SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

This article presents a method based on Jones' “virtual settlement” theory for testing empirically for the presence of virtual community in Internet Relay Chatting (IRC), The conditions for virtual community proposed by Jones are related to the technological context of IRC and formulated as conceptual hypotheses. The author argues that sustained level of co-appearance and nickname stability should be included in testing. Interactivity analysis should include both verbal exchanges and action-simulating messages. Analysis of message references should be done in terms of message content as well as message syntax. Major issues related to research design and implementation are discussed in depth.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

The term “virtual community” is frequently used to refer to the general appearance and gathering of people on the Internet (Hagel & Armstrong, 1997; Hof, Browder, & Elstrom, 1997), in spite of the fact that researchers have difficulty coming up with a generally acceptable definition (Alstyne & Brynjolfsson, 1997; Jones, 1995; Jones, 1997; Smith, 1992; Weinreich, 1997). While the concept of “virtual community” is still an issue of much debate, an even more fundamental problem is what exactly constitutes a “virtual community” and whether the current activities on the Internet can be considered as a “community” in the sociological sense. The problem of virtual community existence was directly and systematically treated the first time in Jones’ work (1997).

Most of the published works on virtual communities are theoretical or philosophical. The few domain-specific empirical studies are merely description of field observation at best. In these studies, a virtual community is generally assumed to be already existing in a specific technical domain, and this assumption is not subjected to any empirical test using a quantitative approach (Baym, 1995; Erickson, 1997; Fernback & Thompson, 1995; Johnston, 1997; Kollock & Smith, 1994; Newhagen & Rafaeli, 1996; Parks & Floyd, 1996; Rheingold, 1993, 1994; Reid, 1991; Turkle, 1996). Jones’“virtual settlement” theory (1997) presents specific conditions for the existence of virtual community. This article uses these conditions to develop a corresponding empirical test in specific technical settings.

Empirical testing for the presence of virtual community must be done in the context of a specific technical domain such as newsgroups, Internet Relay Chatting (IRC), Multi-User Dungeons (MUDs or MOOs), et cetera. This is because virtual communities, if they do exist, reside in specific technical domains of Internet-based group communication. Different technical settings of group communication on the Internet have different supporting mechanisms and communication processes. They do not work the same way in channeling communication messages. Virtual communities in one technological domain may exhibit different characteristics from communities in other domains, and the testing parameters and implementation techniques may therefore need to be altered to accommodate domain differences.

An initial investigation has to start somewhere in a specific domain, and I choose to start with IRC. IRC is a multi-user synchronous communication facility that is available all over the world to people with access to the Internet. An extensive description of the technical and social characteristics of IRC can be found in Danet (1995), Danet, Reudenberg, & Rosenbaum-Tamari (1996), and Reid (1991). The rather simple functionalities of IRC systems define a basic technical framework which has been serving as a wonderful playground for group activities and has provided many possibilities for self-expression and group interaction (Bechar-Israel, 1995; Danet, 1995; Danet, Reudenberg, & Rosenbaum-Tamari, 1996; Reid, 1991; Turkle, 1995). Although the opportunities for virtual community development in IRC are abundant, there has been little research in this vein done specifically on IRC-based group communication.

In this article, we explore the possibility of testing empirically for the presence of virtual community in the environment of IRC, basing the constructs and measures on Jones’“virtual settlement” theory (Jones, 1997). First, we introduce Jones’ theory and relate the proposed conditions for virtual community to the technical context of IRC to derive conceptual hypotheses. Next, we discuss the construct measurement, data manipulation, and parameter setting that are required to make statistical hypothesis testing possible. Finally, we address some major issues related to research design and implementation.

The importance and relevancy of this research lies in that it takes the first step, though a small one, towards putting the often readily accepted assumption of virtual community existence to empirical test. It directly confronts the common problem in current CMC research of simply equating various forms of group-CMC to “communities,” as noted by Jones (1997). Recognizing that a CMC environment has the potential to nurture the development of a virtual community is fundamentally different from taking any group CMC as evidence of community without justification. Any attempt at characterizing a virtual environment and analyzing its online activities without first comparing its properties against empirical signposts of community is methodologically flawed. Although this work does not deal directly with distinguishing virtual communities in IRC from those in other technical settings, it has strong implication for such studies.

The “Virtual Settlement” Theory

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

After reviewing the previous research on virtual and real-life communities, Jones (1997) argues that virtual communities are more than just a series of computer-mediated communication (CMC) messages. A community and its material – cultural artifacts and physical traces created and left behind by the community – need to be examined in order to determine when an aggregate of computer-mediated messages meets the requirements of community. CMC researchers should take an archaeological approach to studying virtual communities, systematically exploring the physical traces of its activities, and examining the cultural artifacts thus created.

Using the term “virtual settlement” to refer to the “cyber-space” (virtual environment) within which a virtual community operates, Jones explains that virtual settlement is a prerequisite condition for the emergence and existence of a virtual community, and that “the existence of a virtual settlement is proof of the existence of a related virtual community” (Jones, 1997).

Jones further stipulates that for a cyber-place with associated group-CMC to be considered as a virtual settlement, it is necessary for it to meet a minimum set of four conditions. They are: (1) a virtual common-public-space where a significant portion of interactive group-CMCs occur; (2) a variety of communicators; (3) a minimum level of sustained stable membership; and (4) a minimum level of interactivity.

My intention is to test for the presence of virtual community by determining empirically if messages generated by participants in an IRC channel meet these minimum conditions. It is unclear from Jones’ article (1997) that these minimum conditions are necessary but not sufficient. In other words, if a series of messages in an IRC channel meets these conditions, these messages may still not be sufficient to constitute a virtual settlement. However, if the series of messages does not meet the conditions, it can definitely not be considered as a virtual settlement. Thus, these conditions can be used to help locate a virtual community by eliminating unqualified virtual places.

Conditions of Virtual Community Existence in IRC

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

We now examine the four conditions of virtual settlement proposed by Jones in light of the technical characteristics of IRC. Note that the subheadings in this section do not always correspond to Jones’ four conditions.

Virtual Common-Public-Space

In Jones’ theory, “common public space” refers to a symbolically delineated place, a virtual space shared by participants to interact and to form relationships. It is easy to see that IRC channels provide such common-public space for interaction and that this condition is met.

The word “public” was not explicitly defined in Jones’ article. In our opinion, it simply means that the shared place is open to everyone in the community and that everyone can come to the place to interact with others. It does not imply that the place is open to everyone outside of the community. Just as a community in “real life” can exclude individuals who possess certain characteristics or properties, a virtual community can purposefully ban others from joining. In IRC, owners (operators) of a channel can make the channel an “invitation-only” place to prevent unwanted individuals from joining. A user can be kicked out of a channel or even banned permanently. But for the remaining members of the community, the space is still public.

Neither does the word “public” suggest that interaction between two participants has to be always visible to everyone there. In a real-life community, members can interact in public places and make their interaction observable to others, but they can also choose to associate in a private setting. Similarly, members of a virtual community may communicate in the shared virtual place where everyone there can observe the interaction, or they can choose to talk privately using the direct messaging function of IRC. Just as private communication in real life is essential for relationships to form in a community, private exchanges between participants in the same IRC channel play a similar role in relationship development and should be considered as part of the activities of the virtual community existing in that channel.

In Jones’ article (1997), it was made clear that “private communication where postings go directly from one individual to another with no common virtual-place” (emphasis added) ought to be excluded from consideration of virtual settlement. However, this is not the case when two participants in an IRC channel send private exchanges to each other while chatting in the same channel. Such private communication happens within a common virtual-place, and therefore, should not be excluded on the ground of Jones’ formulation of the condition.

Variety of Communicators

The second condition – a variety of communicators – is also easy to meet, provided that we focus our attention on well-attended, long-existing channels and ignore those short-lived “ghost towns” (channels) where only a very small number of participants are present.

My interpretation of the second condition with respect to IRC is slightly different from Jones’ original definition. In Jones’ theory, having a variety of communicators means that a virtual place ought to have more than two participants for any meaningful interaction to occur. Jones explained that “authors universally relate the term to group-CMC where there are more than two communicators” and that this condition is needed to “exclude most database queries and database interactions” (Jones, 1997). While Jones’ definition was focused on the minimum number of participants required for interaction, my interpretation extends this condition in two directions. First, the minimum number of participants is increased to exclude groups of very small size. Second, the virtual place (channel) has to be stable for a significantly long period of time. This extension is made for the following reasons.

