Holly's Dissertation



Chapter 3
COMMUNICATION

Although we have come to expect that new technologies are revolutionary, I suspect that at the level of individual innovators and managers the process of cultural construction is often one of "cultural creep." By this I would suggest that inventors and producers often create artifact to fit into cultural spaces suggested by their existing frames of meaning.
W. Bernard Carlson [24]

Whereas the Internet had its beginnings as ARPANET in 1969 to further research and development of technologies for military use [40], Usenet grew from a program by several graduate students in 1979 as a forum for discussion of Unix and Unix troubleshooting [114]. Usenet is now available through most Internet service providers [40]. This chapter situates Usenet among other Internet services and provides a background on Usenet interactions. After a brief history of the Internet and Usenet, I describe the services available on the Internet and compare the characteristics of the interpersonal communication services that include Usenet. The last section is a detailed description of the organization and distribution of the messages in Usenet. From this description I elaborate on Usenet interaction and the social dilemmas of that interaction.

HISTORY

As part of a joint effort in 1969 by the Advanced Research Projects Agency (ARPA), universities and research communities, ARPANET was created so that ARPA-funded researchers could exchange computer data and operate different computers at a distance [114]. In 1983 ARPANET split into two separate networks, ARPANET and MILNET. A computer on ARPANET could exchange information with a computer on MILNET by routing the data through a gateway computer, thus forming a network of networks, the Internet [100]. Other networks were formed about this same time--BITNET for scholarly and academic discussion in 1979, CSNET for computer science in 1981, and in 1986, NSFNET for connecting supercomputer centers [114]. The TCP/IP communication protocols were developed by DARPA (successor to ARPA) to route information between these networks and ARPANET through an inter-network gateway [50]. In March 1990, ARPANET was disbanded and NSFNET became the Internet's main backbone [114]. Currently, the Internet is composed of high-speed backbone networks such as NSFNET and MILNET, mid-level networks such as universities and corporations and stub networks such as individual local area networks [40].

Key to the success of ARPANET and subsequent networks were two fundamental elements--a new way of looking at computers and a decentralized communication system. Computers were viewed as communications devices rather than just computational ones; email became the most popular use of the network [52]. In addition, the purpose of the decentralized communication system was to provide the network the capability to survive a nuclear attack [50]. The impact of this approach is expressed by Rheingold:

Information can take so many alternative routes when one of the nodes of the network is removed that the Net is almost immortally flexible. It is this flexibility that CMC telecom pioneer John Gilmore referred to when he said, "The Net interprets censorship as damage and routes around it.". . . This invention of distributed conversation that flows around obstacles--a grassroots adaptation of a technology originally designed as a doomsday weapon--might turn out to be as important in the long run as the hardware and software inventions that made it possible. ([114], p. 7)

The Unix Users Network (Usenet) grew out of efforts by several graduate students instead of the government sponsored research efforts that gave rise to the Internet. In 1969 Unix was created by Bell Laboratories [128]. In 1977, the Unix-to-Unix Copy (UUCP) utility, a store-and-forward protocol also from Bell Laboratories, was being shipped with the new version 7 of Unix [50]. UUCP allowed any computer running Unix to share information with any other Unix computer via a modem connection [128]. UUCP provided an alternative means of networking for those computers not on the Internet. In 1979, Tom Truscott and James Ellis of Duke University, together with Steve Bellovin of the University of North Carolina, developed a program using UUCP that regularly exchanged information via modem [114]. Although this first version of Usenet news was a forum for discussion of Unix without needing an ARPANET connection (and was advertised as 'a poor man's ARPANET' at the Delaware Summer 1980 Usenix meeting [52]), Usenet grew into thousands of discussions on any topic and began to be carried on ARPANET. This lead to a gradual replacement of UUCP by NNTP (Net News Transfer Protocol), a method of transmitting Usenet on TCP/IP connections [50]. By 1992, 60 percent of all Usenet traffic was carried via the Internet using NNTP, the remainder still using UUCP connections [114].

Other communication media and information services in addition to Usenet are available via the Internet. These include information services such as file transfer (FTP), remote login (telnet), location and retrieval (gopher) and World-Wide Web (WWW), as well as communication services such as electronic mail (email), mailing lists (listserv), internet relay chat (IRC) and multi-user domains (MUD). Communication services are included in the study of computer-mediated communication, yet each has different characteristics. In the next section I describe each of these interactions and compare their characteristics.

