Social Interaction on the Net:
Virtual Community as Participatory Genre*

Thomas Erickson
Apple Research Laboratories
(now at) snowfall@acm.org

Abstract

The phrase "virtual community" is often used to describe long term, computer-mediated conversations amongst large groups. This paper suggests that such conversations may be better viewed as instances of a participatory genre, rather than as community. A genre-oriented analysis is useful because it encourages a focus on the medium within which the discourse is embodied. As an example we analyze an on-line conversation from the perspective of genre: we identify its communicative purpose, regularities of form and substance (such as word play and affirmation), and the situation which gives rise to these regularities. We then examine ways in which the discourse medium supports these regularities and enables participants to establish and reinforce the conversation's underlying conventions. More generally, we believe that genre-oriented analyses such as this can play an important role in the design of participatory media.

1. Introduction

I'm interested in the issue of how to support computer-mediated social interaction among large groups of people, particularly long term, textually-mediated interaction. This type of interaction--which I'll refer to simply as on-line discourse--is often discussed using the conceptual framework of virtual community (e.g., see Rheingold [14] for a survey). Indeed, virtual community has been applied to synchronous chat systems such as IRC [13], asynchronous conferencing systems such as the WELL [15, 17] and netnews groups [2], and systems like MUDs and MOOs that provide both synchronous and asynchronous communications (e.g., [3, 6]).

While virtual community is an engaging and provocative notion, the concept of community is not always well-suited to describing on-line discourse. In particular, the frame of community offers little guidance to those of us interested in designing the infrastructure for supporting on-line discourse.

In this paper, I suggest a complementary framework that provides a useful shift in emphasis: namely, using the concept of genre to analyze and understand on-line discourse. Genre shifts the focus from issues such as the nature and degree of relationship among "community members", to the purpose of the communication, its regularities of form and substance, and the institutional, social, and technological forces which underlie those regularities.

At the same time, genre is not a perfect fit: on-line discourse has some differences from traditional examples of genre. One difference is that on-line discourse is highly participatory in nature. Whereas most genres have a distinction between producer and consumer, or author and audience, in on-line discourse the distinction between the producer and the consumer is blurred. A further distinction is that the turns in such discourse can succeed one another very rapidly, thus allowing the conventions that underlie the discourse to be shaped, reinforced, or renegotiated much more rapidly than in traditional genres. Nevertheless, a genre-oriented analysis is particularly useful because it encourages a focus on the medium within which the discourse is embodied (hereafter the discourse medium), and the way in which that medium allows the participants to understand and shape the underlying conventions of discourse.

This paper begins by laying out the basic implications of community and genre as frameworks for analysis, and arguing that there are many interesting examples of on-line discourse that do not fit well into the community framework. Next, we look at an example of on-line discourse: a conversation from an interactive salon called "Cafe Utne." The genre framework is used to analyze the conversation, and a number of the conversation's characteristics are identified. Particular attention is paid to the ways in which the discourse medium enables participants to establish and reinforce these expectations. Finally, we conclude with some reflections on the design of discourse media, and the utility of genre analysis as a means to that end.

2. Community and Genre

2.1. Community

Let's begin by taking a look at some of the implications of the term "community." While "community" has a vast number of definitions (for example, see [9, 5, 16, 17, 8]), it commonly suggests the following:

• Membership: central to the notion of community are issues of membership and exclusion. Some people are in, others are out. Communities range from being open to anyone who shares particular ideas or interests to communities accessible only to those who meet certain criteria of geography, ethnicity, gender, etc.
• Relationships: Community members form personal relationships with one another. These relationships can run the gamut from casual acquaintance, to friendships, to deep emotional bonds. Yet, communities as generally conceived are too large for everyone to know one another well; instead, a community is best viewed as partially overlapping networks of relationships.
• Commitment and generalized reciprocity: community implies a sense of mutual commitment to the community: one member may help another simply because they belong to the same community, not because of a personal relationship.
• Shared values and practices: community members may share a common set of concerns, values, goals, practices, procedures and symbols. Communities typically have a shared history, and shared artifacts and places.
• Collective goods: Communities participate in the creation, control, and distribution of various collective goods.
• Duration: the aforementioned characteristics take on importance only because the community as a collectivity is expected to have a long existence.