In preliminary observations, it was not unusual to come across IRC channels that had only three or four participants. These participants stayed on the channel around the clock every day and never posted anything. Later, we learned that some were computer programs (“bots”) and others were users who signed up to multiple channels. These users were actually carrying out their chatting business somewhere else although their nicknames were shown in the channel. It makes sense to exclude such channels from a virtual community study.

Even if all of the participants whose nicknames are displayed are human beings and they are actually interacting with each other, it is still questionable to consider such a small group as a community. One certainly would not consider three or four individuals who regularly meet in a nightclub or a coffee shop as a community in the usual sense. It is difficult to determine exactly how many people a group has to include in order to qualify as a community. In addition, one may argue that a virtual community is different from a “real life” one and that it ought to be treated differently in terms of the required group size. Nevertheless, it is fair to say that for a group to be considered as a community, it must have a large number of participants to allow some interaction to happen – not just one or a few incidents of interaction between only two individuals, but a significant level of interaction among many individuals.

To understand why shortly-lived channels should be excluded from virtual community research, one needs to recognize that in IRC anyone connected to a server can open a channel by trying to join a channel that does not exist. (IRC servers do not allow two channels to have exactly the same name.) Although the owners/operators of a channel can change its topic line, no one can change the name of a channel once the channel comes into existence1. The channel is automatically closed by the server after the last participant leaves. “Bot” programs are often used to keep the channels alive when everybody is gone, so that when participants come back, they can always go to the old place to find each other.

What is the implication of a channel disappearing after a short existence, for virtual community research? One possible explanation is that the channel has fallen apart and the associated group is dissolved – meaning that no interpersonal bond has developed from the brief encounter and that participants will no longer interact with each other as a group. In this case, exclusion of such channels from consideration is justified.

An alternative explanation could be that, in spite of the disappearance of their shared virtual space, participants in the group have developed interpersonal bonds and will continue their activities in another virtual place or in face-to-face interaction or both. They may come back later to meet in IRC again and interact as a group in a different channel. Or they may revive the channel when they come online again. That is, whoever gets there first, not aware of the fact that the channel no longer exists, will try to join the channel anyway, and his or her joining command will cause the server to recreate the channel under the same old name.

The former scenario of a group of IRC participants migrating between different channels is unlikely, since it works against the purpose of coming back online to meet each other. The latter scenario of a channel existing only as long as the associated group is present and being revived by the same group again each time they come back is interesting. Although each incident of existence of the channel is brief, it ought to be considered as a stable and relatively long-existing channel. A large enough window of observation will capture its prolonged but interrupted existence, and therefore, the virtual community history of the associated group. For this reason, we conclude that such channels are not excluded by our interpretation of the second condition as proposed by Jones.

Boundary of Virtual Community

Apparently, Jones’ proposal of the third and fourth conditions – sustained stable membership and minimum level of interactivity – was made with reference to one specific virtual community. However, it is not clear how the boundary of a virtual community can be determined in IRC, and we need to address this issue before examining these conditions in depth.

Jones argued that “an IRC server containing hundreds of unrelated channels would not indicate the existence of a single virtual community” and that “a single channel or a small collection of channels could” (Jones, 1997). While it is evident that a single channel may indicate the existence of a single virtual community, the suggestion of several channels indicating the existence of one single virtual community is not convincing.

As a matter of fact, we do not think it is practical or likely that a single community would carry out its activities in several segregated groups situated in different channels. The whole population on an IRC network is organized into channels. An individual has to be on a channel in order to post and receive messages from the channel, and it is impossible for anyone to post to an unsubscribed channel. The possibility of public interaction across channels is prohibited by the technical mechanism of the IRC system. By organizing the IRC population into channels and by limiting visibility of posted messages to the scope of specific channels, the possibility of one single virtual community either emerging from several simultaneously existing IRC channels or being segregated into several channels is greatly reduced.

Thus the boundary of a virtual community in IRC should be constituted by the channel where it resides. This does not, of course, exclude the private exchanges that arise as natural extensions of the community's group communication.

Sustained Stability of Membership

If the boundary of a virtual community in IRC is determined by the channel where it resides, then the membership of the virtual community can be approximated by participants’ subscription to that channel. Although a member of the candidate population has to be a member of the channel, an individual who joins the channel does not automatically become a member of the community. For example, a user who joins a channel but does not initiate or participate in any communication activities may not be considered a member. Testing of membership stability should include only those who demonstrate a certain level of participation.

At first look, limiting membership testing to only those who communicate may seem questionable. One may argue that lurkers are part of the community, and therefore, should be included in the testing. However, the issue here is not whether a lurker should be considered as part, theoretically, of an existing community, but rather how to determine empirically the existence of a virtual community by testing its membership stability. A group of “lurkers” who do not communicate can not be called a community. For a group of individuals to qualify as a community, these individuals have to communicate and interact. A lurker may be considered as part of a virtual community only if the community exists, and the existence of a virtual community is not defined by the presence of lurkers, but by those who do communicate and interact. For this reason, testing for the presence of virtual community (or rather testing for membership stability in this case) should be done on those who communicate, not on those who just lurk.

On the other hand, testing for membership stability and other conditions has to be done on the same group of participants. If we include lurkers in the test population, the population will be made of two distinct groups – those who communicate and those who do not. There is a possibility that those who communicate may not maintain stable memberships and those who lurk may maintain stable memberships but do not communicate. When this occurs, the results of empirical testing will be misleading, and the conclusion we draw from such testing regarding the existence of a virtual community will be wrong.

In Jones’ theory, sustained membership stability has two dimensions which, in the context of IRC research, can be referred to as the qualitative and quantitative aspects. With respect to the latter, a channel should always require a significant number of active participants for a virtual community to emerge. With respect to the qualitative dimension, the candidate population should consist of approximately the same group of participants for a meaningfully long period of time. In other words, while individuals come and go, there should be one or several core groups of participants who appear and stay active on the channel continuously for a sustained period of time. Apparently, in testing sustained stability of membership, we need to monitor not only the total number of participants in a channel throughout the period of observation, but also the identities of participants, the number of sessions of their participation, and the duration of participation in each session.

It is easy to see that the essence of “sustained stable membership,” as a condition of virtual community in IRC, is not continuous participation of individual users in a channel, but the co-appearance of a significant number of participants over a period of time. Unlike newsgroups, where interaction takes place through asynchronous posting of messages, participants in IRC have to co-appear to make any interaction possible. The co-appearance has to be stable enough to allow sufficient opportunity for interpersonal interaction to occur and for interpersonal connections to develop. For the sake of argument, we may say that, theoretically, it is possible for a large group of participants to maintain stable membership on a channel but not have sufficient sustained co-appearance to allow meaningful interpersonal connections to materialize.

Therefore, we suggest that the condition of “sustained stable membership” be supplemented with “sustained level of co-appearance”. The notion of “sustained level of co-appearance” has three dimensions. First, the candidate population should have a significant number of clusters of participants whose co-appearances on a channel show stable patterns. Second, the size of such clusters (i.e., number of participants in a co-appearance group) should be reasonably large. Finally, the patterns of co-appearance should be sustained through a significantly long period of time. It seems plausible that the larger the size of co-appearance clusters and the longer the patterns of co-appearance persist, the greater the opportunities are for interpersonal connections to form, and therefore the more likely a virtual community is to develop.

The issue of co-appearance is essentially related to the dynamics of initial interaction and relationship development in naturally formed non-task groups. An exhaustive literature review failed to locate any research directly addressing our concern. The existing studies on group development and group dynamics are generally focused on task-oriented, consensus-seeking groups, which can not be of much help. There are reasons to believe that naturally formed non-task groups are fundamentally different from those formed to achieve a goal or to make a decision. The literature on interpersonal relationship and initial interaction is for the most part devoted to the study of dyadic relations between fixed pairs (Barker, 1991), and studies of spontaneous interpersonal relationship in naturally formed groups are rare.