INTERNET SERVICES

The Internet information services described above provide access to information and to other computer systems. Both FTP and telnet enable connections to remote computers. FTP provides access to file archives [83]. It allows file transfer between computers and was designed to manage the differences in rules and conventions (protocols) used on different computer systems [40]. On the other hand, telnet allows you to run a program resident on a remote computer without having to copy it to the local system [100]. Gopher is a menu-driven program that allows you to locate and retrieve information, to transfer files using FTP, and to telnet to other computers [40]. Although gopher displays only ASCII files, other formats may be transferred [83]. WWW is a more powerful system than gopher and uses hypertext links instead of a hierarchical set of menus to navigate among resources [83]. With the use of a WWW browser, you have access to documents written in HTML (hypertext markup language) as well as to the other Internet services. Because these services provide only indirect interaction among people, they are not included in the following comparisons.

The Internet communication services provide interpersonal communication opportunities. Both email and listserv offer services similar to a postal system. Email is the electronic exchange of messages among people and listserv is the exchange of messages among many people using an automated mailing-list distribution service [40]. Both services deliver messages to an individual's email address. All listservs require that you subscribe to the discussion list. IRC is an expansion of the Unix talk program and is similar to citizens-band radio in that messages are displayed in real time to people in the same group or channel [40]. A MUD is an interactive role-playing game using text in an interactive talk session format [40] (although building the MUD environment also requires programming skills [22]). Lastly, newsgroups are open discussion forums with messages distributed via Usenet software. A moderated group has a volunteer who screens messages for appropriateness before forwarding them to the list [40]; both listservs and newsgroups may be moderated.

INTERNET INTERPERSONAL COMMUNICATION

There are various ways of looking at the interpersonal interactions that take place on the Internet. One way is to compare the interaction with traditional levels of communication. These levels often are defined in terms of the four aspects of communication--a source or sender, a message, a channel or medium, and an audience or receiver [1]. Although this approach minimizes the role of media and channel [95], it is an instructive beginning for comparison. Littlejohn categorizes communication in four levels--interpersonal, group, organizational and mass communication [82]. Interpersonal communication is private, face-to-face (FtF) interaction, group communication is small group interaction usually in decision-making settings, while organizational communication occurs in large cooperative networks with discernible structure and function. Mass communication or mass media uses a complex institutional organization to send a message from a source to a large audience; the channel makes the message accessible to anyone [6]. Littlejohn stresses that these are not mutually exclusive types of communication but merely convenient mechanisms for organizing communication theory [82]. Another approach by Rogers has only three levels--FtF interpersonal communication, interactive machine-assisted interpersonal communication and mass media [117]. FtF interpersonal communication is one-to-few with the potential for equal control of the communication. The interactive machine-assisted interpersonal communication is characterized as many-to-many (N-N) with the potential for equal control of the communication. Lastly, mass media is one-to-many (1-N) with little control of the message by the receivers.

The interpersonal communication services of the Internet can be classified using the traditional levels of communication although none of the interaction is FtF (see Table 1, adapted from Rogers [117]). Using the Littlejohn model, email is private, interpersonal communication; IRC, MUDs and listservs are group interactions. While newsgroups often feel like small groups with social interaction and attention to the individual, many have large numbers of participants and more closely resemble organizational communication. Size is not the only criteria which determines whether the level is organizational. Many large Usenet newsgroups are far from cooperative with little evidence of structure, goals or function other than the posting of opinions. Because the messages are accessible to anyone, Usenet newsgroups often resemble mass communication, but the source is not a complex institutional organization. Usenet is therefore difficult to place in any one level and is a hybrid of several levels. The Rogers model is less satisfying because all the Internet services are lumped into one category, interactive machine-assisted interpersonal communication, and therefore is not included in Table 1. Because these models ignore any time, space or channel effect on the interaction, they do not adequately describe the differences in these services.