Now, while community does indeed seem to be a suitable frame for some sites of on-line discourse (e.g., certain MUDs and MOOs), there are many sites of discourse where issues of membership, shared values, relationships, mutual commitment, duration, etc., seem to apply weakly, if at all. For example, The Palace(TM) [12] is an internet-accessible graphical environment permitting people to connect, and carry on real-time, textually-mediated multi-way chats while represented as graphical avatars. The Palace sites I have visited (any registered user of The Palace may create their own site) seem a cross between a cocktail party and a cruising zone: the anonymity, the superficial nature of most conversations, and the ephemeral nature of the interactions seem distant from any notion of community--yet, nevertheless, still of interest to designers and users. Another example is Ferndale(TM) [7], an on-line, interactive soap opera accessible via the world wide web. Ferndale features a cast of characters, a virtual place, and an on-going narrative with which visitors can interact by exchanging email with the characters, or by participating in scheduled, on-line chat sessions. Ideas suggested (or roles enacted) by visitors may show up in future episodes of Ferndale. While Ferndale is a fascinating experiment, characterizing it as virtual community likewise seems problematic. Ferndale's self description as an interactive soap opera--with the implied separation between the audience and the cast--seems much more apropos.

More generally, on-line discourse may be useful and engaging to its participants even if the participants form no lasting relationships, even if they share few values, and even if they know that in a pinch they can't count on one another. On-line discourse may be of value even if--to take a hypothetical case--each person participated once, and only once, and then never returned. What is important, in many cases, is the communication itself--the shared informational artifact that is created by the participants--rather than a real or perceived bond among the participants in the communication. In these cases, I believe that genre provides a more useful framework for making sense of the sort of activity that is happening. Genre analysis is also useful because it shifts the focus from the participants and the putative relationships among them to shared artifacts (that is, instances of the genre). This stance is particularly useful to designers of on-line systems because it focuses on the very things over which designers have some control: the conventions of form and content that typify a genre.

Let's take a closer look at the concept of genre.

2.2. Genre

The concept of genre has changed considerably over the last several decades. Traditionally genre has been defined as a classification of types of spoken and written discourse in terms of their form and substance. Examples include elegies, epics, and encyclopedia articles. However, more recently, scholars from a variety of fields have taken a more situated approach to the definition of genre (see Bazerman [2] for some of the converging streams of work). Miller [10], in a seminal paper, has argued that genre are typified rhetorical actions carried out in response to socially defined, recurrent situations. That is, genres provide ways and means of accomplishing social actions in particular situations; it is recurrent situations--with their similarities of purpose, audience, and other constraints--that give rise to regularities of form and substance in genre as writers or speakers attempt to achieve their communicative ends. Swales [18] has built on this conception of genre by developing the notion of discourse communities. Members of a discourse community are those who participate in a genre: they have shared goals, they communicate with one another, and they use various participatory mechanisms to provide information and feedback. So we have arrived back at the notion of community, albeit one that is weaker than the traditional conception: discourse communities are more in the background, as mechanisms for supporting conversation, rather than as an end in themselves.

A variety of analyses have been based on this notion of genre. Most relevant to this paper's focus is work by Yates and Orlikowski [20] on the evolution of business memos. They trace the evolution of the business memo genre as it emerges from the business letter genre around the turn of the century, in response to institutional changes (larger companies, new managerial philosophies), technological changes (the advent of typewriters and vertical filing), and social pressures (mandates from upper management; the development of customs in the new professional of typist). Similarly, they examine changes in the memo genre brought about by the shift from typewritten paper memos to email memos, again calling attention to a complex interplay of institutional, technological, and social forces. Their overall claim, that genres evolve over time through reciprocal interaction between institutionalized practices and individual human actions is of particular interest here, because on-line interaction has the potential to greatly speed up the evolution of genres.