Nevertheless, findings from the research on initial interaction between strangers may be useful in this case (Berger & Kellermann, 1983; Douglas, 1987, 1990, 1994; Martin & Rubin, 1998; Redmond & Vrchota, 1997). For instance, Douglas (1990) measured subjects’ levels of uncertainty at various intervals of initial interaction and found that the uncertainty scores at the close of a six-minute interval were significantly lower than the scores of those who met for shorter periods. He also found that more questioning occurred during the first two minutes of initial interaction. Redmond & Vrchota (1997) observed that a large percentage of subjects changed in the level of attraction between two assessment periods six minutes apart. Their findings seem to suggest that the initial encounter between two strangers needs to last at least for four to six minutes for an interactant to gather enough information to decide if he or she wants to continue interacting with the other. Thus, we can reasonably assume that co-appearances of IRC participants have to be at least four minutes long to be meaningful for any prospect of relationship development. Furthermore, computation of mean ratios of co-appearance should only include co-appearances longer than four minutes.

Sustained Level of Interactivity

Jones (1997) argues eloquently that a minimum level of (sustained) interactivity is a critical pre-requisite for virtual community. There is no question about the relevance of this condition. However, in order to test this condition, we need to have a good understanding of interactivity in the specific context of IRC.

The issue of interactivity in computer-mediated communication has been treated extensively by researchers. Quoting others (Rafaeli, 1988; Rafaeli & Sudweeks, 1997; Sudweeks & Rafaeli, 1994), Jones explains that interactivity is the extent to which later messages in a sequence relate to each other, and especially the extent to which later messages recount the relatedness of earlier messages. This explanation is based on Rafaeli's work, and we will review Rafaeli's treatment of this issue later. But for now, it is sufficient to point out that in the context of computer-mediated communication, this definition does not refer to general interactive behavior of participants, but only to the content nature of verbal messages. (By “verbal messages,” we mean messages as spoken words, not messages as whatever is keyed in and appears on the computer screen.) The fact that this definition is limited to verbal messages is clearly shown in the statement that “[I]nteractivity describes and prescribes the manner in which conversational interaction as an iterative process leads to jointly produced meaning” (Rafaeli & Sudweeks, 1997).

However, IRC communication is not made up of verbal messages alone. In IRC, unlike some other CMC environments such as newsgroups, it is possible for participants to “talk” to others as well as to “act” out imagined actions – actions that would have been conducted by their “virtual being” should they have a body of flesh and bones in the virtual space. This online behavior of “acting” is known as action simulation.

Two techniques of action simulation have been identified in computer-mediated communication. One is by enclosing an action-describing verb with asterisks such as “*inhales deeply*” (Danet, Reudenberg-Wright, & Rosenbaum-Tamari, 1996). This technique of action simulation and its effects were thoroughly described in the authors’ analysis of a virtual party that took place in IRC. The other is by issuing certain commands to cause lines like “Jane waves!” to appear on the computer screen (Danet, 1995). The technique was mentioned in connection with character manipulation (verbal puppetry) in MUDs and MOOs.

Actually, the second technique of action simulation is also used in IRC to achieve similar effects of “verbal puppetry” as in MUDs and MOOs. In IRC, such “acting” is made possible by certain commands such as “/describe” and “/me” etc. For example, if a participant nicknamed “BigBoy” enters the command “/me presents a rose to PrettyGirl”, the message will appear as “*** BigBoy presents a rose to PrettyGirl”, to everyone on the channel.

Although both “talking” and “acting” are presented as lines of texts on the computer screen, the generated messages are of different nature, and they represent different kinds of interaction in the virtual space. Analysis of interactivity in IRC communication should be done not only on “talking” messages, but on “acting” messages as well, and these two kinds of interaction should be treated differently. For ease of discussion, from this point on, we will refer to the former as verbal messages and the latter as action-simulating messages.

Interactivity of Verbal Messages

We now focus on the interactivity issue of verbal messages. Since Rafaeli's work (later in cooperation with others) specifically deals with interactivity of verbal messages, we start by reviewing Rafaeli's contribution in this area.

Interactivity was first defined as a formal communication concept by Rafaeli in 1988 as an attempt to characterize communication of various forms and to bridge mass communication and interpersonal communication (Rafaeli, 1988). Although the concept was intended to be general and independent of specific communication forms or media, Rafaeli's effort of explicating this concept was mainly based on consideration of computer-based technologies, and therefore, of computer-mediated communication. Further, it seems that his effort was tilted toward verbal messages as neither his earlier or later works mention non-verbal messages generated in computer-mediated environments (Rafaeli, 1986; Rafaeli, 1988; Rafaeli & LaRose, 1993; Sudweeks & Rafaeli, 1994; Rafaeli & Sudweeks, 1997).

Initially, interactivity was defined as “an expression of the extent that in a given series of communication exchanges, any third (or later) transmission (or message) is related to the degree to which previous exchanges referred to even earlier transmissions” (Rafaeli, 1988). This definition was later restated in a slightly different form in an article written jointly by Rafaeli & Sudweeks (1997). To further clarify the concept, Rafaeli (1988) identified three levels of communication in terms of increasing degrees of interactivity. They are: two-way non-interactive communication, reactive (or quasi-interactive) communication, and fully interactive communication. These categories were later applied in Rafaeli's work (in cooperation with others) to investigate interactivity in CMC, specifically, to classify newsgroup messages (Rafaeli & Sudweeks, 1997).

Rafaeli (1988) defined interactivity as the relatedness of message content, not as the process of communication or as the characteristic of a communication medium. Consider telephone conversation as an example. The physical process of bi-directional flow of messages through the telephone wire does not, for Rafaeli, demonstrate interactivity. Neither does the process of turn-taking as required by the mechanism of dialogue and enforced by the physical properties of the medium. For the same reason, one person's act of directing his or her message to the other person – meaning “I am talking to YOU”– shall not be thought to demonstrate interactivity. Rather, interactivity is shown by what one person says about what the other person said earlier and how that was related to what was said even earlier.

While content interactivity is essential and Rafaeli's definition stands in its own right, we argue that by discounting message directing/targeting along with media- and process-related characteristics, an important aspect of group interaction is neglected and the power of this concept is weakened. In cases such as radio broadcasting and telephone conversation where a message has to be intended for and directed to the other person(s) by default, message directing does not carry any significance. However, the same can not be said about group communication, computer-mediated or not. In cases where the speaker has a choice of whether to speak indiscriminately to all of the audience or specifically to one particular individual, message directing assumes much significance as an indication of in-group interaction. It becomes an effective indicator of what sorts of interpersonal relations have formed within the group, what power relations are present, and how strong the within-group interpersonal bonds are.

Take newsgroup messages – one kind of computer-mediated group communication – for example. A message may refer to what someone said in the previous message, even mention that person's name, but not be specifically directed to anyone in the sense of “I am talking to YOU”. Or in contrast, the message may contain lines like “Jane Doe, did you mean …?”. Message targeting in newsgroups may not be as obvious as in IRC (which is to be explained later) since the intended audience is often defaulted to the whole group. Nevertheless, this aspect of group interaction was noted in a recent study on interactivity in newsgroup communication by Rafaeli and Sudweeks (1997). Among the coding items used to analyze newsgroup messages, one item (COALT3, line 39 in the Appendix) was included to characterize whether a message directly addresses any person on the newsgroup list. If we understand correctly, this item refers to message directing. Even though this factor was included in their analysis, the article did not explain how it relates to interactivity.

It is contended here that these two aspects of group interaction in CMC – message directing and relatedness of message content– are distinct, and that they are equally important for understanding interactivity in group communication. Analysis of interactivity in computer-mediated group communication should therefore include both aspects. In the following two sections, we will examine these two aspects in greater depth.

Message Directing/Targeting

In IRC, participants may direct their messages to the whole group or to a subgroup by posting the messages indiscriminately to the public forum. Or, they may target their messages, still in the public forum, to one particular individual by including that person's nickname at the start of the line. A participant, upon reception of a message directed to him or her, may choose to respond either by sending a private message or by posting a message in the channel and specifically targeting the message at the other person. The practice of addressing messages to a specific participant in the public forum by including the person's name at the start of message line, in combination with message references in content, creates complicated patterns of interaction. For instance, a message can be targeted at one specific person and refer to what that person said in his or her previous message. On the other hand, a message can be specifically targeted but not refer in any way to anything that was said in the past, or it can be directed to one person and refer to something said by another.

Analysis of message directing aims at revealing surface patterns of participants’ interaction within any episode of IRC communication. Each line of IRC exchange can be classified as either targeted (directed to a specific recipient) or untargeted (not directed to any particular individual). Targeted messages can be further classified as unidirectional (directed but not responded to by the targeted recipient) or dyadic (responded to by the targeted recipient). Based on such simple coding, it is possible to identify higher-order patterns of message directing such as dyads, triads, and quadruples among others. Such patterns indicate the intensity of within-group interaction at the surface level.

