and its many threads to begin off-topic discussions of “military follow-up of potential hatemongers” and a “letter in opposition to Omnibus Terrorist Act.”

Topic Derailment

Three messages from another participant, Dave, offer particular insights into the dynamics of the OKLABOMB list and its ultimate destiny. In each message, the author does not adhere to the group’s topic or to the subject line, but he tries to create the appearance that he is following the threads of conversation about the bombing by including a previous message by another subscriber with each of his own messages, to which he responds. In this sense he deliberately “misunderstands” and then appropriates past messages for his own agenda. (For further discussion of strategies of quoting previous messages in asynchronous CMC, see Severinson-Eklundh, this issue.) Dave also created and sent these posts within minutes of one another, in effect overloading the conversation with his viewpoint and preventing others from getting an electronic word in edgewise.6 The topic of Dave’s messages could be classified as anger about the outcome of the Branch Davidian siege, which is far afield from the Oklahoma bombing (emphasis added to time of all three messages):7

Examples (7) through (9) each have subject lines borrowed from other threads, but these have little or nothing to do with the new content of each posting. In terms of local topic development, these subject lines might be seen as misleading topic openers. Consider as well the following three examples, in which all messages use the same incorrect subject line borrowed from a previous thread, and all messages cohere with one another in terms of topic, even though that topic has nothing to do with the discussion’s global topic:

Some OKLABOMB participants registered dismay at the way subject lines did not match content expectations:

Of the 332 messages analyzed for this study, 253 subject lines (76%) have at least a thin connection to the content of their messages, while the others (24%) have no connection at all. Nonetheless, most participants negotiate the differences between subject lines and content, and many threads of conversation continue with all participants discussing the same issues, even though the subject lines no longer apply. The “Michigan Militia” messages above serve as good examples of this phenomenon, in which “a single expression may carry with it a complex mosaic of meaning” derived from the inferencing processes that accompany real-life conversations (Simpson, 1993, p. 132). In other words, these participants constructed new meaning, sometimes even conflicting meaning, that went beyond the surface meaning of the subject lines.

Eventually, participants’ metadiscursive comments on the disintegration of the discussion become a primary topic of conversation, another topic that does not fit within OKLABOMB’s stated purpose. In response to posts containing offensive language or ideas, participants were either outraged or intimidated. As illustrated in example (9), many participants expressed their indignation, and some also unsubscribed from the list. Others expressed less confidence in their own viewpoints in response to the polemical ideas contained in some messages by domineering group members. One example is a woman’s response to messages that link Oklahoma City’s bombing to the Branch Davidian siege in Waco, Texas, such as example (9). On April 28, she begins her post with a hedge: “I may be wrong, but I haven’t heard anyone suggest that there is a comparison between Waco and OK City” (Dierdra). Two other female members express their disgust with a message that had been forwarded to the list, written by someone angry at President Bill Clinton and his attempts at brokering peace in Northern Ireland. However, the first, example (14), is met with silence; the second complaint by a woman, included in example (15), is answered in an abusive way by a list administrator and the listowner himself:

Within a week, the listowner himself was warning participants to stay within the parameters of the bombing topic. On the discussion’s first day, however, he rejected complaints by those who wished to stay on topic or to avoid offensive material. He also presupposed in example (15) that the woman who complains is a “newbie,” or someone who has little Internet experience, while aggressively turning aside her comments and prohibiting her future participation by unsubscribing her from the discussion. Even though the listowner allowed almost any topic to be discussed, which was against his own rules, and seemed to eschew his own role in the hierarchy, he definitively asserts his authority in the above message. In other words, people who complain—in this case, women—will be censored. Another instance of a male denigrating the contribution of a female participant with an ad hominem attack is found in example (4). Generally speaking, male contributors denigrated females who questioned their assertions or who attempted to serve in peace-making or rule-enforcing roles (cf. Smith et al., 1997). In contrast, when female participants denigrated male contributions to the discussion, they did so because they considered these posts to be offensive or “sick.”

Powerful Participants and their Effect on Topic Development

Exchanges with the listowner (see examples 14-15) provide insight into the gendered interactions about topic on OKLABOMB. The male listowner and his male administrative associate at first offered no responses to women’s complaints about an off-topic and off-color message with the subject line of “re: O-KabOOOOOOOOOOOOma.” This precisely fits the pattern suggested by Henley and Kramarae (1994), who claim that men who ignore rules are “exercising a common prerogative of power” (p. 391). While the woman in example (14) seems mostly concerned with the obscenity used in the message, the woman in example (15) is clearly focused on the flouting of the global topic. Both seek rule enforcement from the listowner (and rule-maker), but these women are first ignored and then unsubscribed. More important to this study, crudeness and off-topic messages received a kind of sanction from the listowner, despite his own stated rules for the list. When these rules were not enforced, the purpose and topic of the discussion became less clear.