This is a very brief sketch of the concept of genre, but it is enough to provide the background for an analysis of on-line discourse. In summary, looking at on-line discourse through the frame of genre suggests a focus on:

• the communicative purpose of the discourse
• the nature of the discourse community
• the regularities of form and content of the communication, and the underlying expectations and conventions
• the properties of the recurrent situations in which the genre is employed, including the institutional, technological, and social forces that give rise to the regularities of the discourse

3. An Example: Cafe Utne

In the next two sections of the paper, I describe an example of on-line discourse and analyze it in terms of genre. Because my ultimate goal is to understand how to design effective systems for supporting on-line discourse, the principle focus will be on the way in which the underlying technology embodied in the discourse medium shapes the characteristics of the discourse.

3.1. About Cafe Utne

Cafe Utne is an on-line conversational salon run by the Utne Reader magazine. The Cafe is billed as an "informal gathering place for people who want a fun, relaxing, and harassment-free atmosphere," and who "like discussing ideas & issues in a thoughtful, respectful manner." [4] In actual fact the Cafe lives up to its billing, with conversation there being generally polite, friendly, and thoughtful.

Begun in October of 1995, Cafe Utne's current membership is over 8000, and it is said to be one of the busiest "Web conferencing communities" on the Internet, with over 124,000 messages posted in 2,000 different conversations. In an electronic newsletter optionally sent to participants, its management writes: "In August, 1,744 different people (45% female) visited the Cafe, adding 21,822 posts. Just over 35% of the visitors posted messages." [19]

Membership in Cafe Utne is free, although people must register to join. One of the policies of Cafe Utne is to maintain a gender balance of at least 1:2; this policy is implemented by having a waiting period--which can range from a few days to about two weeks--for males who wish register. Once they are members, people can connect to the web site and, by entering a name and password, can then navigate a set of web pages, each of which contains a conversation (or meta-information such as a list of conversations about a particular topic).

In general, members can participate in any conversation, and anonymity is not permitted, although the Cafe has been experimenting with a few private and/or anonymous conversations. While the primary usage model is a place where people come, and find conversations in which they're interested in participating, Cafe Utne also regularly invites special guests to participate in on-line panels and open discussions.

The Cafe appears to be supported by sponsorships, and has also begun soliciting people to become voluntary paid members as is done by public access radio and TV. The Cafe also draws on and refers to content from the Utne Reader magazine (available by paid subscription), and thus undoubtedly plays a role in publicizing and stimulating interest in the magazine.

3.2. A Note on Method

This is by no means a comprehensive study of Cafe Utne. While I have participated in the Cafe from its beginning, my attention has been confined to about half a dozen conferences, and to one to half a dozen conversations within each conference. Thus, this paper does not reflect a comprehensive analysis of the content of conversations within Cafe Utne. Given the size of Cafe Utne this would be a considerable task. The analysis of content described here draws upon a single conversation, in a single conference. While this is appropriate for the purposes of this paper--namely, exploring the utility of genre as an analytic framework--the reader should be wary of generalizing from the particularities of the analysis to Cafe Utne as a whole, or on-line discourse in general.

3.3. The Cafe Utne Interface

The interface to Cafe Utne is provided by a software application called Motet [11], which generates html that may be viewed through forms-capable web browsers. The interface is principally textual buttons arrayed on a web page (see figure 1 for an example). The controls of the user's web browser can also be used to navigate among pages already displayed.

The content of Cafe Utne is structured in terms of a simple hierarchy. There is a main page, which provides access to the conferences (as well as general information); each conference has a page (e.g., figure 1) that provides access to its of topics (i.e. conversations); and each topic has a page which contains the conversation (e.g., figure 2).