Interactivity in Message Content

Rafaeli's three levels of communication (two-way, reactive, and interactive) can be adopted for analyzing the degree of content interactivity of IRC messages. In Rafaeli and Sudweeks’ study (1997), a newsgroup message was coded as reactive if it responded to previous messages and as interactive if it contained references to the manner in which previous messages related to those preceding them. The coding of both reactive and interactive messages was done based on message references.

The problem of how to implement these levels in operational terms needs to be addressed. There is always some time difference between the referring message and the referred-to message, whether a message refers to the immediately previous message or to something posted months ago. While the issue of time difference may be irrelevant in newsgroup messages, it is a more important concern in IRC Communication. In newsgroups and other CMCs where previous messages are archived, one can easily retrieve and refer to what was said in the past, and the time difference between the referring message and the referred-to message may not mean anything. However, in IRC communication, messages are not archived, and words disappear permanently as soon as they scroll off the top of a computer screen. When a reference is made to what was posted weeks or months ago, the message reference shows not just how well one can remember the past,. but more importantly, it may indicate how long these individuals have been associating and the level of their interest in each other – so much that their past interaction made deep impression or so little (close to none) that nothing of the past registered.

The difference between a message referring to the immediately previous message and those that refer to messages posted even earlier but within a short period of time may be trivial. However, the implication of a message referring to what was posted a long time ago cannot be ignored, and the factor of time difference should be taken into consideration in analyzing message references in IRC communication. If nothing else, we at least need to differentiate messages referring to postings in the same sequence (session) from those that refer to postings in earlier sessions. We may call the former “within-session reference” and the latter “cross-session reference.” Message references within one particular session indicate within-session interactivity, and message references extending to earlier sessions indicate cross-session interactivity.

A related question is how to determine the beginning and ending of a session. Ideally a session should be defined as the beginning and ending of an encounter. While this can be done with messages involving only two participants, it is difficult and impractical when more than two participants are involved. Alternatively, a session can be defined as the episode in which a cluster of participants join and stay active on the channel. This is meaningful only if IRC participants get online on a regular schedule and their participation in the channel shows a stable pattern. Even when that is true, it will still be difficult to determine the beginning and end of a session. Individual participation is more likely to have many fractionated overlaps, which makes it almost impossible to identify clear dividing points in a continuous sequence of messages.

Fortunately, in the actual study of IRC communication, interactivity analysis comes down to line-by-line evaluation and coding, and we do not necessarily have to divide a continuum of message flow into separate pieces. While within-session references can be determined by examining previous lines of messages in a reasonably short time frame, cross-session references often can be identified by looking for explicit wordings or by catching clues in the content of a message itself.

Action-Simulating Messages

Action-simulating messages are lines of text posted by an IRC participant to describe what he or she is doing, or rather what his or her virtual being is doing or wishes to do, had it been given a physical body. It is possible that an individual might be actually doing what he or she claims to be doing in the posted message, while all at the same time busy with typing. However, it is more likely that the described action may only take place in the virtual space at the rhythm of finger tapping. There is no way to determine which is the case. The only thing we know is that the action-simulating message is posted to act out a play and that it ought to be taken into consideration in studying interactivity of IRC communication.

Just as verbal messages can be targeted at specific recipients or not directed to anyone, an action message can either be directed to an individual or be narcissistic in nature. They can be either unidirectional – directed to an individual but ignored, or dyadic – directed to an individual and responded to by the targeted person either with another action-simulating message or a verbal message. Directed action-simulating messages are almost always indicated by inclusion of the targeted individual's nickname, and the directness of such messages is self evident. Therefore, it should not be difficult to determine if an action-simulating message is targeted and to whom it is directed. By relating to the context, we can easily determine whether it is unidirectional or dyadic.

Finally, we have the problem of whether action-simulating messages should be analyzed separately from verbal messages or in correlation. The degree of interactivity in any session of IRC communication is not evidenced by verbal messages or action-simulating messages alone, but by both together. On the one hand, they are technically different, and they play different roles in group interaction. Participants often combine verbal and action-simulating messages in an interesting way to carry out playful drama without a script (Danet, Reudenberg-Wright & Rosenbaum-Tamari, 1996). On the other hand, a verbal message may be met with an action-simulating message or vice versa. If verbal and action messages are analyzed separately without relating to each other, a targeted message will be coded as unidirectional even though it was responded to by the targeted individual in a different mode. Therefore, the two kinds of messages should be treated as different categories conceptually but be examined in the context of each other in actual analysis.

Nickname Stability

Nickname stability, an issue not addressed in Jones’ theory, is apparently relevant to the problem of virtual community presence in IRC, as interpersonal relationships form the fabric of a community and interpersonal relationships cannot develop when individual identities are not discernible. (The term “identity” here refers to the physical existence of a participant, not his or her psychological identity in the sense of how this person views himself or herself and how he or she is perceived or wants to be perceived by others.) In IRC, all activities are carried out under nicknames, and nickname changes certainly affect how individuals relate to each other. The more nicknames each participant uses, the more difficult it is to trace individual identities. The more frequently participants change their nicknames, the more difficult it becomes to develop and maintain interpersonal connections, and therefore, the more unlikely it is that a community will emerge.

The problem of nickname instability (and thus the potential confusion of identities) arises directly from the fact that the rules governing nickname use are extremely liberal. The lack of restriction opens a number of possibilities that can be exploited by participants. They may use different nicknames each time participating in IRC, or choose to stick to the same nickname. They can change their nicknames whenever they want to and for whatever reason. They may purposefully change their nicknames to avoid unwanted attention or resume old nicknames to make themselves identifiable to those they met in the past. Finally, although any participant can use only one nickname and any nickname is associated with only one individual at any time, it is possible for several participants to log on with the same nickname at different times.

It is not clear whether IRC participants have been exploiting these possibilities, and to what extent if they have. Bechar-Israeli (1995) claims that IRC participants choose and maintain the same nicknames for themselves for a long period of time and that there is a relatively stable correlation between nicknames and participant identities, but little empirical evidence is given in support of such claim. After a thorough review of existing literature, we were unable to locate any additional research that points either one way or the other.

Given what is said above, we have to ask this question: how could an IRC participant possibly deal with the potential chaos and manage to keep track of other participants’ identities? If there exists an effective means for keeping track of individual identities, then even if nickname use in IRC is unstable, it would still be possible for the participants to develop interpersonal relationships. However, if such a means does not exist, then nickname stability will be a critical factor for virtual community formation and maintenance. In the following paragraphs, we examine all of the possible avenues, and we argue that none of them is sufficient for identity tracking.

At first look, IP string (user ID and IP address) may seem to be a solution. However, IRC servers announce participants’ IP strings only at the time of joining a channel. After that, every time a person changes his or her nickname, the server sends out an announcement saying that “X is now known as Y”, without specifying the person's IP string. The announcement is sent only to the currently-subscribed channels. Anyone who wants to keep track of this person's identity will have to monitor such announcements constantly. When many participants are coming and going and the computer screen is jammed with all kinds of messages, following the thread of nickname change quickly becomes difficult. Even if this approach works, it does so only when nickname changes happen within a session, but not when a participant signs up with a different nickname.

It is true that the system provides certain commands that can be used to verify the IP string associated with a nickname. However, effective use of such commands requires some expertise that average participants often do not have. Even if a participant is equipped with such skills and is willing to make the effort, he or she has to start with a nickname, and the best shot is to check out the nicknames that have been used in the past by the person whom he or she wants to meet again. When the number of nicknames associated with a person increases and many participants start enjoying the game of metamorphosis, this approach will become impossible.

The above discussion implies that IP string is a unique identifier of individual identity, which is not necessarily true. A participant may connect to a server under different IP addresses and login names, and an IP string may be associated with more than one participant. However, this imperfection only complicates the situation to a very small extent, and IP string remains a reasonably effective means of identity differentiation for research purposes. (This issue will be discussed in greater detail later in a separate section.)

Alternative approaches may exist that IRC participants can employ to maintain and to verify individual identities in the chaos of nickname changes. For instance, if a participant really wants another individual to know who he or she is, the participant can announce his or her real name and currently-used nickname via private exchange. The participant can even “prove” his or her identity either by revealing some personal details or by relating what was shared in the past. The participant may start the conversation by saying “hi, XYZ, I am the one who did this and that and who is a doll maker, remember?” or something like “tell me where I am if we have met and you really do know me.” However, for this tactic to work, at least one of the involved parties has to maintain relative nickname stability to make the initial contact possible. When both parties have changed their nicknames frequently or have used too many different nicknames, it will be too difficult to determine whom to approach in the first place.