Table 1.  Levels vs. Aspects of Communication

          inter-       group       organizational         mass
          personal
          ----------------------------------------------------------
sender     one          one          one or more     initially one, 
                                                   maintains control

receiver   one       few, may         large            large 
                    take turns       audience         audience

channel    FtF or      FtF or         FtF or          mediated
          mediated    mediated       mediated

examples    FtF,      lecture,       corporate       TV, radio,
          letters,   discussion      networks          books
          telephone   listserv
           email

Another comparison of Internet interactions uses types of conversation. The traditional types of conversation are monologue, dialogue, and discussion [124]. A monologue consists of one sender, one or more passive receivers and a message that may be delivered FtF or via mass media. A dialogue contains a sender and an active receiver taking turns with the sender; traditional dialogue can be oral or written messages. In a discussion, one person starts as the sender with multiple receivers who take turns as senders, but the initial sender retains control of the conversation. Because traditional conversation does not consider the effect of computer mediation, Shank added the category of multilogue for the Internet conversation with an initial sender, multiple receivers who take turns as senders, but with loss of control of the conversation [124].

Again, the Internet interpersonal communication services can be categorized using the conversation model although none of the interaction is FtF (see Table 2). Email is usually dialogue but if messages are ignored, it can be a monologue. Listservs or Usenet newsgroups may be monologues if they are moderated and if the moderator allows only a single source to produce the messages. Because control of the conversation may be retained by the moderator, moderated groups can become discussions, although interruptibility and turn-taking take on different meanings from the FtF model [108]. Unmoderated listservs and newsgroups are generally multilogues as are IRC and MUDs. Again, the conversational model ignores any time, space or channel effect on the conversation and does not fully describe the differences in these services.

Table 2. Type of Conversation vs. Aspects of Communication

          monologue     dialogue      discussion       multilogue
          ---------------------------------------------------------
sender      one           one        initially one,   initially one,
                                    maintains control   no control

receiver  one or more,  one, active   one or more,     one or more,  
            passive                     active           active

channel    FtF, mass      FtF or        FtF or          computer-
          media, other  mediated       mediated          mediated
           mediation

examples  lecture, TV,     FtF,          FtF,           IRC, MUD, 
          books, radio,  letters,      moderated       newsgroups,
          mailing list   telephone,     groups          listserv
                           email

A third comparison uses the time/place taxonomy, shown in Table 3, that was originally used to classify groupware systems [37]. This approach emphasizes the messages as well as the people involved. The four categories are FtF, asynchronous, synchronous distributed, and asynchronous distributed interactions. If the sender and receiver are in the same place and the message is received at the same time it is sent, then the interaction is FtF and synchronous. If the sender and receiver are in the same place but the message is received at some later time than it was sent, then the interaction is called asynchronous. Leaving a note on the refrigerator for your child is an example of asynchronous interaction. If the sender and receiver are in different places yet the message is received at the same time it is sent, then the interaction is synchronous distributed. A telephone conversation is synchronous distributed interaction. Lastly, if the sender and receiver are in different places and the message is received at some later time than it was sent, then the interaction is called asynchronous distributed. Sending (and receiving) a letter is asynchronous distributed interaction.

Table 3. Time/Space Taxonomy

           Same           Different  
           Time           Times
           ---------------------------
Same       FtF            asynchronous
Place      (synchronous)

Different  synchronous    asynchronous  
Places     distributed    distributed

Because the interpersonal communication services of the Internet are computer-mediated, none are FtF. Email, listservs and newsgroups are asynchronous distributed interactions; the senders and receivers are in different places and messages are delivered some time after they are sent. The collocation metaphor, the collocated feel of the interaction and the intense involvement of the participants in IRC and MUDs belie the fact that these are not collocated interactions; for some, these interactions are more real than real (FtF) life [22]. Considering the sender's and receiver's physical locations and the real-time delivery of messages, IRC and MUDs are synchronous distributed interactions.

Next, I present a socio-visibility taxonomy that I developed that looks at the number of senders and receivers and whether the interaction is public or private. I use the term dialogue for 1-1 communication. Broadcast is then one-to-many (1-N) and multicast is many-to-many (N-N). Public and private interaction refers to the reception of the message. If the message is received only by the receiver, then it is private. If the message is displayed or made available to an audience other than the sender and receiver, then it is public. A conversation between two people on a busy street corner is a public dialogue; a telephone conversation is a private dialogue. Radio and television are public broadcasts; an advertisement circular is a private broadcast. Computers have made multicast interaction possible.

In this model of interaction, different relationships become apparent (see Table 4). Email is a private dialogue. With a distribution list on an email service, a single message can be delivered to emailboxes of multiple receivers--private broadcast interaction. In a listserv multiple people send messages that are delivered to the emailboxes of multiple receivers, therefore listservs are private multicasts. Because the multiple messages of newsgroups, IRC, and MUDs are publicly viewable, that is viewable by more than just the intended recipient, these interactions are public multicasts.