Figure 1. Part of the Cafe Utne interface: a topics page.

3.4. An Example Conversation

Now lets look at a segment of a conversation (see figure 2). This is an from the "Introduce Yourself" conversation in the "InfoAge" topic. (Note: names and identifying information has been changed; conversation quoted with permission of participants.)

In the example below, each turn in the conversation is separated by dashed lines. Each message (also known as a posting) begins with a header that specifies the conference name; topic and comment numbers; the user's "handle" (usually the user's name--though it can be changed when the user enters the conference) and ID; the date and time of the comment; and the length of the comment. The header is followed by the body of the message which is relatively unstructured. We'll consider features of its form and substance in the next section.

Figure 2. Excerpt from Cafe Utne conversation

---

InfoAge.2.59: Kitty Feldman (kf ) Mon, 27 Nov 95 18:20:39 CST (14 lines)

Hello all. I've just found the web and cyberspace in general after 13 years of computing in the mainframe environment (I feel like I've come out of a cave!) I work at a medium-sized midwestern university providing support for administrative clients using student-based information.

I'm very interested in the political aspect of the virtual community thread that's been started here and look forward to jumping into that discussion.

To all the welcomers: thanks for the encouragement!

My web page:http://www.*.edu/*/*/*/kf.html

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

InfoAge.2.60: Harold Stevens (hs ) Mon, 27 Nov 95 22:50:50 CST (7 lines)

Welcome aboard, Kitty!

Programming mainframes. That sounds serious . . . like housetraining dinosaurs.

I like cats too. But feel right at home and jump to any topic you like. It's a pretty friendly environment.

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

InfoAge.2.61: George Smith (gs ) Wed, 29 Nov 95 14:30:41 CST (8 lines)

Greetings, y'all. I haven't been down in the WELL. I haven't swimmed the River. I'm a new Cafe lurker wondering "where's the context for virtual community?" IRL, I'm a dad of a 2-

year-old son; 32-year-old communications designer for a music store in Orlando; musician; writer; artist; activist. Virtual Community is an interesting concept, but does it further isolate us from physical real life? I'll check out these links and hope for insight. Thanx to all for intelligent postings! Peace, George

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

InfoAge.2.62: RD Hoffman (rdh ) Wed, 29 Nov 95 18:14:56 CST (4 lines)

Hi everyone. I'm also floating in here from the Well. I'm a reporter based down in LA and am enjoying an incredibly sunny November. I tend to be a bit quiet, so I'll probably only pop into the discussion every once in a while. This cafe definitely looks promising so far.

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

InfoAge.2.63: Lori Jones (lj ) Wed, 29 Nov 95 22:30:24 CST (1 line)

A reporter in LA, eh? Do you know woodrow?

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

InfoAge.2.64: RD Hoffman (rdh ) Thu, 30 Nov 95 19:07:57 CST (1 line)

Lori: Wilson? He's not up for commenting on much these days.
------------------------------------------------------------

---

Figure 2. Excerpt from Cafe Utne conversation

3.5. Cafe Utne as a Genre

Is Cafe Utne a virtual community? While the term "community" is used in describing the cafe, it is easy to raise questions about its applicability. The Cafe's size, at 8000 members, is awfully large for a community: while it might very well be a convivial assemblage of strangers, it seems likely that hallmarks of community like overlapping networks of relationships, generalized reciprocity among members, and the absence of a shared history and symbols, are absent or at least weak. However, the point of this paper is that the question of community is irrelevant for our purposes. Clearly, judging by its activity and continued popularity over the first year of its existence, the Cafe does provide value to its membership, and seems a site of activity worthy of investigation.

Let's look at the Cafe Utne conversation as an instance of a genre. For the purposes of our analysis, I'll focus on four aspects of genre: the situation to which the genre is a response; the genre's communicative purpose; and its regularities of form and substance.