Conceptual Hypotheses

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

The sections above carefully examined the four conditions for virtual community proposed by Jones, and proposed changes as well as additional factors that need to be considered in investigating the possibility of virtual communities in the environment of IRC. Each condition is given as a general conceptual hypothesis, below, and then restated in operational terms.

1. Stability of Membership:
    H1. The channel has a significantly large number of members whose participation is relatively stable throughout a reasonably long period of time:
        H1a. The channel has a significantly large number of frequent participants;
        H1b. Participants maintain a significant level of co-appearance in the channel.
    2. Interactivity:
        H2. Messages (including both verbal and action-simulating messages) posted in the channel demonstrate a significant level of interactivity:
        H2a. A significant level of cross-session reference is present in messages;
        H2b. A significant level of within-session reference is present in messages;
        H2c. A significant portion of messages are directional (explicitly targeted) in syntax and responded to by the targeted participants.
        H2d. A significant number of messages are fully interactive.
3. Stability of Identity:
      H3. Participants on the channel maintain stable identities as represented by nicknames:
        H3a. Within the scope of the channel, the number of nicknames associated with more than one participant is insignificant;
        H3b. Within the scope of the channel, the number of participants who ever used more than one nickname is insignificant.

Testing for nickname stability on the basis of one nickname per participant may be too restrictive. If empirical evidence warrants otherwise, the condition may be relaxed and the testing may be done on the basis of two or three nicknames per participant instead. Since transforming these conceptual hypotheses into statistical null hypotheses is trivial, we now move on to consider some major issues related to research implementation.

Differentiation of Participant Identity

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

Testings on co-appearance and nickname stability both require effective differentiation of participant identities. IRC servers identify participants by their associated IP strings such as “~theresa@208.164.194.136”. Although full names are also recorded in the database, IP string is the only element used for participant identification. This is because a user can fake his or her real name by configuring the client software, but not the user login ID and IP address. Therefore, the IP string is the only information available from the server that can be used for identity differentiation. (The sender's IP string is included as part of every message packet received from the server at the network level.) But are IP strings sufficient for differentiating participant identities throughout a period of time? More exactly, is an IP string always associated with the same participant? And does a participant always have the same IP string? Unfortunately, the answer to either of the last two questions is no.

An IP string gets associated with different identities when an Internet account is shared by more than one individual. This happens either when one connection is made from the account and several persons chat through the same program, or when several persons connect to the server using the same account at different times. Note that IRC chatting is often done for personal amusement during leisure hours and in solitude. While one may chat together with somebody else through the same connection once or twice, it is unlikely that he or she would keep doing this over time. Since our goal is to test for established patterns, occasional occurrences of such incidents will not invalidate the findings. As to the possibility of one account being shared by several individuals, we differentiate the case of sharing a host account issued to one person from the case of sharing a service account issued by an Internet Service Provider (ISP). Sharing an individual's host-based Internet account is against the policy of computing. Although policy violation has been reported occasionally in the past, the likelihood of this happening during the period of observation is small. In the case of a family sharing one ISP service account, a different login name is often created for each family member. When a family member accesses the Internet using his or her own login name, the IP string associated with his or her appearance will be different from that of other family members.

The possibility of one participant being associated with multiple IP strings arises when a participant uses two or more Internet accounts for online chatting. The person may have access to several computers each of which has a unique IP address, or the person may have access to several accounts on one single networked machine that has a unique IP address. Since the second scenario is almost impossible, we concentrate on the first one.

The problem of a participant's being associated with multiple IP strings is most likely to occur in two possible situations. One is when a participant changes his or her Internet Service Provider. The possibility of someone keeping more than one ISP account is unlikely and will not be considered here. The other is when a participant maintains an ISP account and has access to one or several networked computers. In view of the current infrastructure of the Internet, it is reasonable to assume that most likely the ISP account is used at home and the networked computers are used at workplace. The possibility of using networked computers in the workplace for online chatting can be discounted, given the facts that online chatting is often done in private settings and that anyone who is caught doing it during work hours endangers his or her employment.

This leaves only the possibility of ISP switching for further discussion. On one hand, the business of providing Internet access has moved out of the initial stage of chaos and both service features and price structure have reached a point of stability and uniformity. There is little to gain by frequently changing ISP, either in terms of savings or in improvement of service quality. On the other hand, a user may still change his or her ISP for various reasons. Nevertheless, changes of ISP are not likely to be frequent and widespread, for the simple reason that people are often discouraged from doing it by the trouble of software installation and computer reconfiguration.

From the above analysis, we conclude that although IP string may not always be a perfect identifier of participant identity, it will be sufficient and serve reasonably well as a means of identity differentiation most of the time.

Measurements & Parameter Setting

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

Measurements of most constructs are straightforward. For a given sample of IRC messages, the level of cross-session/within-session references can be measured by the ratio of referencing messages to the total of messages in the sample. Nickname associations can be extracted for each nickname (or participant) on a nickname-by-nickname (or participant-by-participant) basis and then compiled into overall frequencies. The only problem that needs to be addressed at greater length is how to measure the level of participants’ co-appearance in a channel.

Measurement of Co-appearance Level

First, co-appearances can be categorized by the number of participants co-appearing – co-appearances of two-participants, three participants, et cetera. Second, for any number of participants, their co-appearances can be characterized either by the total number of times these participants co-appear for a minimum duration, or by the accumulated amount of time throughout the period of observation. While the former emphasizes overall patterns of co-appearance, the latter indicates the intensity of co-appearance. To combine the strengths of both, average lengths of co-appearance can be computed.

An ideal measurement of co-appearance should take into consideration the number of co-appearing participants as well as the intensity/frequency factor. One possible solution is to identify all of the participants who have been active during the period of observation on a specific channel, and then for each possible grouping of participants, compute the average length of co-appearance. For instance, we can start by computing the average length of co-appearance by any two participants, and then by any three, any four, and so on. For each level of combination, a mean ratio of co-appearance length to total connection time can be computed for each participant to facilitate statistical analysis. It is reasonable to assume that the larger the size of the groups that co-appear in an IRC channel, the more likely it is that a virtual community is emerging or already in existence.

Establishment of Significance Thresholds

In the conceptual hypotheses, phrases such as “significant levels” are used to describe a desired degree of certain properties, but the meaning is not specified in measurable terms. In other words, we have the problem of how to define the critical point of significance. Ideally, the threshold should be set according to empirical estimation. Random samples of messages may be gathered from channels where virtual communities are known to exist, and relevant ratios can be calculated from such samples and then used for testing. While this approach sounds convincing, it assumes that we already have a way of testing for the existence of virtual communities, and therefore suffers from the logical problem of circular reasoning.

One way to avoid the logical fallacy is to compute threshold parameters from samples of online messages produced by small groups that have already established stable interpersonal relationships. It is known that people from the same neighborhood or students from the same school do come online to chat. Although they often chat in private channels, arrangement can be made to observe their online activities. Data gathered from such small groups can be used to estimate the critical levels of nickname stability and message references.

However, this technique cannot be used to estimate the critical level of co-appearance. This is because relationships within the group were developed in face-to-face interaction and later transplanted into the virtual place of an IRC channel. The level of co-appearance required for forming initial bonds may be fundamentally different from what is necessary for maintaining existing relationships. When a relationship goes through the stages of initial contact, intimacy, and possible dissolution (DeVito, 1995), the emotional state of involved individuals will work in combination with social norms to dictate their behaviors toward each other. Their desire for and avoidance of contacts will be reflected in their online participation and cause changes in the level of co-appearance. The critical level of co-appearance computed from such a sample would be inconsistent with the fact that a virtual community encompasses not just established and successful relationships, but relationships at all different stages, including those that have already died or are yet to form.

Alternatively, the critical level of co-appearance can be determined by means of mathematical reasoning. Note that as we mentioned earlier, for any given group size of co-appearance, the co-appearance level of every single participant can be measured in terms of the mean ratio of co-appearance length to total connection time. The problem here is to find the critical value of mean ratio that separates stable co-appearances from co-appearances that have no established patterns. The critical value can be mathematically derived from the assumption that co-appearances of participants in the research population are absolutely random. The details of calculation are given in the Appendix.