Table 4.  Socio-Visibility Taxonomy

         1-1            1-N           N-N
         -------------------------------------
Public   conversation   political     Usenet,
         on the street  speech        MUD, IRC

Private  telephone,     distribution  listserv
         email          list

In providing detailed units of analysis for the study of CMC, December [31] argues that protocol, time, distribution scheme and media type be included in distinguishing characteristics of Internet communication. December uses TCP/IP as the only Internet protocol with other networks using TCP/IP or other protocols [31]. He uses protocol to distinguish between Internet and LAN (local area network) communication. The time delay of messages varies between nearly instantaneous, called synchronous in the time/place taxonomy, and persistent. Messages that are available for on-demand retrieval are called persistent communication and are characteristic of Internet information services. Distribution schemes include point to point, point to multipoint, point to server broadcast, point to server narrowcast, server broadcast and server narrowcast. The latter two refer to Internet information services and are included only for completeness of the descriptions. In point to point distribution a single sender transmits a message to a single receiver. In point to multipoint a single sender or application sends a message to many receivers. Because the client-server model of Internet is integrated into the distribution schemes, these schemes are more than message flow indicators of 1-1 or 1-N. In point to server broadcast, a single sender or server transmits a message to a server which makes the message available to other servers or to receivers with the appropriate client software; in point to server narrowcast the message is available only to a specific group of receivers, identified with passwords and/or IDs. In the server broadcast scheme, the server has stored information that is available to anyone for retrieval with an appropriate client; the server narrowcast scheme again restricts the information to an authorized set of users. Media type includes text, sound, graphics, images, video, executable files, hypertext and hypermedia [31].

Email uses a point to point distribution scheme; listserv and email distribution lists, point to multipoint schemes. IRC and Usenet use the point to server broadcast and MUDs use point to server narrowcast. The dominant media type used in all of these services is ASCII text. With the programming capabilities in MUDs, executable files, graphics, images and sound are being introduced.

The next category for comparison of Internet services is interactivity. In their investigation of computer-mediated groups, Rafaeli and Sudweeks investigate the degree of interactivity among participants in newsgroups [108]. The degree of interactivity with its emphasis on the message is in contrast to the other categories which emphasize the relation between sender and receiver. Interactivity is a process-related construct about communication that describes the extent to which messages are related to each other. Rafaeli and Sudweeks argue that interactivity is a continuum--declarative or one-way communication on one end, fully interactive communication at the other end, and reactive or two-way communication somewhere in between. The focus on interactivity also supports their argument that FtF conversation not be used as the standard of comparison for group CMC. One-way declarative communication has unrelated messages relayed from a source to an audience. Two-way reactive communication is similar to a dialogue with the sender and receiver taking turns, but the new message is in response to the previous one. In full interactivity later messages in any sequence not only take into account the previous messages but also the reactive manner of the previous messages. Thus all communication falls short of full interactivity [108].

Although individual sets of interactions from the interpersonal communication services of the Internet can be ranked as more interactive than other sets, the services are not easily classified as one type of interactivity or another. Email is often two-way reactive, but it can resemble one-way communication when messages are ignored. It also has mechanisms built into the service that facilitate interactive communication. These services include subject lines and markers for included text from other messages. Listservs and Usenet newsgroups have these same services and therefore the interactions on them may be in any mode along the continuum. IRC and MUDs have no built in mechanisms to facilitate interactive communication other than the textual trace of conversations that remain on the screen. As such, they tend to be two-way reactive, sometimes one-way and only limitedly interactive.

Each of the above comparisons does not fully describe the differences among the interactions in these services, therefore additional characteristics need to be specified. I assert that the form of messages, the skill level required for full participation, the environment and the metaphors used in these services provide further insights to the types of interactions. The form of the message may be encapsulated, fragmentary or operative. An encapsulated message is a bounded whole, resembles a letter or traditional written communication, is asynchronous and uses contextual support mechanisms such as address spaces and subject line. A fragmentary message is delivered in pieces, resembles traditional spoken communication, is synchronous and has no contextual support other than the screen display. An operative message is a message that carries out some action and is accomplished via a program or macro. Skill levels vary from low to high. An example of low skill requirements is basic editing skills with few commands. High skill levels require programming expertise. Interpersonal communication environments are usually provided but may be individually constructed. Lastly, the underlying metaphor of each service (e.g., rooms) frames the interactions.