3.5.1. The Situation

The participant in the conversation (I use participant to refer to either reader or contributor) is located in front of a computer, reading a screen of text. The size of the audience is unknown to the participant, although we may assume it to be large and composed primarily of strangers. Other participants are not present: this is true in the obvious sense that they are not physically co-located, and also true in the computer-mediated sense, in that participants (even if on-line at the same time) have no representations within the Cafe. The only signs of participants are the messages they may type (this may be contrasted with a MUD, where there is a textual representation of a participant that is "visible" to anyone else in the same area). The consequence of this is that the participant is therefore free of many of the social pressures to participate that are present in a face to face conversation.

Yet, for the conversation to be a success, the participant must move from the role of reader to that of contributor. There are a number of barriers that must be overcome: the participant must have a desire to communicate; the participant must have an opening for his or her contribution; and the participant must actually be able (in the sense of being able to use the software) to contribute. As we shall see, much effort--on the part of conversational participants, as well as on the part of the system administrators and designers--is devoted to overcoming these barriers.

3.5.2. Communicative Purpose

In general, the communicative purpose of Cafe Utne is simply to have polite, friendly, and thoughtful, topic-oriented conversations.

In the conversation excerpted in the previous section, the purpose of the conversation is for people interested in the "Info Age" conference (a set of conversations on digital culture and virtual community) to introduce themselves. It's purpose is explicitly stated as "Say a little bit about yourself and your interests in this area. Extend a welcome to other newcomers. Include a link to your homepage, if you have one...." (By convention, the first item in each conversation describes its purpose; similarly, "introduce yourself" conversations like this are a feature of each conference in Cafe Utne.)

3.5.3 Regularities of Form

Now let's look at some of the regularities of form found in Cafe Utne conversations. Most, though not all, of these regularities are reinforced by the software, and so are found throughout conversations in the Cafe.

• Text-Based. One property of Cafe Utne conversations is that they are textual in general, and presented as world wide web documents in particular. Thus, participants can enter typed text, and also create or follow links to other web documents either inside or outside the Cafe. Although in principle users could link pictures and other non-textual media of relevance to their Cafe conversation, in practice I haven't seen this happen except in links to outside home pages, where the pictures, sounds, etc. are related to the home page, rather than the particular conversation. In general, the use of links seems relatively rare.
• The Message Body. Messages posted in the cafe tend to be relatively short: a couple dozen to a couple hundred words. A facility enabling the conversant to "hide" a long posting by making it a separate document accessed by a link is rarely used, though sometimes needed. The body of the message has little structure: while the conversational contributions shown frequently begin with greetings, this is particular to the "introduce yourself" conversation, and is much less frequent in other Cafe Utne conversations.
• The Message Header. As in the example, each conversational contribution is begun with a header that includes name, user ID, time stamp, and length of comment. The contents of the header is under the control of the software, although the user may change their name when entering the general conference, in my experience this has been infrequent.
• The Sequential, Linear Structure of the Conversation. The messages posted to a conversation are presented as a single sequence of contributions, all within the same window. Messages frequently refer and respond to the previous message; reference is usually an integral part of the message content, as opposed to explicitly quoting parts of a previous message as is done in email.

3.5.4. Regularities of Substance

Aside from the fact that Cafe Utne conversations generally adhere to the topic of conversation (though there may be considerable topic drift over time), I've observed a number of regularities in the substance of the Info Age "Introduce Yourself" conversation.

New participants often make moves to control expectations of their performance. One type of move is the declaration of inexperience. Newcomers, to the conversation, the Cafe, or to on-line discourse in general, often begin by stating their inexperience ("I'm a new Cafe lurker..." comment 61). Another move is the participation hedge, in which participants a try to downplay expectations about the degree to which they'll participate: "I tend to be a bit quiet, so I'll probably only pop into the discussion every once in a while."