The assumption of randomness implies that participation in a channel shows absolutely no pattern of co-appearance, which is clearly unrealistic. Activities in a channel may show a level of co-appearance above the threshold but demonstrate no trace of a virtual community either forming or in existence. Nevertheless, the mathematically derived threshold sets the minimum requirement. It constitutes a necessary but not sufficient condition.

Research Design and Related Issues

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

Obviously, the data (IRC scripts) for condition testing have to be collected over a sufficiently long period of time – long enough to reveal interesting patterns and to even out random fluctuations in participation by channel members. Although there are no hard rules on how long the observation period has to be, some field experience would be sufficient to make such a decision. One day, or even a week, is apparently too short, because many IRC participants may only get online once a week. Our preliminary observation suggests that three to six months would be long enough to generate sufficient data to make a study meaningful.

Selection/Sampling of Channels

Online chatting is done on ISP-provided programs as well as on host systems of the Internet. The systems (IRC servers) typically operate in groups. Servers in the same group are connected to form an network. Participants connected to the same server or to different servers on the same network are able to communicate with each other, but not with those on a different network. Some networks are much more popular than others. A popular IRC network may have hundreds of channels, of which some are heavily populated while others are not. Note that, as we explained earlier, IRC channels may disappear or be re-created, but they do not change names or transmogrify.

Selective or random sampling of IRC channels can be done in two possible ways. One is to combine channels from existing IRC networks into one general pool for sampling. The other way is to first select IRC networks and then select channels on the chosen networks.

Message Scripting

Two possible methods may be used to observe activities on IRC channels. One is to intercept and script participants’ postings from the server side. The other is to do it from the client side by being present at chosen channels. We call the former the “server-side approach” and the latter the “client-side approach”. Regardless of which approach is taken, if interception is done at the network level, every line of unparsed message will have the following structure: “:trueGal!~theresa@208.164.194.136 PRIVMSG #california :hey guys!”, in the format of nickname, IP string, action keyword, channel destination, and posted message. Each line of message can be time stamped.

The client-side approach can be implemented with a small program installed on a personal computer. The program can be configured to stay on several channels at the same time and log every line of messages coming from the IRC server into channel-specific files. However, this approach suffers from some major problems.

First, the observation cannot be truly unobtrusive. Even though the observation is absolutely passive and the “observer” does not contribute anything to the public forum, one can still argue that the very presence of the observer changes the group dynamics. This is true especially when the “observer” is singled out and approached by other participants. The researcher will have to face the dilemma of either responding to the inviting participant or refusing to be engaged in any conversation. Either way, the researcher's decision will affect the approaching individual.

Second, the scripting process may be interrupted either by technical difficulties or by human interference. The server may cut off the connection after a certain time of inactivity or when it detects a collision of nickname. The monitoring program may get kicked out, or even banned, from a channel when someone with power becomes tired of a constant “presence” which never says anything. In any case, the researcher should be aware of these problems and be prepared to deal with them.

Third, invited-only (private) channels also present a problem. The monitoring computer program probably will never get invited into a private channel. The researcher may get online in person, try to gain the trust of a channel owner/operator, and manage to get invited. Then, the researcher can either script the channel manually or try to smuggle the computer program into the channel under the same IP string. Although this method may not necessarily work with all invited-only channels, it at least provides a chance to get some idea of what is going on there.

Finally, monitoring private exchanges from the client side remains the biggest problem. It was noted on several occasions earlier that private exchanges among participants in a channel are important evidences of interpersonal interaction and that they should be included in the research. However, if the client-side approach is taken, it is impossible to intercept and script private messages, because private messages are directly sent from the server to the recipient. A possible remedy is to recruit voluntary participants to script their private exchanges for the research. While scripts of private exchanges gathered this way may reveal what is going on underneath the surface to some extent, they cannot provide a complete picture of private interaction in IRC.

Compared to the client-side approach, the server-side approach seems to offer much more control and flexibility. With a piece of software installed on the IRC server, all of the messages passing through the server – including private exchanges and messages meant for “invited-only” channels – can be captured. The process is seamless and the involved parties will know nothing about it. Although the server-side approach requires a great deal of assistance and co-operation from the system administrator and some system administrators may not be willing to help at all, the promise of highly complete and unobtrusive research data makes this approach extremely attractive.

Nevertheless, the decision of which approach should be used to collect research data can not be made solely on the basis of data quality. In the following section, we examine the ethical aspect of this research and focus our discussion on ethical issues in collecting unobtrusive and naturalistic data on human interactions in IRC.

Ethical Issues

While researchers in the field of computer-mediated communication are deeply concerned about the ethical issues of conducting research on human interactions in cyberspace, they have difficulty reaching agreement on common ethical guidelines (Allen, 1996; Bohlefeld, 1996; Garton, Haythornthwaite, & Wellman, 1997; Herring, 1996; King, 1996; Reid 1996; Paccagnella, 1997; Thomas, 1996a, 1996b; Waskul, 1996). Even if common ethical guidelines exist, following the guidelines to the letter does not automatically result in responsible and ethical research. It is not a matter of following codes, policy or procedure, but a matter of commitment to protect the participants in one's study from potential harm (Waskul, 1996). The individual researcher has the ultimate responsibility for keeping the best interests of the research participants in mind.

Recognizing that ethical issues may arise anywhere throughout the course of a research, we can nevertheless divide the ethical concerns of this research into three parts: ethical issues in data collecting, in data handling, and in reporting of findings respectively. It seems that ethical concerns in data handling and reporting of findings can be dealt with by following common practices in the field of CMC research. We can always stay on the cautious side by guarding the confidentiality of data unconditionally, by avoiding direct quotation of postings, and by not referring to any particular IRC participants in our writing.

In contrast, the data collecting part of this research is much more problematic from the ethical perspective. In this research, we need to gather data on mainly three kinds of CMC communications in IRC: public postings in public IRC channels, public postings in private IRC channels, and finally, private (one-to-one) exchanges in IRC. The three kinds of data present different sets of ethical problems, and we examine each of them separately below.

The ProjectH Research Group's ethics policy (Paccagnella, 1997; Rafaeli, Sudweeks, Konstan and Mabry, 1994) seems to take a reasonable position on the issue of how to study human communications in public domain ethically. Although the ProjectH ethics policy was specifically designed for studying newsgroup postings, it can be adopted for this research and be used to guide our practice in collecting and managing postings in public IRC channels.

Conversations in publicly accessible IRC channels, like messages posted on newsgroups, are public acts deliberately intended for public consumption. Recording, analyzing, and reporting of such content, where individuals’ identities are shielded, is not subject to “Human Subject” constraints. It should not be necessary to inform the participants or to obtain explicit consent before collecting their public IRC messages for research purposes. Nor is it necessary for the researcher to take any more precautions in handling the collected data than in any study of real-life public activities.

The problem of how to observe participants’ interactions in a private (invited-only) channel seems to be more complicated. The degree of “perceived privacy” (Herring, 1996; King, 1996) in a private IRC channel is relative to the number of participants. When there are only two or three participants chatting in the channel, the situation becomes essentially equivalent to private conversations. When the channel has a large number of participants, the degree of perceived privacy reaches the minimum. Since we are only interested in channels with a large number of participants, it should not be necessary to treat public postings in private channels much differently from those in public channels. However, message content and participant membership should be kept confidential, and the research report should not include any information that may be potentially used to identify either individual participants or the private IRC channel.

One-to-one messaging in IRC, regardless of whether it is carried out in a public channel or in a private channel or outside of any channel, is private communication. General research ethics requires that one should have informed consent from the research participants before recording their private communications, or gathering personal information from them. While it is relatively easy to obtain informed consent in studies that rely only on a few voluntary informants, it is rather difficult in this research. The goal here is to investigate the overall interactivity level in an IRC channel sustained through a period of time, and we can not possibly accomplish this by relying on only a few voluntary informants. On the other hand, it is impossible to anticipate who is going to join or leave an IRC channel and who is going to “message” whom at any particular moment. One may suggest that we should ask every participant for informed consent as soon as he or she joins the channel, which is hardly practical, or periodically post an announcement to inform the participants about the recording. However, as noted by King (1996), the act of requesting for informed consent, or even the compromised attempt of informing participants about the recording, will be grossly disruptive to the group process and make the results of this research questionable.

