Less is more

the benefits of low bandwidth in virtual communities

by John Unsworth

Part of "Scholarly Communication on the Internet: A Retrospective Look at H-Net on its Tenth Anniversary" Hilton, Continental Room C, Chicago, IL, January 3, 2003. Sponsored by the AHA Research Division, H-Net: Humanities and Social Sciences OnLine, and the National Endowment for the Humanities

As my title indicates, my topic today is the benefits of low bandwidth in virtual communities, as a way of understanding the past success and future prospects of H-Net, in whose honor this session is organized. Before I do that, though, I want to provide some historical context for the activities of H-Net, and some personal context for my remarks.

H-Net’s historical context:

(from Hobbes' Internet Timeline, http://www.zakon.org/robert/internet/timeline/)

·                     1970: First cross-country network link established (56K, between UCLA and BBN)

·                     1972 (20 years before H-Net): The first machine-to-machine email message.

·                     1979: First MUD; USENET established

·                     1982: first definitions of an "internet" as a connected set of networks, specifically those using TCP/IP

·                     1983: First desktop workstations

·                     1984: Number of hosts breaks 1,000

·                     1987: Number of hosts breaks 10,000

·                     1988: IRC

·                     1989: Number of hosts breaks 100,000

·                     1991: Gopher, WWW both released

·                     1992: number of hosts breaks 1,000,000; first MBONE (multicast audio and video

That brings us up to the year of H-Net's birth. Over the decade intervening between then and now, the most significant events have to do with higher-bandwidth uses of the internet, for example:

1993: Mosaic, the first graphical web browser. WWW proliferates at a 341,634% annual growth rate of service traffic. Gopher's growth is 997%.

1995: banner ads, Java, Real Audio, introduced; AOL, Compuserve begin to provide internet access; the internet is commercial--no longer government-sponsored; 10,000,000 hosts

1998: XML

2000: Napster suits; 100,000,000 hosts

2001: Debut of the first live distributed musical -- The Technophobe & The Madman -- over Internet2 networks; First uncompressed real-time gigabit HDTV transmission across a wide-area IP network takes place on Internet2

Personal context:

Although I've been an H-Net subscriber, I'm not active in the organization; I think Mark asked me to speak on this occasion, in spite of that fact, because I've watched H-Net grow up while I've been working in the same area--electronic scholarly communication--for the past ten years.

My first project in this area was Postmodern Culture--the first peer-reviewed electronic scholarly journal. I started PMC in 1990 with a couple of colleagues, and my motivation was not unlike what you'll hear from an H-Net subscriber a little later on--I was newly appointed to a department where I was the only person teaching contemporary literature, and I was trying to establish an intellectual and professional community. We actually didn't intend for PMC to be an electronic journal, but the state was in the middle of a budget crisis, and a print journal was out of the question. We were fortunate to be at a university with a very engaged library and strong campus computing: we went to the library for help, they put us in touch with campus computing, and they said "there's this new software called Listserv, and we think it might be interesting to use it for this." So we did--and when we sent out our first issue, in September of 1990, all 300 pages or so, we had our first lesson in limitations: we crashed the mailboxes of nearly every one of our subscribers, since one issue of the journal (even in plain ascii text) exceeded the average mailbox quota. In the second issue, we switched to just sending the table of contents, and letting people request individual articles, and shortly thereafter (1991, I think) we started distributing the journal by ftp and gopher as well as email, letting people retrieve some or all of the issue or (with gopher) browse the articles online. We also ran a non-peer-reviewed discussion list, called PMC-Talk, for the first few years, and I'll have a little more to say about that later on.

In those early days, we made some critical decisions--decisions that, in retrospect, were consonant with the spirit of H-Net: we decided for access over presentation, foregoing formatting (even underlining, italics, etc.) in favor of plain ascii, which everyone (Mac, PC, Unix) could read, and which we could easily and dependably distribute by email.  When the Web came along, PMC was an early adopter, publishing its first web issue in January of 1994 (vol. 4, no. 2), and converting all of the back issues to html at that point. To this day, though, we continue to distribute tables of contents by email, and we continue to produce a text-only version of everything the journal publishes (for free, although the journal is also part of Project Muse, which licenses the whole collection in html, with multimedia elements).