These moves are met by regular participants, who respond with their own moves of welcoming and encouragement. Thus, in comment 60, we have "Welcome aboard, Kitty!", expressed by a regular participant, later followed by "But feel right at home and jump to any topic you like. It's a pretty friendly environment."

It is also common for participants to respond positively to posts that are particularly thoughtful or self-revealing with affirmations or thanks. These posts usually directly follow the target message.

Finally, jokes and word play are not uncommon, and make the conversation more pleasant and engaging, as well as decreasing its formality. These are typically very short comments that play on immediately preceding comments. Thus comment 2.60 ("Programming mainframes. That sounds serious... like housetraining dinosaurs.") plays off 2.59, as does 2.64 on 2.63.

While none of these conversational moves--except for welcoming--are directly connected to the conversation's topic, they are important for encouraging participation, and in compensating for the lack of direct social pressure implicit in face to face conversational situations.

4. Designing Participatory Genre

In this section I discuss ways in which the regularities of form and substance of the Cafe Utne conversations arise out of, or are facilitated by, the nature of the discourse medium. By discourse medium, I mean to focus on the constraints and affordances of the technology which supports the conversation (I prefer this term to"user interface" because interface is often understood as referring to icons and menus, and because it does not include the content). I am not so much concerned with the physical level aspects of the technology--although the requirement to access and use keyboards, computers, displays, phone lines, and modems imposes a variety of physical, social, and financial constraints--as with the functional nature of the conversational system. The goal is to gain insight on how to design discourse media to support on-line discourse.

4.1. Properties of the Discourse Medium

Let's begin by looking at the properties of the discourse medium used by Cafe Utne. As evident in the discussion of regularities of form, the Cafe Utne discourse medium primarily supports text: only text may be entered, though since URLs may be entered, links can be created to other media types. The discourse medium also provides structured textual headers for each message.

Another property of the discourse medium is that it is persistent. Unlike synchronous chat systems, where a conversational contribution is seen only participants who are on-line at the moment it made, contributions are preserved (in an easily accessibly form) indefinitely, and so conversations can (and normally do) occur among people who are not connected at the same time.

Perhaps the most important aspect of the discourse medium is that it uses a conversation-as-document model. That is, in contrast to mailing lists and certain bulletin board systems, in which each contribution to a conversation is a separate document that can (and must) be opened and closed independently, the Cafe Utne conversational system treats each conversation as a single entity which is opened or closed as a whole. There are three important properties of the discourse medium spring from this:

• all participants see the same thing
• newcomers to a conversation see the entire conversation before reaching a point where they can contribute
• sequentiality is preserved (that is, if I add a comment, it will appear following the last comment I see on the page)

(To be strictly accurate, none of these last three properties is actually guaranteed . Due to the mechanics of web-page updating, sequentiality will occasionally not be preserved, and all participants may not see the same thing, at least simultaneously. However, this happens only in the case when the on-going conversational turns are close together in real time; so far, most conversations appear to have stayed asynchronous. More arguable is the claim that newcomers to a conversation see the whole thing. While this is possible, and while it's the easiest thing to do, it is possible to request that the system only display comments that were entered since a particular date. Nevertheless, I shall assume for the purposes of this paper that newcomers generally read at least a portion of the preceding conversation. My only evidence for this, besides frequent references to preceding messages, are occasional apologies from new participants about not having read the entire conversation.)

4.2. How the Discourse Medium Supports Participation

4.2.1. Providing an Overview
Because a conversation is represented as a single, persistent document with the oldest items first, newcomers can quickly get an overview of norms that govern the conversation by skimming through it. Similarly, infrequent participants can refresh their memories about what has transpired. Consider what can be gleaned from skimming through the conversation "document":

• Length. A newcomer to the conversation can very quickly see--by scanning down the conversation page--various structural aspects conversation: most obviously, it is evident how long conversational contributions are.
• Conversational Rhythm. Participants in the conversation can get a feel for the tempo of the conversation by inspecting the time stamps. If they post a question, they can form an estimate of how quickly they're likely to see a response.
• Content. As the reader skims down the conversation, a quick idea of the content can be gathered as well. Comments of special interest are often identifiable because they may be followed by several, brief, laudatory comments.
• Participants. Inspecting the names and user IDs in the headers reveals who the regular participants are, and how often they participate. It gives a quick idea of how large the conversation is, and whether there are steady participants or a series of transient contributors.