It seems that one possible solution to this ethical dilemma is to record only who is sending one-to-one messages to whom, but not the actual content, without explicitly seeking for informed consent from either the sender or the recipient. By not recording the actual content of private messages, we manage to protect the essence of private communication. Although the identities of individual participants and their acts of private messaging are recorded in the scripts, the potential of putting any participant at risk of harm as a consequence of this research is minimized. Once the scripts are compiled into quantitative data without human intervention, all of the related computer files should be destroyed and no permanent records should be kept. The generated data on occurrences and patterns of one-to-one communication in IRC will be sufficient for the purpose of this research, i.e., to statistically test for virtual community existence in IRC.

However, one may argue that with the approach described above, even though the content of private messages is not recorded in scripting, the process of monitoring private communications without informed consent itself constitutes a violation of the participants’ perceived privacy. Our response to this argument is that the importance of complete and unobtrusive observation in this research outweighs the necessity of acquiring “informed consent”. Determination of whether a research approach is ethical should not be based on whether informed consent is sought from research participants, but rather on whether the researcher makes his or her best effort to protect the research participants from any harm, and on whether the participants may suffer from any potential damage as a result of the research. On such matters, however, there will continue to be debate and discussion.

Data Extraction and Message Coding

Once enough scripts are collected, a computer program can be employed to extract the necessary data and to compute mean co-appearance ratios for each participant based on different factors of combination. Although the bulk of data to be processed will be huge, a well-designed program running on a reasonably powerful machine should be able to handle it.

While IP string-nickname associations shall be extracted from the whole collection of scripts and mean co-appearance ratios may be computed on a channel-by-channel basis, analysis of interaction (i.e., message references and message targeting) does not have to include every line of script. Since such analysis has to be done manually, it would be terribly expensive and impractical to try to code every single piece of script. Instead, the analysis may be limited to a good number of segments randomly selected from the whole collection.

For every channel chosen for inclusion in the study, the continuum of scripts for the whole period of observation may have weak spots where the channel does not have much activity going on. The weak spots form natural divisions of the continuum and can be used to define segments. Once the continuum of scripts is divided into segments of manageable size, a good number of segments can be randomly selected for the channel. The resulting random samples specific to IRC channels can be either analyzed separately or combined into a larger pool of samples, depending on what specific research questions we ask.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

The purpose of this article was to explore the possibility of empirically testing for the presence of virtual community in IRC. The focus of the article was neither on how to analyze characteristics of IRC-based virtual communities statistically nor on how to distinguish virtual communities in IRC from those in other technical settings, but rather, on how to test empirically for the presence of virtual community in a sinlge IRC channel.

In this article, Jones’“virtual settlement” theory (1997) was adopted as a general framework. Jones' four conditions of virtual community were related to the technical context of IRC and extended to account for the particular characteristics of IRC-based group communication. The results of this discussion were stated as conceptual hypotheses as a way of recapitulation. Issues of construct operationalization, measurement, research design, and data manipulation were addressed.

Several important factors peculiar to virtual community in IRC were identified. It was argued that the condition of membership stability can be accurately tested in terms of participant co-appearance and that nickname stability should be included as an additional condition of virtual community existence in IRC. Interactivity analysis of IRC communication needs to include both verbal and action-simulating messages and to be done on message content as well as on message directing. The time difference between a referring message and the referred-to message should be incorporated into analysis and coding of message references.

This article demonstrates that it is possible to test empirically for the existence of a virtual community instead of simply assuming its existence in a technical domain as done in the past. The article contributes to the field of virtual community research by clarifying and operationalizing concepts developed by other researchers from theoretical analysis and field observation. It paves the way for more systematic investigation of virtual communities and for building stronger empirical/quantitative research in the field of virtual community research. Although the emphasis is on testing for the existence of a virtual community in one particular IRC channel, the ideas presented in this article can be useful for implementing similar studies with different orientations.

The author is currently implementing software instruments to for the systematic investigatation of virtual community based on the ideas and methodology described in this article. Findings of the project will be reported soon.

Footnotes
  • 1

     Note that here the word “channel” refers to a virtual place created on an IRC server, not the social gathering of a group of individuals in the place. When a group of people who regularly chat on #channel A decide to abandon this channel altogether and to move to a new place #channel B, A and B are considered to be technically two separate places. Instead of saying that they have changed the name of their virtual gathering place, we believe that from the technical point of view it is better to say that they have moved their meeting from place A to place B.