Other patterns of gendered interaction were also evident in the discussion, consistent with the findings of other studies of gender in CMC (Carstarphen & Lambiase, 1998; Herring, 1993; Hobman et al., 2002; Lambiase, 2001a). Foremost among these is the participation rate of men, who posted almost double the number of messages of women. Two hundred and six messages (62%) were posted by men, 112 messages (34%) were posted by women, and 14 messages were from participants of unknown gender. Moreover, men were in general more adversarial, more likely to flame, and more tolerant of adversarial or obscene discourse (men in OKLABOMB registered no complaints about the “O-KabOOOOOOOOOOOOma” message, for example), whereas women were less tolerant of rule-breaking and were ignored or unsubscribed when they did complain.

At the same time, active subscribers of both genders displayed participation styles that were aggressive, flamboyant, persistent despite flaming, and copious in number. These attributes may be tied to hegemonic male values and also to linguistic survival within a large group (see Hobman et al., 2002). The four women who participated the most—Kathy, Nancy, Audrey, and Tracy—posted 60 messages, more than half the total for messages by females. Their prolific output masks the under-participation of other women in the group. (In contrast, the four men who participated the most—Bobby, Dan, the listowner, and Charles—posted 53 messages, about one-fourth of the total messages for men.) These four women and their messages were at the heart of the discussion’s give-and-take, matching most of the men’s online communication style by using the same strategies to guide local topics to their interests. On April 27, Nancy contributed one-fourth of the total messages for that day (seven out of 28 overall), more than any other participant, and six of these seven are off the main topic of the discussion; on April 24, she contributed nine messages of 48, or 19% of the total, with four of these off-topic. Of Kathy’s 11 contributions on April 23, eight are not on topic; on the same day, Bobby contributed eight messages, the most of any male during the nine days studied, with half of those off topic. These most copious contributors quickly established their online identities to gain the floor by using the usually successful strategies of posting “clever or controversial messages” and “multiple messages” (Herring, this issue). The four women who participated most increased their odds of response by posting numerous messages (and in Kathy’s case, because she posted so many messages she received negative responses from men and women). Several men posted messages containing offensive language and extreme ideas, and these always received attention. Altogether, the four men and four women who participated more than any other subscribers were responsible for contributing 113 messages, more than one-third of the 332 messages analyzed for this study.

At the peak of the discussion group’s membership on April 22, 497 people had subscribed; by May 2, almost two weeks after the start of the discussion, that number had fallen to 337. By late May, the listowner posted a message asking for suggestions about whether the electronic discussion should be continued. One of the dominant contributors responded:

Two other respondents also replied, one writing “close it” and the other “I think it would be good to leave this list open until the apprehension and trial of the suspects are over.” By early summer, the discussion group did close, as by then political and media attention to the bombing and its suspects had subsided.

Almost from its inception, the OKLABOMB electronic discussion showed signs of discontent from many of its subscribers regarding topic management, even from its most frequent contributors. On April 23, Bobby wrote that “as long as we’re engaged in meta-discussions about the list, I would prefer that people not send tracts . . . . I want to hear what you folks are thinking about the events surrounding the Oklahoma City bombing (*Not* a call for censorship).” Kathy herself recognized when she did not adhere to the main topic, by responding with an emoticon on April 23 to someone who posted about attacks on an anti-abortion clinic. She joked, “Maybe it’s an artificial insemination clinic? ;-> Now back to our regularly scheduled topic.”

Three members who participated more moderately made comments on topic management as well, all posted on April 24. In response to another subscriber about his continued focus on Waco, Audrey wrote that “It’s about time someone edits this list, or ‘Kathy’s List,’“ and she also wrote that “the point of this list is to discuss OKC—not Waco.” Paul commented: “I spend my time trying to participate in a forum which discusses the Oklahoma bombing. If you want to flame people, don’t do it on the rest of our time. Write to them directly. Let’s just move on already, people.” And Timothy responded to a tirade about media censorship endangering American democracy, writing that “the Internet is world-wide, and that sort of comment can be seen as arrogance by (A)merica in the other 149 countries of the world. Care in language would be appreciated, on this list and others. Sorry to go off-topic, all.”