PMC was also an early experimenter with MUDs (multi-user dungeons) and MOOs (a nested acronymn, standing for MUD, Object-Oriented). PMC-MOO (which still runs today, on an ancient unix server in a corner of my office) was a very active virtual community from about 1992 to about 1996. MOOs are an interesting parallel to email, where the subject of low bandwidth is concerned, and they also provide some illuminating contrasts. Like email, MOOs are text-only (at least, in their original, purest form), and therefore pretty low-bandwidth. Unlike email, though, they provide real-time interaction, which means they are not quite so fault-tolerant as email: if the network is lagging, you may be struck dumb in midsentence.

About the same time I was getting involved in MOO programming, in 1992 (the year of H-Net's birth), Neal Stephenson published a novel called Snow Crash, in which he imagined a future internet where people and information mingled in real-time virtual reality. In this world, called "the Metaverse," there are visible distinctions between rich and poor, as there are in our world--but in the Metaverse, the difference is a difference in bandwidth, and therefore, in resolution. In the world of Snow Crash, the disadvantaged are

black-and-white people--persons who are accessing the Metaverse through cheap public terminals, and who are rendered in jerky, grainy black and white....Talking to a black-and-white on the Street is like talking to a person who has his face stuck in a xerox machine, repeatedly pounding the copy button, while you stand at the output tray pulling the sheets out one at a time and looking at them.

Of course, the problem Stephenson describes here wouldn't exist if real-time fully rendered virtual presence weren't the norm for participation in this networked community--if, say, the community were based on email rather than on avatars.

When I moved to the University of Virginia to run the Institute for Advanced Technology in the Humanities, in 1993, I took with me the lessons I had learned from PMC, and chief among those was the importance of communication in scholarship: it was so important, in fact, that the ability to communicate quickly, cheaply, and interactively was, I believed, worth sacrificing most other things to obtain. The Web was a great way to disseminate research results, and it offered staggering opportunities for publishing text with images, publishing out of databases, building complex hypertext and multimedia collections, etc.. Still, project management required communication among individuals, often across time zones. In the Blake Archive, for example, editors were in New York, North Carolina, and California, with project manager and IATH staff in Virginia. Initially, I thought we might use a MOO for this, and I even put the Blake editors through a training session in MOO use. However, somewhat to my surprise, the MOO was a failure--never caught on as a channel for collaboration among these editors. Instead, they gravitated to email, and over the past eight years, they have generated thousands of mail messages, sometimes at the rate of dozens a day.

What's so special about email?

Why does email work so well for these scholars? Time-shifting is one important part of the answer: real-time communication (like MOOs, or conference calls, for that matter) require all participants to be available simultaneously--and that can present a considerable logistical challenge, especially in international collaborations (for example, I recently organized a conference call that took place at 5 am on the west coast, 8 am on the east coast, 1 pm in England, Norway, and Germany, 2 pm in Spain and Croatia, and 10 pm in Japan). With email, people can participate when it suits them. Also, it's easy to automatically log the exchanges, turn them into html, and make them searchable and browsable on the web, for future reference. And although email may qualify as a "hot" medium, by comparison to the Web, it is a much cooler one than MOOs, IRC, or other text-based realtime communication technologies, and much less noisy.

The two killer apps of the internet have been, and continue to be, email and the world-wide web. Both fundamentally depend on the existence and extent of the network, but so do a lot of other applications. Why did these two succeed?  Two answers, I think: they are each very good at supporting a very general-purpose activity, and neither requires real-time participation.  Email is, I think, still the best form of reciprocal human-to-human communication: It is almost instantaneous and almost free, and it doesn't require the recipient to be on hand to receive the message, or to respond immediately. It is ascii-based, so it crosses platforms with little difficulty. All of these things made it an instant success--it constituted 75% of network traffic one year after it was introduced--and email continues to demonstrate its vitality as a medium, for example as one of the most successful applications on new handheld devices (like your cellphone, PDA, or blackberry).

