Cite as: 10 Harv. J.L. & Tech. 495 (Summer 1997)

The Rise and Fall of Sysopdom


Jonathan Zittrain*


I. Introduction


"Sysop" has gone from a term of art known only to the bleeding-edge few to a dusty anachronism known only to the bleeding-gums few, without the usual years-long general linguistic acceptance and respect in between. In case the reader is not among the bleeders: sysops (from "system operators") run electronic areas accessible by typing furiously on one’s networked computer, through which one can meet, talk to (well, at least type at), and develop nuanced social relationships with other people similarly typing and reading. Few know what a sysop is because these electronic areas — aspirationally, and sometimes accurately, known as "online communities" — have never quite flourished and today are in decline.

Indeed, "online community" joins "sysop" in the oversize dustbin of trite or hopelessly esoteric, hence generally meaningless, cyberspace vernacular. Not that "online community" is obscure, like "sysop"; rather, the term’s emptiness results from its abuse. "Online community" is used by Internet companies the way a motivational speaker uses "excellence," an academic uses "new paradigm," or a lawyer uses "justice": it represents something once craved and still invoked (if only as a linguistic placeholder) even as it is believed by all but the most naïve to be laughably beyond reach. Since it’s applied to almost anything, it now means vague warm fuzzies and nothing more.

The craft of sysoping and the phenomenon of online community (non-hollowly defined) have gone down together even as the Internet has burgeoned, and I want to explain what has happened to sysops as a way of explaining what has happened to the truly great and transformative promise of online communities. Law has played a major role in two distinct ways. First, sysops and the members of the communities they lead have struggled through lawlike reflection to arrive at just solutions to the disputes that inevitably arise in the course of their interactions. This struggle is a large part of what has made the communities so interesting. Second, fear of the formalistic application of the machinery of the real-world legal system is threatening to drive the amateur sysop to extinction and thereby to destroy what’s left of online community.


II. Beyond Sending Notes, Pointing, and Clicking


Today, going online generally connotes some mix of two activities: exchanging e-mail and surfing the World Wide Web. However, there’s nothing about the architecture of the Internet mandating that e-mail or web surfing be the most popular or available pastimes, any more than the nature of an electrical outlet determines that it be used primarily for hair dryers and refrigerators, or the nature of a road requires that it be used just for trips to friends’ houses and the supermarket. E-mail and web surfing aren’t the only ways to use the Net, and as it turns out they aren’t uses that foster online communities. To describe online communities thus requires showing why their mechanics aren’t much like what most everyone is now used to online.

E-mail is primarily one-to-one: one finds a friend’s e-mail address and begins a correspondence. A company finds a hapless consumer’s e-mail address and sends an offer to participate in a multi-level blue-green algae distribution scheme. To be sure, that company could be sending a similar offer to thousands or millions of other e-mail users — but the recipient wouldn’t know it, nor would it particularly affect the content of the communication if he or she did. The one-to-one nature of e-mail makes it difficult for a community to be built upon it, just as it would be tough to build a community exclusively out of a series of one-to-one telephone calls.

The Web is primarily one-to-many. The one can be an individual hosting a home page, typically presenting a fairly static set of facts (or lies) about some subject of interest to the host (e.g., "All About Me!"), like a carrel at an expo or convention. The one can also be a corporation hosting a site, typically a place where facts (or lies) about the company’s products can be gleaned and one can perhaps place orders for those products. With few exceptions, someone visiting a home page or a corporate site has no idea how many other people are visiting at the same moment, and no way to communicate with any other visitors. It’s like being in a movie theater without a view of the other seats. As the Internet’s bandwidth increases, the text and pictures of today’s static web sites are giving way to moving pictures and live sound — but they’re still one-to-many, enforcing a distinct separation between constellations of increasingly professional producers ("content providers") and masses of silent, shuffling consumers who register their presence only by the fact of a turnstile-like "hit" upon each web page they visit, the purchase they may make, or the occasional one-paragraph feedback form they deign to fill out.