A male subscriber, who had complained about Kathy’s domineering tactics on April 22, and who started the chain of postings about censorship in examples (10-12), responded to the above-mentioned posts, writing that their concerns were valid:

up until one person starts drowning out everyone else. As long as Kathy doesn’t do that, I agree she should be welcome. I think I overreacted to the number of postings as much as anything else. But this is a worry—I’ve seen lists and newsgroups where one or two people take over and drown out the rest. Hopefully it won’t happen here.

His call for the listowner to “please shut Kathy up,” along with his more reasoned follow-up message, above, illustrates the dilemma of asking for constraints on over-participation while at the same time tolerating a kind of cyberlibertarianism where anything goes. The over-participation itself may silence more passive and overwhelmed participants, but the constraints on more copious participants may also censor them.

Discussion and Conclusions

Topic shifts of the kind experienced in the OKLABOMB electronic discussion list are normal and expected in any conversation; discussions in cyberspace are no different in this regard. Many electronic discussions stray from established global topics, and research has shown the difficulty of enforcing rules and topics in electronic newsgroups. As Kollock and Smith (1996, p. 122) note,

groups routinely wander off of their declared topics, and are frequently invaded by those who are either ignorant of the goals of the group or who actively seek to disrupt them. In some of these cases, groups have decided to deal with a social dilemma by turning over authority for the management of a collective good to a particular member or group of members, trusting these leaders to manage the resources well.

In their study of online communities, Johnson-Eilola and Selber (1996) examined the ways participants “learn and internalize discourse laws as they operate in online forums and how they learn the consequences of breaking these laws” (p. 270). In terms of topic development, they suggest that “generally agreed-to prohibitions on certain topics are viewed by most participants as a sign of civilization: Our lives are too complicated and the voices so many that we must agree to avoid discussion of certain topics because they are too complex or do not relate to the task at hand” (p. 283).

Hobman et al. (2002) found that in short-term CMC, early conflicts were related to later conflicts, causing a “conflict spiral,” and that large groups and male-female groups had more conflict than small groups and single sex groups. Their observations are relevant to the present study in several ways, especially as regards topic development. Participants flouted rules about the global topic early in the discussion, with no response from the listowner to complaints about rules or topic consistency. For a group of disparate people to come together and engage in a meaningful and democratic electronic discussion, an agenda is needed to guide their contributions and responses, and a listowner must follow through with threats of dropping those who do not adhere to the main topic or who break other rules.8 When a small group of participants chooses to flout that purpose by discussing subjects that are off-topic, so much so that fewer than half (and on April 27 and 28, fewer than one-third) of messages are on the main topic of the list, they do so at the risk of silencing participants and losing subscribers. Moreover, the overall rate of participation was low. Although there were nearly 500 subscribers shortly after the list was created, on April 23, the day with the largest number of messages (72), only 36 people posted messages. By May 2, when 337 people were still subscribed to the discussion, only 13 people posted messages that day. Even the listowner expressed wonderment at the large number of lurkers in the discussion (see example 12).

In addition to topic degeneration, other explanations for the rapid decay of the OKLABOMB discussion list include lessened media and public attention to the bombing event. This certainly contributed to OKLABOMB’s waning activity and participant ambivalence. Yet long-term, public discussions of national events are not impossible. In an Internet forum sponsored by a Chilean newspaper, La Tercera, 1,670 postings from more than 1,000 people were made over a two-year period concerning the arrest of Augusto Pinochet in London in 1998 (Tanner, 2001). In her study of these messages, Tanner found that the newspaper provided a vigorous, open forum about Chilean justice, truth, democracy, and history that was centered around Pinochet, his arrest, and his legacy. Journalists at the newspaper posted every email letter received, but at first edited each one for misspellings and obscenities; when this proved too time consuming, the forum became more like a synchronous chat room, in which every letter was posted “as is” with only obscene postings removed (Tanner, p. 388).