4.2.2. Supporting Form
Note that the ability of the conversational document to serve as an overview is supported by the discourse mediums provisions of automatic headers with a regular structure and format. This regularity is what makes it easy to skim: the reader can choose which parts of the header to focus on, because of their spatial regularity, and the conversational turns are clearly separated from one another.

4.2.3 Facilitating Substance
It's interesting to note that the discourse medium can also support the content of the conversation. Not that it directly shapes the content, but rather in that it makes certain types of conversational moves easy or difficult. For example, the type of word play observed in Cafe Utne conversations is rare in the medium of mailing lists. In a mailing list, to play off a preceding remark, requires quoting it and playing off the quote, a more cumbersome move. Similarly, for the reader of a mailing list, having to open a separate message is a bit more effort, and the one-liner which could have been easily skimmed or skipped in the context of a conversation document is more likely to be annoying than amusing. Similar arguments hold for other sorts of short responses such as affirmations and thanks.

It is important to recognize that no discourse medium is entirely supportive of participation. Rather, any medium facilitates certain conversational moves and inhibits others. For example, because this discourse medium supports a sequential, linear conversation, it is likely to inhibit responses to posts that are further back in the conversational sequence. In contrast, a mailing list, for example, makes it easy to respond to any contribution to a conversation. Thus, as in any other area of design, discourse media embody tradeoffs between alternatives.

4.3. Extending Participatory Discourse Media

Having carried our analysis this far, it is tempting to speculate on ways in which the discourse medium might better support on-line discourse.

One approach is to look at the regularities of form and substance that facilitate participation, and design the discourse medium so as to support them more effectively. Thus, a designer intent on improving Cafe Utne, might examine ways of providing better conversational overviews. Another avenue would be to explore ways of opening up the conversation, so that new contributions were not primarily responses to only the few most recent responses. While a general solution to this issue is undoubtedly complex, simply providing a way for conversants to attach one-line affirmations or thanks to any contribution, might markedly change the feeling of the conversation, as well as giving inexperienced participants a very easy and low profile way to participate.

A complementary approach is to look at the communicative pressures inherent in the genre, and look for other ways to respond to them. For example, early in our analysis of Cafe Utne, we noted the absence of social pressures for participation that are found in face to face participation. It would be interesting to explore ways to bring these social pressures back into play. One way would be to represent the presence of both readers and contributors in the discourse medium, so that participants who were on-line at the same time would have some indication of the others presence, as is currently done in other discourse media like MUDS. Another approach would be to investigate ways of portraying the audience, those who are reading the conversations without contributing. Knowing that a contribution would be read by very many, or only a few, would doubtless affect potential contributors in complex and varied ways. Along with such a tack, it might be advisable to create limited audience conversations, for those desiring to talk with only a small number of people.

Clearly, genre analysis doesn't provide easy answers here. There is considerable work for the designer. However, by highlighting the communicative purposes of the genre, and the way in which the discourse medium facilitates or inhibits the achievement of those purposes, genre analysis does offer an approach that can be applied to a wide variety of systems that support on-line discourse.

5. Conclusions

As a designer, my ultimate goal is to design more effective on-line systems. Genre appears to be a useful frame for analyzing and designing on-line conversational systems because it allows the foregrounding of the discourse medium, and encourages the examination of ways in which characteristics of the medium shape the practices which are conducted within it.

In this paper I've looked at one example of how properties of a discourse medium are used to shape conversational interactions. In this particular case, it's interesting to note the importance of the discourse medium's sequentiality. Obviously, designers would do well to examine other discourse media, and arrive at a better understanding of how these media can best support the establishment, maintenance, and negotiation of the basic expectations underlying conversations.