These four characteristics help complete the descriptions of the interpersonal communication services available on the Internet. Email, listserv and newsgroups have encapsulated messages; IRC and MUDs have fragmentary ones. In addition, MUDs have operative messages that construct the environment and provide actions for the participants. Because email, listserv and newsgroups require only basic editing skills and only a few commands for reading, sending, and replying, they have low skill requirements. Besides talk, IRC has multiple commands for emoting, movement between rooms, reading and posting messages to a message board, locating people, creating channels, and removing people from channels. Depending on the commands used, the skill level for IRC ranges from low to moderate. In addition to the conversation, MUDs have an additional programming component to build the environment, to construct agents, and to structure the action. Depending on the set of commands used, the skill level for MUDs ranges from moderate to high. The environments of MUDs are constructed; all other services use the existing setting provided. The underlying metaphors, the structure of the interaction provided by the software, also differ. Email and listserv use a postal service metaphor; IRC uses a rooms metaphor; Usenet is a discussion or a conversation and the constructed environments of MUDs provide fantasy worlds. Table 5 gives a summary of the characteristics presented.

Table 5. Characteristics of Internet Interpersonal Communication Services

        email        listserv     IRC         MUD         Usenet
        -----------------------------------------------------------
leve    inter-       group        group       group       hybrid
        personal
 
conver- usually      monologue,   multilogue  multilogue  monologue,
sation  dialogue     discussion,                          discussion,
                                                          multilogue

source  one          many, can    many        many        many
                     be one

re-     one          many         many        many        many
ceiver

channel computer     computer     computer    computer    computer

distri- point to     point to     point to    point to    point to
bution  point        point        server      server      server  
                     multipoint   broadcast   narrowcast  broadcast

media   text         text         text        text,       text
type                                          executable

message encapsulated encapsulated fragmentary fragmentary encapsulated
                                              operative

tem-    asynchronous asynchronous synchronous synchronous asynchronous
poral

spatial distributed  distributed  distributed distributed distributed

so-     dialogue     multicast,   multicast   multicast   multicast
ciality              can be
                     broadcast

visi-   private      private      public      public      public
bility

inter-  usually      ranges       usually     usually     ranges
active  reactive                  reactive    reactive 

skill   low          low          low to      moderate    low
                                  moderate    to high

envi-   existing     existing     existing    constructed existing
ronment

meta-   postal       postal       room        fantasy     discussion, 
phor                                          world       conversation

Thus, each of the interpersonal communication services of the Internet are very different services with very different interactions. December argues that CMC research "as not led to much successful theoretical integration or cross-study comparisons" [31] due to the disparate units of analysis. By characterizing each of the Internet interpersonal communication services, I provide basic features for distinguishing among CMC systems and research settings.

USENET

Although Kehoe describes Usenet as "the set of machines that exchange articles tagged with one or more universally-recognized labels, called newsgroups" [70], it is considered more complex by its participants. According to the alt.culture.usenet FAQ (Frequently Asked Questions) [120], Usenet consists of the newsgroups, networks of computers that transmit news via various protocols, and the community of people who read news. In this section, I describe the organization and distribution of the messages in Usenet. From this description, I elaborate on Usenet interaction and the social dilemmas of that interaction.

Usenet has been called an "anarchic global conversation" [114]. Anarchy refers to the lack of any central governing authority on either policy or technical levels. Other descriptions include a world wide public conferencing network, an electronic magazine, and an electronic town meeting [52]. The readership data in Table 6, posted to news.lists and published on the Internet, give an indication of the size of Usenet news service [110]. Although Usenet was originally formed to discuss Unix, it now distributes thousands of conversations on thousands of topics [114] ranging from abdominal training, books, computer languages, food, politics, raising children, sex, ZyXel modems and almost everything in between. There are at least 6,000 newsgroups [7] (this number does not include regional and local newsgroups that have limited distribution).