In OKLABOMB, obscene and inflammatory messages were allowed to coexist with postings that focused on the bombing and its aftermath. This “flaming” did affect topic coherence and perhaps also discussion viability. Online discussion breakdown can occur due to flaming, as illustrated by an event in August 2000. Two months before an election vote between U.S. senate candidates Hillary Clinton and Rick Lazio in New York, a readers’ forum at The New York Times’ website about that contest was shut down because discourse within the forum had become extreme and full of personal attacks against the candidates and among the participants’ themselves (Lambiase, 2001b). Many people find flaming strategies offensive and intimidating, while others justify them as an integral and, therefore, acceptable part of discussion. Yet in OKLABOMB, flaming and obscenities affected group dynamics, since some members (especially women) were offended, while other men and women seemed unaffected, nonplussed, or energized. Whatever its affect on participants, flaming usually causes more focus to be on the tone of the discourse, rather than on message content about a global topic or its local threads.

As this study illustrates, the issue of topic development often is tangled up with gender differences and/or powerful tactics used by some participants (considered by some to be based in male agonistic values). The tension between content and form was recognized early in the history of online communication, when Sutton (1996) observed that

the dominant group on the net (both in sheer numbers and in level of use), as in society in general, (is) male, and males see adversarial behavior as friendly; flaming is okay. Individualism is paramount, and the truth will eventually be shaken out of encounters built on direct conflict. (p. 181)

Moreover, dominant groups establish normative behaviors, so that “hierarchies determine whose version of the communication situation will prevail . . . whose language style will be seen as deviant, irrational, and inferior; and who will be required to imitate the other’s style in order to fit into the society” (Henley & Kramarae, 1994, p. 384).

In OKLABOMB, dominant group members shared the following characteristics: They were ready and willing to flame or to accept flaming by others, were liable to “misunderstand” previous discussions in order to appropriate them for their own agenda, and were likely to overload the group with multiple postings on a topic. These strategies became normative communication behaviors for many participants, but they exacerbated the disintegration of the global topic, and therefore, the discussion itself. Although these tactics worked to the advantage of a handful of participants for a few weeks, many more group members were not willing to flame or to risk being flamed by posting too many messages or by posting at all. Unlike the long-lived Chilean forum about Pinochet, this discussion was not able to achieve the conditions of a robust egalitarian community. Thus OKLABOMB demonstrates that dominant communicators can drive out the “good” in the marketplace of cyberspace by willfully straying from the topic more than they stay on topic and by being “bad” sorts of participants by valuing individualism over community concerns. The dreams of cyberphilosophers notwithstanding, egalitarian forums for democratic discourse are hardly possible under such conditions and in other contexts where hierarchies determine who speaks and who is silenced.


  1. These authors created a canon of their own work, similar to Jane Tompkins's observations about Nathaniel Hawthorne's reputation being a product of “political and social processes” (1985, p. 4). She writes that a canon evolves “because the groups that have an investment in them are culturally the most influential” (p. 5). All of these earliest authors writing about cyberspace, as well as many of their circle, are white males.

  2. Habermas's reformulation of Robert Alexy's rules of reason, which should be observed to ensure that discourse is democratic, is: “1.) Every subject with the competence to speak and act is allowed to take part in a discourse; 2a.) Everyone is allowed to question any assertion whatever; 2b.) Everyone is allowed to introduce any assertion whatever into the discourse; 2c.) Everyone is allowed to express his (sic) attitudes, desires, and needs; and 3.) No speaker may be prevented, by internal or external coercion, from exercising his (sic) rights as laid down in (1) and (2)” (from “Discourse Ethics,” qtd. in Herring 1993, p. 1).

  3. I analyzed all messages that I received during the first nine days of the OKLABOMB discussion.

  4. On April 19, 1993, a seige by the United States Bureau of Alcohol, Tobacco, Firearms and Explosives (ATF), the FBI, and the Texas National Guard of the compound of the Branch Davidian religious sect near Waco, Texas, ended in the deaths of the Branch Davidian leader and 82 other people, including children and four ATF agents. Some people blamed U.S. Attorney General Janet Reno, who had approved recommendations by the FBI to mount an assault, for the deaths (http://en.wikipedia.org/wiki/Waco_Siege, retrieved December 30, 2010).

  5. All email addresses have been deleted by the author, and names and other identifiers have been changed, except for the listowner, who identified himself in an interview with The New York Times (see note 8 for further information).

  6. This could also be caused by a list server program that bundles and sends messages sporadically, although the consistent content and tone of these three messages suggests otherwise.

  7. Of course, after Timothy McVeigh's conviction in 1997 based on prosecution assertions about his anger over the Branch Davidian seige, it is obvious after the fact that there was a link between the Oklahoma bombing and prior events in Waco, Texas. In the first nine days after the bombing, however, this connection was just speculation.