Online communities, like their real-world counterparts and unlike most relationships based on e-mail and the Web, are built upon many-to-many communication. There is a public component to them, an unusual sense of getting upon a stage where one is both speaker and audience, where no special preparation is needed to be a producer as well as a consumer.

For example, an early Internet application now known as "chat" allows the many to type at the many, a soundless telephone party line with strangers constantly picking up and dropping additional extensions. Chat addicts — and their cousins who use the more sophisticated software of MUDs1 — have been known to describe their electronic realms as communities, and they might be right. If one chats enough in the same venue, the same people can be encountered again and again, and long-term interactions can develop in the way that regularly hanging around the same convenience store at the same hours can spark new friendships, or at least make the other loiterers seem familiar. Like hanging around, chat is live by nature — one meets other participants only as they are participating at the same moment in time. This creates three nontrivial hurdles for community-building: First, interactions are limited by one’s typing speed. Having a conversation with a slowpoke or even a normalpoke can be akin to watching paint dry; there’s just not enough payoff to a thirty-second wait when the result is a single sentence, however carefully crafted. (Experienced chatters seem to get around this problem by participating in several chats simultaneously, or by undertaking other computer-based work while chatting; one’s attention is thus otherwise occupied while others plod away at typing.) Second, the conversational exchanges are burst-like, disjointed. This problem is in a way the opposite of the first: too many active participants in a chat room (in practice about eight or more) can make it difficult to follow any one thread of conversation. Third, what one says is in any event ephemeral, appreciated only by those present, and discussions evolve and devolve in real time — spend some time away from the screen, and the thread is lost.

Another Internet application loosely known as "newsgroups" is, like chat, many-to-many, but without chat’s live character and its attendant drawbacks. Within a newsgroup one composes and then posts messages that look a lot like e-mail. These messages can, however, be addressed to all comers, known and unknown. To a newsgroup about dogs, for example, one might post a question seeking advice on various breeds — and receive answers from others who read the same group and have some thoughts to offer on the question. Those answers, while earmarked for the original seeker of knowledge, can also, in turn, be posted publicly to the newsgroup. Others browsing the group can then see both question and answer, jumping in to offer their own questions, clarifications, or contrary views. In this manner a conversation unfolds over the course of hours, days, or weeks, with those who started it perhaps not even participating later on. The conversational coherence stays intact because the original question remains visible to all, along with the follow-up answers and discussion.

This newsgroup-style architecture enables something extraordinary: useful and satisfying contact between people who are utter strangers. In the physical world we typically encounter strangers in situations in which our roles and scripts are highly defined and the scope of our communication with them is duly limited. Deeper relationships develop slowly, fitfully, and rarely in a group. The few real-world analogies to the type of interaction possible within newsgroups and their kin — clubs, churches, neighborhood cafes — are, sadly, in decline.2


III. From Friendly Gatherings to Restless Mobs to Friendly Gatherings: The Growth, Trashing, and Revitalization of Online Communities


A. Friendly Gatherings


Newsgroups enable a physical distance that suggests safety, a self-selection of topic that entails some shared interest among those who participate, and a mode of communication that mates the relative permanence and subtlety of the carefully written word with the dynamism of constantly interacting with people. Their use tends to create horizontal relationships among individuals (dog lover meets dog lover) instead of vertical relationships between information consumer and producer (dog lover reads book about dogs, or browses Alpo’s web site). Further, in the early days, newsgroup participants brought similar backgrounds and norms to the table. After all, regardless of a newsgroup’s topic, early participants were, naturally, computer experts with technical talent and equipment that was still beyond that of the average computer owner, much less anyone else. In much the way that amateur radio enthusiasts of an earlier era already had some shared background — ham radio’s legal participants each had passed government-sponsored tests of varying difficulty in applied physics — most early newsgroup participants were cut from similar cultural cloth, and it was a culture of pioneering and experimentation that took readily to relationships based on lending a helping hand to a passerby.