  • 2

     Note that we choose to compute the mean ratio only over the non-zero values, by dividing the sum with m instead of n-1. It is easy to demonstrate mathematically that if it is calculated on all ratios, including the zeros, the mean ratio will be always 1/(n-1).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix
  • Allen, C. (1996). What's wrong with the “Golden Rule”? conundrums of conducting ethical research in cyberspace. The Information Society, 12(2), 175187.
  • Alstyne, M., & Brynjolfsson, E. (1997). Electronic communities: Global village or cyberbalkans?. MIT Sloan School.
  • Barker, D. B. (1991). The behavioral analysis of interpersonal intimacy in group development. Small Group Research, 22(1), 7691.
  • Baym, N. K. (1995). The emergence of community in computer-mediated communication. In S. G.Jones (Ed.), Cybersociety: Computer-mediated communication and community (pp. 138163). Thousand Oaks , CA : Sage.
  • Bechar-Israeli, H. (1995). From <Bonehead> to <cLoNehEAd>: nicknames, play, and identity on Internet Relay Chat. Journal of Computer-Mediated Communication[Online], 1(2). Available: http://www.ascusc.org/jcmc/vol1/issue2/bechar.html.
  • Berger, C. R., & Kellermann, K. (1983). To ask or not to ask: Is that a question? In R.Bostrom (Ed.), Communication Yearbook 7 (pp. 342368). Newbury Park , CA : Sage.
  • Boehlefeld, S. P. (1996). Doing the right thing: ethical cyberspace research. The Information Society, 12(2), 141152.
  • Danet, B. (1995). Playful expressivity and artfulness in computer-mediated communication. Journal of Computer-Mediated Communication[Online], 1(2). Available: http://www.ascusc.org/jcmc/vol1/issue2/genintro.html.
  • Danet, B., Reudenberg, L., & Rosenbaum-Tamari, Y. (1996). “Hmmm…Where's that smoke coming from?”: Writing, play and performance on Internet Relay Chat. Journal of Computer-Mediated Communication[Online], 2(4). Available: http://www.ascusc.org/jcmc/vol2/issue4/danet.html.
  • DeVito, J. A. (1995). The interpersonal communication book. Harper Collins College Publishers, 7th ed.
  • Douglas, W. (1987). Question-asking in same- and opposite-sex initial interactions: The effects of anticipated future interaction. Human Communication Research, 14, 230245.
  • Douglas, W. (1990). Uncertainty, information-seeking, and liking during initial interaction. Western Journal of Speech Communication, 54, 6681.
  • Douglas, W. (1994). The acquaintanceship process: An examination of uncertainty, information seeking, and social attraction during initial conversation. Communication Research, 21(2), 154176.
  • Erickson, R. (1997). Social interaction on the net: Virtual community as participatory genre. In J. F.Nunamaker & R. H.Sprague (Eds.), Proceedings of the Thirtieth Hawaii International Conference on System Science, 6 (pp. 2330). Los Alamitos , CA : IEEE Computer Society Press.
  • Fernback, J., & Thompson, B. (1995, May). Virtual communities: Abort, retry, failure? Presented as “Computer mediated communication and the American collectivity” at the Annual Meeting of the International Communication Association, Albuquerque, New Mexico, May, 1995. Available: http://www.well.com/user/hlr/texts/VCcivil.html.
  • Garton, L., Haythornthwaite, C., & Wellman, B. (1997). Studying online social networks. Journal of Computer-Mediated Communication[Online], 3(1). Available: http://www.ascusc.org/jcmc/vol3/issue1/garton.html.
  • Hagel, J., & Armstrong, A. (1997). Net gain: Expanding markets through virtual communities. Boston , MA : Harvard Business School Press.
  • Herring, S. (1996). Linguistic and critical analysis of computer-mediated communication: Some ethical and scholarly considerations. The Information Society, 12(2), 153168.
  • Hof, R. D., Browder, S., & Elstrom, P. (1997). Internet communities. Business Week European Edition, May 5, 1997, 3847.
  • Johnston, E. (1997). The community in cyberspace[Online]. Available: http://www.acs.ucalgary.ca/~dabrent/ 380/webproj/commun.html.
  • Jones, Q. (1997). Virtual-communities, virtual settlements & cyber-archaeology: A theoretical outline. Journal of Computer-Mediated Communication[Online], 3(3). Available: http://www.ascusc.org/jcmc/vol3/issue3/jones.html.
  • Jones, S. G. (1995). Understanding community in the information age. In S. G.Jones (Ed.), Cybersociety: Computer-mediated communication and community (pp.1035). Thousand Oaks , CA : Sage.
  • King, S. A. (1996). Researching Internet communities: Proposed ethical guidelines for the reporting of results. The Information Society, 12(2), 119127.
  • Kollock, P., & Smith, M. (1994). Managing the virtual commons: Cooperation and conflict in computer communities. In S.Herring (Ed.), Computer-mediated communication. Amsterdam : John Benjamins.
  • Martin, M. M., & Rubin, R. B. (1998). Affinity-Seeking in Initial Interactions. The Southern Communication Journal. 63(2):131143.
  • Newhagen, J. E., & Rafaeli, S. (1996). Why communication researchers should study the Internet: A dialogue. Journal of Communication, 46(1):413. Also see Newhagen, J. E., & Rafaeli, S. (1996). Why communication researchers should study the Internet: A dialogue. Journal of Computer-Mediated Communication [On-line], 1 (4). Available: http://www.ascusc.org/jcmc/vol1/issue4/rafaeli.html.
  • Paccagnella, L. (1997). Getting the seats of your pants dirty: Strategies for ethnographic research on virtual communities. Journal of Computer-Mediated Communication[Online], 3(1). Available: http://www.ascusc.org/jcmc/vol3/issue1/paccagnella.html.
  • Parks, M., & Floyd, K. (1996). Making friends in cyberspace. Journal of Communication, 46, 8097.
  • Rafaeli, S. (1986). The electronic bulletin board: A computer driven mass medium. Computers and the Social Sciences, 2(3), 123136.
  • Rafaeli, S. (1988). Interactivity: From new media to communication. In Sage Annual Review of Communication Research: Advancing in Communication Science (pp. 110134), 16. Berverly Hills : Sage.
  • Rafaeli, S., & LaRose, R. J. (1993). Electronic bulletin boards and “public goods” explanations of collaborative mass media. Communication Research, 20(2), 277297.
  • Rafaeli, S., & Sudweeks, F. (1997). Networked interactivity. Journal of Computer-Mediated Communication[Online], 2(4). Available: http://www.ascusc.org/jcmc/vol2/issue4/rafaeli.sudweeks.html.
  • Rafaeli, S., Sudweeks, F., Konstan, J., & Mabry, E. (1994). ProjectH overview: A quantitative study of computer-mediated communication [Online]. Technical Report, Key Centre of Design Computing, University of Sydney. URL: ftp://ftp.arch.su.edu.au/pub/projectH/papers/techreport.txt.
  • Redmond, M. V. & Vrchota, D. A. (1997). The effects of varying lengths of initial interaction on attraction. Communication Reports, 10(1), 4753.
  • Reid, E. M. (1991). Electropolis: Communications and community on Internet Relay Chat[Online]. Honors Thesis, Department of History, University of Melbourne. Available: http://people.we.mediaone.net/elizrs/electropolis.html.
  • Reid, E. (1996). Informed consent in the study of online communities: A reflection on the effects of computer-mediated social research. The Information Society, 12(2), 169174.
  • Rheingold, H. (1993). The virtual community: Homesteading on the electronic frontier. Reading , MA : Addison-Wesley.
  • Rheingold, H. (1994). A slice of life in my virtual community. In L. M.Harasim (Ed.), Global networks: Computers and international communication. Cambridge , MA : MIT Press.
  • Smith, M. A. (1992). Voices from the WELL: The logic of the virtual commons. Unpublished master's thesis, University of California, Los Angeles. Availble http://www.cpsr.org/cpsr/sociology/virtcomm.txt.
  • Sudweeks, F., & Rafaeli, S. (1994). How do you get a hundred strangers to agree: Computer mediated communication and collaboration. In T.Harrison & T. D.Stephen (Eds.), Computer Networking and Scholarship in the 21st Century University. SUNY Press.
  • Thomas, J. (1996a). Introduction: A debate about the ethics of fair practices for collecting social science data in cyberspace. The Information Society, 12(2), 107117.
  • Thomas, J. (1996b). When cyber-research goes awry: The ethics of the Rimm “Cyberporn” Study. The Information Society, 12(2), 189198.
  • Turkle, S. (1995). Life on the screen: Identity in the age of the Internet. New York : Simon & Schuster.
  • Turkle, S. (1996). Virtuality and its discontents: Searching for community in cyberspace. The American Prospect, 24, 5057.
  • Weinreich, F. (1997). Establishing a point of view towards virtual communities. Computer-Mediated Communication[Online], 3(2). Available: http://www.december.com/cmc/mag/1997/feb/wein.html.
  • Waskul, D. & Douglass, M. (1996). Considering the electronic participant: Some polemical observations on the ethics of on-line research. The information society, The Information Society, 12(2), 129139.

Appendix

  1. Top of page
  2. Abstract
  3. Introduction
  4. The “Virtual Settlement” Theory
  5. Conditions of Virtual Community Existence in IRC
  6. Conceptual Hypotheses
  7. Differentiation of Participant Identity
  8. Measurements & Parameter Setting
  9. Research Design and Related Issues
  10. Conclusion
  11. References
  12. Acknowledgments
  13. Appendix

Appendix: Measurement and Significance Level of Co-appearance

We start with the case of co-appearances by any two participants. Assume that we have a total of n participants in the research sample. Then for any given participant X, there will be n-1 possible pairs of combination that include participant X as a member. An average length of co-appearance can be computed for each pair.

Consider a combination of any two participants X and Y. Assume that participant X has been active in the channel during the period of study for an accumulated amount of time t, and that the average length of co-appearance with participant Y is r. Then, a ratio of r/t can be calculated for participant X in reference to this particular pair. The ratio r/t is 0 if X and Y have no co-appearance with each other and 1 if participant X joins the channel and stays only as long as participant Y is there.

A ratio of r/t can be computed for the given participant X in reference to each of the n-1 possible pairs of combination. Therefore, for any given participant, we have n-1 ratios of r/t, of which some will be zero and others will be within the interval of (0, 1]. Suppose that out of these n-1 ratios we have m non-zero figures. By adding up all of these non-zero ratios2 and dividing the total by m, we have a mean ratio m which is the measure of overall dyadic co-appearance level of participant X. Needless to say that such a mean ratio of dyadic co-appearance can be computed for every single participant in the research sample.

We now proceed to estimate the critical value of dyadic co-appearance level that is required for significance testing. Since the statistical test is a mean-comparison analysis with the hypothesis being that the population mean of the data is greater than a pre-defined critical value (μ > μo), our goal here is to estimate the critical value μo.

Assume that occurrences of co-appearance are purely random. That is, every participant is equally likely to co-appear with anyone else in the research population; thus the average length of co-appearance by one participant with any other participant is the same. This is possible only if the participant (1) always co-appears with every other participant at once, or (2) co-appears with different participants at different time, and the connection time is equally distributed among the n-1 possible pairs of combination. Since the first condition is rarely true in reality, we only consider the second condition from now on.

Again, let t be the total amount of time that a given participant has appeared on the channel during the period of study. Then the average length of this participant's co-appearance with anyone of the n-1 other participants will be t/(n-1), and the ratio of his or her average length of dyadic co-appearance to the total connection time, in reference to any of the n-1 possible pairs of combination, will be 1/(n-1). The mean ratio of this participant's dyadic co-appearance, over all of the possible pairs of combination, is also 1/(n-1). Therefore, the critical value – expected mean ratio of dyadic-co-appearance for any participant in the research population – is 1/(n-1).

Now we consider co-appearances by more than two participants. The problem can be solved using the same approach. Let f be the factor of combination – the number of participants whose co-appearances as a group are under consideration, and n be the total number of participants identified on a channel through the period of observation. Then, for any given participant, the total number of possible combinations that include this participant as a member would be C(n-1, f-1) = (n-1)!/ [(f-1)! (n-f)!]. Again, assuming that the participant's co-appearance with any of these groups is completely random, the mean ratio of his co-appearance at the level of combination factor f is 1/C(n-1, f-1) = [(f-1)! (n-f)!] / (n-1)!, which is also the critical value μo, or the expected value of mean ratio of co-appearance at the level of combination factor f.