  8. Aaron Dickey was the listowner of OKLABOMB. Many listowners of online discussions also serve as moderators and control content, but Dickey did not. By not moderating contributions to the list, Dickey allowed participants to control topics and to flout the list's purpose, which he himself had established.

    Dickey carried many identities. He was founder of the OKLABOMB list; he was the “chew toy” of someone or something “named” Ariel; and he also used at least two other email addresses (one with the log-in name of Kiernan, and one of his accounts included a signature file identifying him as an employee of a news service). He explained in a message to list subscribers that he was not a working member of the media, just in support services; this explanation followed flaming by participants of any reporters who might be lurking on the discussion, and he quickly denied that this included him. However, in an interview with The New York Times that appeared in the print edition on April 25, 1995, Dickey was identified as “a freelance writer in New York who contributes to Wired magazine ... (and) runs a couple of small listservs for 'generation Xer's'—people in their 20s” (Miller, p. A21).


Bolter, J. D. (1991). Writing space: The computer, hypertext, and the history of writing. Hillsdale, NJ: Lawrence Erlbaum and Associates.

Carstarphen, M. G., & Lambiase, J. (1998). Domination and democracy in cyberspace: Reports from the majority media and ethnic/gender margins. In B. Ebo (Ed.), Cyberghetto or cybertopia? Race, class, and gender on the Internet (pp. 121-135). Westport, CT: Praeger Publishers/Greenwood.

Chafe, W. L. (1994). Discourse, consciousness and time. Chicago: University of Chicago Press.

Dorval, B. (Ed.). (1990). Conversational organization and its development. Norwood, NJ: Ablex Publishing.

Ess, C. (1994). The political computer: Hypertext, democracy, and Habermas. In G. Landow (Ed.), Hyper/text/theory (pp. 225-267). Baltimore: The Johns Hopkins University Press.

Fairclough, N. (1992). Discourse and social change. London: Polity Press.

Gattiker, U. E. (2001). The Internet as diverse community: Cultural, organizational, and political issues. Mahwah, NJ: Erlbaum.

Goodwin, C., & Goodwin, M. H. (1990). Interstitial argument. In A. Grimshaw (Ed.), Conflict talk: Sociolinguistic investigations of arguments in conversations (pp. 85-117). New York: Cambridge University Press.

Goutsos, D. (2005). The interaction of generic structure and interpersonal relations in two-party e-chat discourse. Language@Internet, 2, article 3. Retrieved December 30, 2010 from http://www.languageatinternet.de/articles/2005/188/index_html/

Grossman, L. (1996). The electronic republic. New York: Viking.

Henley, N., & Kramarae, C. (1994). Gender, power, and miscommunication. In C. Roman, S. Juhasz, & C. Miller (Eds.), The women and language debate (pp. 383-406). New Brunswick: Rutgers University Press.

Herring, S. C. (1993). Gender and democracy in computer-mediated communication. Computer-Mediated Communication, special issue of the Electronic Journal of Communication, 3(2). Retrieved December 30, 2010 from http://ella.slis.indiana.edu/~herring/ejc.txt

Herring, S. C., Ed. (1996). Computer-mediated communication: Linguistic, social and cross-cultural Perspectives. Amsterdam and Phildelphia: John Benjamins.

Herring, S. C. (1999). Interactional coherence in CMC. Journal of Computer-Mediated Communication, 4(4). Retrieved December 30, 2010 from< href="http://jcmc.indiana.edu/vol4/issue4/herring.html

Herring, S. C. (2003). Dynamic topic analysis of synchronous chat. In New Research for New Media: Innovative Research Methodologies Symposium Working Papers and Readings. Minneapolis, MN: University of Minnesota School of Journalism and Mass Communication. Retrieved December 30, 2010 from http://ella.slis.indiana.edu/~herring/dta.2003.pdf

Herring, S. C., & Nix, C. G. (1997). Is “serious chat” an oxymoron? Academic vs. social uses of Internet Relay Chat. Paper presented at the American Association of Applied Linguistics, Orlando, FL, March 11. Retrieved December 30, 2010 from http://ella.slis.indiana.edu/~herring/aaal.1997.pdf

Hobbs, J. R. (1990). Topic drift. In B. Dorval (Ed.), Conversational organization and its development (pp. 3-22). Norwood, NJ: Ablex Publishing.

Hobman, E., Bordia, P., Bernd, I., & Chang, A. (2002). The expression of conflict in computer-mediated and face-to-face groups. Small Group Research, 33(4), 439-465.