While this paper has focused on the properties of the discourse medium for shaping participatory genre, it is important not to neglect other forces which shape participatory genre. The very features of the discourse medium that support word play and joking, could also effectively support sarcasm, parody, and other negative forms of response. Without a doubt, social factors--such as the nature of the discourse community--and institutional factors--such as the policies carried out by the Cafe's management--are critical in the creation of a participatory, convivial site for on-line discourse.

6. Acknowledgments

Thanks to Laura Gurak, Carolyn Miller, and Chuck Bazerman for pointers into the literature on genre theory. Thanks to Griff Wigley of The Utne Reader for discussions of Cafe Utne and its administration. And thanks to the (disguised) participants in the example conversation for permission to use their words.

"The Palace" is a trademark of Time-Warner Communications. "Ferndale" is a trademark of Songline Studios, Inc.

7. References

[1] Baym, N. K. The Emergence of Community in Computer-Mediated Communication. In Cybersociety: Computer-Mediated Communication and Community (ed. S. G. Jones). Thousand Oaks, CA: Sage Publications, 1995.

[2] Bazerman, C. Genre and Social Science: Renewing Hopes of Wingspread. Paper presented at the Rhetoric Society meeting, Tucson, May 1996.

[3] Bruckman, A & Resnick, M. Virtual Professional Community: Results from the MediaMOO Project. Proceedings of the Third International Conference on Cyberspace , Austin, Texas, March 1993.

[4] Cafe Utne. www.utne.com. (Site for the Cafe Entrance and The Utne Lens on-line publication.)

[5] Cohen, A.P. The Symbolic Construction of Community. London: Routledge, 1985,

[6] Curtis, P. "Mudding: Social Phenomena in Text-Based Virtual Realities." Proceedings of DIAC '92 .

[7] Ferndale. Songline Studios, Inc. http://www.ferndale.com.

[8] Fowler, R. B. Community: "Reflections on Definition." New Communitarian Thinking: Persons, Virtues, Institutions, and Communities (ed. A. Etzioni). Charlottesville: University Press of Virginia, 1995.

[9] Hillery, G. A. Definitions of Community: Areas of Agreement. Rural Sociology, 20, 111-123. 1955.

[10] Miller, C. "Genre as Social Action." Quarterly Journal of Speech , vol. 70, pp. 151-67. 1984. (Reprinted in Genre and the New Rhetoric (eds. A. Freedman & P Medway) London: Taylor and Francis, 1994.

[11] Motet. http://www.sonic.net/~foggy/motet/.

[12] The Palace. Time Warner Communications. http://www .thepalace.com/

[13] Reid, E. "Electropolis: Communications and Community on Internet Relay Chat." Honors Thesis, Department of History, University of Melbourne, 1991.

[14] Rheingold, H. The Virtual Community: Homesteading on the Electronic Frontier. Reading, MA: Addison-Wesley, 1993.

[15] Rheingold, H. "A Slice of Life in My Virtual Community." In Global Networks: Computers and International Communication (ed. L. M. Harasim). Cambridge, MA: MIT Press, 1994.

[16] Selznick, P. The Moral Commonwealth: Social Theory and Promise of Community. Berkeley: The University of California Press, 1992

[17] Smith, M. "Voices from the WELL: The Logic of the Virtual Commons." Master's thesis, Department of Sociology, UCLA, 1992.

[18] Swales, J. Genre Analysis . Cambridge: Cambridge University Press, 1990.

[19] The Utne Buzz, September, 1996. (A monthly electronic newsletter for members of the Utne Cafe.)

[20] Yates, J. and Orlikowski, W. J. "Genres of Organizational Communication: A Structurational Approach to Studying Communication and Media." Academy of Management Science Review, Vol. 17, No. 2, 299-326. 1992.