By contrast, the Web is not reciprocal human-to-human communication. It is more like a publishing medium, and its strengths are hypertext linking, cross-platform network accessibility, and the incorporation of images (it's hard now to recapture the striking advance that seemed to represent in 1993). Oddly enough, though, for all the wonders of the Web, email is still a better means of creating virtual community, because it is reciprocal. MOOs, IRC, blogging--these are all reciprocal communication media, and they have all been successful means of creating virtual communities--but none more so than email.

Email's low bandwidth has made it ubiquitous, but it has also made it the beach-head application, first to reach the outer limits of the network, usually the first application experienced by new adopters, and generally our last link to our networked lives when we're on the road, dialing in from hotel rooms.

Low-bandwidth, high-impact virtual communities

Email discussion groups, like those at the heart of H-Net, have played an important part in academic life over the last decade--as Richard Jensen says in "Internet's Republic of Letters," "[a]nnual conferences and conventions [create a community of scholars] a couple days a year, rather like Brigadoon. H-Net communities live every day." And although we don't count participation in these groups for any kind of professional credit, the contribution to scholarship, and to academic community, that's made by through these email-based communities is very real. Jensen's 1997 essay goes on to say that

[t]he hundreds of pages of evaluation reports H-Net received from subscribers indicate [that] . . . . H-Net has sharply increased the quantity, quality and diversity of communications among historians, especially those who otherwise would be remote from the centers of scholarly activity. While professors at the major graduate universities are already well-connected, the H-Net lists provide a scholarly lifeline to independent scholars, librarians and archivists, public historians, graduate students, and teachers at smaller or less well-funded colleges, community colleges and universities, offering intellectual stimulation and a chance to maintain a dialogue with leading scholars in their fields of interest."
--from "Internet's Republic of Letters: H-Net for Scholars," by Richard Jensen

Today, in 2002, H-Net Discussion networks comprise "over 100 E-mail lists spanning various fields of study" from American Religious History to the History of Childhood and Youth to Lusophone African Studies to Labor Studies, and many others. H-Net itself, of course, also includes syllabi and other teaching resources, a job guide, announcements, and reviews.

So far, I've been arguing that less is more, that email--the oldest, lowest bandwidth application on the internet--has a special virtue where virtual communities are concerned, because it is broadly accessible, requires minimal connectivity and nothing more than the most basic end-user hardware and software, and because it is reciprocal, as well. But to create and sustain a virtual community requires more than email--it requires participation, direction, and structure. In fact, virtual communities are very like real ones in their social structures and their organizational requirements. For any one of the H-Net lists to succeed requires dedicated involvement from a number of list-managers as well as from key members of the community (two categories that may obviously overlap). Even so, the truth about virtual communities is that they have life-spans; they come and go; they live and die. There are exceptions, of course: Humanist, Willard McCarty's humanities computing email discussion group, has been around since the late 1980s. But PMC-Talk is a more typical example: it grew from a few members to a reasonably large and active group, continued successfully for a couple of years, and then died. There's a reason for this pattern: virtual communities are communities of interest, and interests shift, definitions and terms change, and conversations finish. Sometimes, as in the case of Humanist, if the topic is broad enough, and if there are one or two people making sure to stir the pot at regular intervals, then you may see a longer life-span. But the weakeness of PMC-Talk, and even of Humanist, is that these are one-offs, stand-alone endeavors. One of the great strengths of H-Net is that it consists of many such communities, not one, and when one of its constituent communities dies out, there is a framework that makes it easy for another to spring up to replace it.

Behind that one word, "framework," are some carefully crafted documents--the bylaws and constitution of H-Net--which are the real key to H-Net's longevity and success, in my view, as well as being the objective evidence that a virtual community has actually come into being here. It may strike some of you as depressing to think that bylaws are evidence of community, but they are--if you look at the H-Net bylaws, you can almost hear the echoes of the community crises that brought them into being. For example, consider Section 2.02, Dispute resolution procedure:

Academic debate and discourse inevitably, and usually constructively, invite controversy. The editors of H-Net's networks normally apply common, if evolving, professional standards in judging the value of content for publication. All network welcome messages will indicate any departure or special application of these standards to the network's mission. Those standards entail the balancing of intellectual freedom with the need for civility and restraint. Messages that in the judgment of the editors harass, defame, slander, or libel others, that abuse intellectual property rights, are subject to rejection or revision to remove the offending passages.