Johnson-Eilola, J., & Selber, S. (1996). Policing ourselves: Defining the boundaries of appropriate discussion in online forums. Computers and Composition, 13, 269-291.

Kollock, P., & Smith, M. (1996). Managing the virtual commons: Cooperation and conflict in computer communities. In S. C. Herring (Ed.), Computer-mediated communication: Linguistic, social and cross-cultural perspectives (pp. 109-128). Philadelphia: John Benjamins.

Lambiase, J. (2001a). Like a cyborg Cassandra: Hacking into the internet’s misbegotten rhetorical situation, in L. Gray-Rosendale and S. Gruber (Eds.) Alternative rhetorics: Challenges to the rhetorical tradition (pp. 111-125). Albany, NY: SUNY Press.

Lambiase, J. (2001b). Media, readers, and politics: Communicative components of letters to the editor and online forums. Presented at Internet Research 2.0, Minneapolis, MN, October 12.

Landow, G. (1993). Hypertext: The convergence of contemporary critical theory and technology. Baltimore: Johns Hopkins University Press.

Lanham, R. (1993). The electronic word: Democracy, technology, and the arts. Chicago: The University of Chicago Press.

McLaughlin, M. L. (1984). Conversation: How talk is organized. Beverly Hills: Sage Publications.

Miller, S. (1995, April 25). The explosion echoes in cyberspace. The New York Times: A21, 1.

Negroponte, N. (1995). Being digital. New York: Knopf.

Palen, L., Vieweg, S., Sutton, J., Liu, S. B., & Hughes, A. (2007, October 7-9). Crisis informatics: Studying crisis in a networked world. Third International Conference on e-Social Science, Ann Arbor, Michigan. Retrieved September 16, 2010 from http://citeseerx.ist.psu.edu/viewdoc/download?doi=

Poster, M. (1990). The mode of information: Poststructuralism and social context. Chicago: University of Chicago.

Rollman, J. B., Krug, K, & Parente, F. (2000). The chat room phenomenon: Reciprocal communication in cyberspace. Cyberpsychology and Behavior, 3(2), 161-166.

Reichman, R. (1990). Communication and mutual engagement. In B. Dorval (Ed.), Conversational organization and its development (pp. 23-48). Norwood, NJ: Ablex Publishing.

Sacks, H. (1972). Unpublished transcribed lectures. University of California, Irvine.

Scholz, T. (2008). Market ideology and the myths of Web 2.0. First Monday, 13(3). Retrieved September 10, 2010 from http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/viewArticle/2138/1945

Selfe, C. L., & Meyer, P. R. (1991). Testing claims for on-line conferences. Written Communication, 8(2), 163-192.

Simpson, P. (1993). Language, ideology and point of view. New York: Routledge.

Smith, C. B., McLaughlin, M. L., & Osborne, K. K. (1997). Conduct controls on Usenet. Journal of Computer-Mediated Communication, 2(4). Retrieved September 16, 2010 from http://jcmc.indiana.edu/vol12/issue4/smith.html

Sutton, L. (1996). Cocktails and thumbtacks in the Old West: What would Emily Post say? In L. Cherny & E. R. Weise (Eds.), Wired women: Gender and new realities in cyberspace (pp. 169-187). Seattle: Seal Press.

Tannen, D. (1990). You just don’t understand: Women and men in conversation. New York: William Morrow.

Tanner, E. (2001). Chilean conversations: Internet forum participants debate Augusto Pinochet’s detention. Journal of Communication, 51(2), 383-403.

Tidwell, L. C. & Walther, J. B. (2002). Computer-mediated communication effects on disclosure, impressions, and interpersonal evaluations. Human Communication Research, 29(3), 317-348.

Tompkins, J. (1985). Sensational designs: The cultural work of American fiction, 1790-1860. New York: Oxford University Press.

Biographical Note

Jacqueline Lambiase ( j.lambiase "at" tcu "dot" edu) is an associate professor of strategic communication in the Schieffer School of Journalism at Texas Christian University. Her published research uses qualitative methods to scrutinize gendered media images, public relations ethics, rhetorical strategies for branding, and social media.


Any party may pass on this Work by electronic means and make it available for download under the terms and conditions of the Digital Peer Publishing License. The text of the license may be accessed and retrieved at http://www.dipp.nrw.de/lizenzen/dppl/dppl/DPPL_v2_en_06-2004.html.