Table 6. Usenet Readership Summary Report of July 1995 
         (Estimated) [105]

Sites:                                        330,000
Users with accounts:                       30,329,000
Netreaders:                                11,033,000

Average traffic per day (megabytes):              586
Average traffic per day (messages):           127,446

The contents of a newsgroup are electronic letters called posts or articles. To help readers find the posts that interest them, Usenet is divided hierarchically into newsgroups, which indicate the topic of conversation [100] as well as the computer directories in which the articles are stored [9]. There are seven major categories whose newsgroups are created by a clearly defined set of guidelines involving formal discussions and a voting procedure [79]. The categories are "comp" for topics in computer science, "misc" for topics not easily classified, "news" for topics on Usenet news, "rec" for recreational activities, "sci" for topics in science and technology, "soc" for topics on societies and social issues, and "talk" for debate-oriented discussion [80]. Additional categories include alt, gnu, bit, local categories for universities and regional categories for states and countries [52]. Newsgroups in the alt hierarchy do not have alternative topics but are created without having to go through any discussion or vote [9]. AGM is a newsgroup in the alt hierarchy; the "good.morning" part of the name is an indication of the discussion topic in the newsgroup. Thus the newsgroup name serves multiple purposes. It is the address for messages and the topic of conversation. It is used to refer to the messages themselves, to the discussion and to the community of people who read those messages.

Not all newsgroups are carried at all sites and not all messages posted to a newsgroup get distributed to all sites which carry that newsgroup. Some sites carry only a subset of the technical newsgroups and some sites do not carry the more controversial newsgroups, especially those in the "talk," "soc" and "alt" hierarchies [80]. Usenet distribution does not guarantee that every site will receive every message; some sites may send messages to the next Usenet site but not send them to any previous sites in the distribution cycle. Because Usenet sites store messages for different lengths of time before forwarding them, delay can accumulate and messages can expire [12]. Some sites have limitations on the broad distribution of messages causing some messages to stay within a localized area. In addition, computer systems need maintenance or fail and messages do not get distributed to those sites in a timely fashion. Some Usenet newsgroups are now available using WWW browsers as an alternative access route to the articles.

Usenet interaction is carried out using software called newsreaders. Newsreaders organize and display posts, filter posts, keep track of which posts you have read, and allow you to reply to posts or to create new messages. Thus, Usenet interaction takes two forms, reading and posting. Lurkers limit themselves to reading without contributing or contributing only rarely; posters write the messages. Readership data supports the contention that lurking is the principal mode of participation on Usenet [110]. This is also the case with AGM, with an estimated 6,100 readers [7] and yet during a six-week detailed analysis, there were only 427 posters; half (214) of them posted only one message.

Not all messages that are posted to a newsgroup are relevant to that discussion. Crossposting is sending messages to multiple newsgroups by entering more than one newsgroup name in the newsgroup line of the header [40]. Many crossposts are the result of replying to an originally crossposted article [61] increasing the traffic on all newsgroups. A closely related phenomenon is spamming. Spam originally referred to commercial messages that were crossposted to many and sometimes all newsgroups [40]. Massive crossposting is now referred to as spamming whether or not the message is commercial. Netiquette (net etiquette) suggests that only five or six groups have similar enough interests to warrant crossposting and the sender should warn the readers that the article is crossposted [48]. Newsreader software can be configured to display only those articles with a preset limit to the number of newsgroups in the newsgroup line, filtering out articles sent to large numbers of groups. A campaign was recently started by a site administrator to transmit crossposted articles to no more than a specified number of newsgroups; he recommends fifteen as the limit [68]. MCI and other Internet service providers have recently published policies to discourage spamming on their systems, recognizing that the abuse of resources wastes time and money [46, 62, 140]. Another set of irrelevant posts, flames, flame-wars and flamebait, are related types of messages. Flames are defined as messages "intended to insult or provoke" [120], "vitriolic on-line exchanges" [34], "verbal conflagrations" [40] or "electronic diatribes" [9]; flame-wars occur when large numbers of flames are exchanged; flamebait is a posting designed "to elicit a vituperative reaction" [40]. Netiquette suggests that flames are appropriate only to alt.flame [91]. The last type of irrelevant post is the troll. Trolling is posting a message specifically worded to generate followups on a trivial topic; the intent of a troll is not to start a flamewar but to "instruct readers to ignore obvious drivel by making the replyers feel utterly stupid" [120].

Other irrelevant posts are the gratuitous reply [48] and the question for which answers are readily available off-line [61], both of which are considered a waste of bandwidth [91]. The gratuitous reply or "me, too" post is better sent by private email [91]. Most questions fall into two categories--questions about the newsgroup and computer system questions. Questions about the newsgroup are easily answered by reading the FAQ; systems questions are usually answerable by reading the manual [61].