This stuff is important: PMC-MOO, for example, disintegrated for lack of a well documented dispute resolution procedure, as you may see from reading "Fuck Art, Let's Kill!: Towards a Post Modern Community, or Terrorism in Cyberspace," a retrospective screed described by its compiler as an "authorless/Subjectless text recounting the deconstructive campaigns of the virtually important (and infamous) Post Modern Culture MOO Terrorists. By exploiting the tensions of community, and questioning the inherent assumptions of MOO Space, the PMC Terrorists illuminate many of the ironic ambiguities of virtual existence." (see ftp://jefferson.village.virginia.edu/pub/pubs/pmc/pmc-moo/MOO.Terrorism)

H-NET has had its own brush with this sort of thing, in the January 20, 1995 issue of the Chronicle of Higher Education, where Jesse Lemisch took it to task for its requirement that participants practice civil discourse. Lemisch's argument was that the "ridicule of ridiculous ideas . . . [is] a legitimate debating strategy," and that flaming was a legitimate form of dissent (http://www.usao.edu/~facshaferi/CYBERWAR.HTML).

My point is that in order for a community to persist over time, and even in order for it to function effectively as a community in the short run, it needs to have ground rules, and these need to be clearly stated, and well thought-out.

Spam, spam, spam, spam...

H-Net has done well to survive this long, and to evolve the policies and procedures that make it a productive part of academic life for so many of its participants. But lest I leave this panel on too cheerful a note, there is a rising threat to H-Net's bucolic, low-bandwidth, email-based virtual communities, and its name is spam. According to a recent article in the cyberatlas at internet.com (http://cyberatlas.internet.com/big_picture/applications/article/0,,1301_1555831,00.html),

"More than 30 percent of all e-mail is unsolicited and MessageLabs [a Managed Service Provider (MSP) specializing in e-mail security] predicts that spam will continue its exponential growth into 2003, surpassing the amount of non-spam e-mail by around July."

That's right--by July 2003, there will be more spam than legitimate email banging around the net. That same article cites "a survey of 1,000 consumers conducted for Symantec Corp." that found

37 percent of respondents receive more than 100 spam e-mails each week at home and work;

63 percent receive more than 50 spam messages weekly at home and work

And spam aside, there may just be too much email:

"In July 2001, Ferris Research predicted that 2001 would see a 50 percent rise in the number of emails business users would receive, with further growth of between 35-50 percent during 2002 .... In September 2001, IDC predicted that there will be 1.2 billion email mailboxes by 2005, up from 505 million in 2000. It also predicted that by 2005 there will be 36 billion person-to-person emails sent worldwide every day." (From Gerry McGovern, "Email: too much of a good thing?" New Thinking, November 05, 2001: http://www.gerrymcgovern.com/nt/2001/nt_2001_11_05_email.htm)

Even discounting the 18 billion of those email messages that will be advertising human growth hormone, herbal viagra, and low mortgate rates, that's still an awful lot of email--a dozen or so legitimate email messages for every one of those 1.2 billion email mailboxes. According to my own email program, I receive an average of 69 emails a day (and that's after vigorous spam filtering), and I send an average of 21 messages a day. That's today, in 2002. I confess that I spend so much time doing person-to-person communication, and just plain business, through email, that I rarely contribute to email discussions any more. The biggest threat to H-Net may be the success of the very medium on which its core functions depend--good old email.

On the other hand, a more optimistic forecast might predict that this very glut of email will make filtered, focused, high quality email discussions like those on H-Net all the more valuable: I hope that's how it turns out, because we do need H-Net's virtual communities, and they do enrich our academic conversations, our teaching, and our research.