Relevant posts are dependent on the topic and social norms of the newsgroup. Little scholarly work has been done on the typical, mundane posts of newsgroups [13]. The ProjectH database, Baym's study of rec.arts.tv.soaps and this study of AGM are representative of the few academic studies of newsgroups. ProjectH and Baym's work are covered in Chapter IV, while Chapters V through VIII contain the results of this study.

Reading messages is not the same experience for every person in a newsgroup. The asynchronous nature and differences in delivery times of messages to some sites causes messages and their followups to be read in different sequences. Followups to a post may precede the original message and may not include the text of the original. The sequence of messages is also dependent on how often the newsgroup is read and whether the newsreader supports threading. A thread is a group of messages with the same subject line [40]. Messages can be read sequentially as they arrive or in threads. Reading messages in bulk, once a week or so, is not the same as reading daily. Messages accumulate until they are read. If the newsgroup is very active and the site has limits on how many messages are retained, earlier messages may be deleted. Threaded messages read in bulk can present a more coherent conversation because larger numbers of messages are grouped in the thread, yet reading this multilogue in bulk loses the dynamism [114] of the message exchange that is experienced by frequent readers and posters. Reading messages daily, whether threaded or sequential, can be a more disjointed experience because fewer messages are in a thread. When you read messages sequentially, you have to construct the conversation beyond conversational involvement [139] to include your knowledge of previous posts, what the interaction is about, who the post came from, what their intentions may be and where the post fits in the context of the newsgroup interactions. Since all messages are not received at all sites, some posters include the full original post in their reply, often at end of their reply, which aids contextualization. Even though this practice wastes bandwidth, it is considered helpful by participants at sites with inconsistent message delivery.

Other aspects of Usenet interaction are anonymity, identity and the social nature of posting. Anonymity has been linked to the removal of social norms and increased flaming [149] and is discussed in Chapter IV. At the outset, identity on Usenet is determined by the email addresses of posters [35] which provide a type of anonymity. Newsreader software provides additional information about posters as do the posters themselves. Identity is a constructed attribute on Usenet and is discussed in Chapter VI, as is the social nature of Usenet.

These various features of Usenet frame a unique form of communication that is rife with contradictions. Active participation by posters in a newsgroup is essential to its success yet this participation must be balanced against restraints on bandwidth, both the limited capacity of Usenet to store and transmit articles as well as the capacity of subscribers to read and absorb that information [71]. Global in scope, Usenet continues to universalize community [41] by providing connections among geographically dispersed people who would otherwise never meet [12] with its tolerance of a wide diversity of opinion [114]. At the same time it reflects the "fragmentation, hierarchization, rigidifying social boundaries, and single-niche colonies of people who share intolerances" [114]. Although newsgroups are organized by topic, the asynchronicity of the interaction, with people reading the group at different times and rates, with simultaneous multiple threads, and with discussions spanning days or weeks, presents considerable obstacles to topical and group coherence [12]. It mirrors the distinctions between the technically sophisticated users and the less knowledgeable new users [78], the rich and the poor, the highly educated and those deprived of advanced education [41]. Usenet provides a means to get to know people while maintaining a distance from them [114]. It provides a means to obfuscate identity but within which people are more intimately self-revealing [114, 151]. These contradictions define the social space of Usenet and the social dilemmas within that space.

Usenet interaction is chaos and order. It is noisy free speech and quietly civilized discussion. Usenet is a place without spatiality, distributed among nodes of various networks in numerous countries. "Usenet" is like a herd of performing elephants with diarrhea--massive, difficult to redirect, awe- inspiring, entertaining, and a source of mind-boggling amounts of excrement when you least expect it'' [129].

This chapter has been an overview of Usenet, its place on the Internet, and its interactions. Usenet is a hybrid of interpersonal, group and mass communication. It can look like a monologue, a give and take discussion or a multilogue. It is a many-to-many asynchronous, distributed multicast transmitted by computers in a point to server broadcast scheme. The messages are encapsulated, public texts. Usenet is relatively simple to participate in and provides a range of interactivity dependent on the energies and inclinations of its members. It is a global conversation, a global town-meeting complete with all the contradictions of any social space. In the next chapter, I present previous research on CMC and Usenet.


Next Chapter
References
Back to Contents
Holly Patterson, September 1996.
Comments to Author: hollyp@falcon.tamucc.edu
http://www.sci.tamucc.edu/~hollyp
Copyright © 1996, Holly Patterson



|| SITE GUIDE || Updated October 1997 || WEBMASTER ||
 
This page is maintained by SDL/Spud