In these early days newsgroups simply ran themselves.3 Some people, through lucid and frequent contributions to the discussions, would emerge as leaders of particular groups. There weren’t elections, and the leaders had no budget to control or soldiers to command. But it was clear that the leaders wielded authority: their counsel was sought, their views respected, and others’ behavior was self-modified to avoid their criticisms. Under such guidance, or merely through the feedback loop of path dependency, a group could develop its own implicit manifesto, both substantively and procedurally. For example, the newsgroup neutrally designated for discussions of Scientology evolved into a forum for those opposing Scientology; contrary views were sufficiently ridiculed and similar views lauded that most dissenters took their conversations elsewhere. In other groups there wasn’t so much a party line on substance as there was on procedure — those who raised their electronic voices a little too sharply were quickly and embarrassingly called to task. Recalcitrantly bothersome members found their posts without audiences as other members configured their software to filter out messages from those sources; without an audience the "troublemakers" as often as not would move on to another group or beg forgiveness from the existing one. Social forces functioned brilliantly to maintain order so long as participants craved each other’s respect.

B. Restless Mobs


Long was not forever. As mere VCR owners became Internet users (and were about as handy with the new technology as they were with the old), newsgroup participants found themselves beset by (1) lots of (2) people who had no particular sense of awe or wonder about a community they might encounter online. They often had no diffidence about entering and then rapidly hijacking a conversation that had its own distinct lilt and cadence born of its group’s culture. It was as if ham radio enthusiasts had awakened to the nightmare of their rarefied bandwidth summarily thrown open to the more plebeian CB crowd (the latter-day one, no less, bred on the brief Hollywood fascination with "smokeys" and "10-4s").4 The sheer din of new messages from so-called "newbies" made it difficult for old-timers to see and respond to the messages of familiar friends. Further, many new participants had no particular obsession with earning the respect of the venerable erstwhile leaders of the group. It was as if a relatively stable country were suddenly flooded with unchecked immigration. In response, xenophobia became the order of the day as old salts tried to ostracize and ridicule new arrivals to the groups rather than absorb and socialize them.

To make matters worse for the old-timers and the communities to which they belonged, commerce found the Internet. Given the current state of the Net it’s hard to believe, even as one who saw it firsthand, that the old Net enjoyed an obeyed if unenforced ban on commercial solicitations and advertisements. Those who strayed into mass advertising in newsgroups were dealt with mercilessly, and the strayers cared enough about the views of their peers (or, if one is more cynical, about the overstuffing of their own e-mail boxes with their peers’ angry complaints) to desist. Indeed, it was a seismic event when an otherwise-obscure law firm explicitly decided to start advertising within newsgroups, complaints be damned.5 Normative forces are simultaneously strong and vulnerable, perhaps unchallenged and dominant for years yet suited to near-instant unraveling in the presence of a few unchecked norm-flouters: once a few people cut in line or walk carelessly across the groomed Commons, a stampede is not far behind.6

Today’s newsgroups are post-stampede: topic names are primarily aspirational, conversations fleeting, and multilateral relationships non-existent as carpets of unrelated messages bomb groups indiscriminately, mutating like viruses to appear of interest and relevance until the moment their entreaties to MAKE MONEY FAST!!! or KILL ALL FORENERS!! [sic] are encountered. Someone who posts substantive messages despite that din now finds his or her "return address," often included so others can identify the author, added to hundreds of junk e-mail lists. It is a textbook fall from innocence: today’s bombed-out, graffiti-laden newsgroups stand as testaments to the archetypal progression from nothingness to order to chaos in the absence of real authority exercised — or perhaps merely revealed — with wisdom.


C. A Little Government Goes a Long Way


Enter sysops. Newsgroups happen to be wired to make the exercise of authority difficult: posted messages don’t repose in one place where everyone looks at them; rather, they’re automatically and ceaselessly shipped around the world to thousands of sites — servers — making it quite difficult to eliminate a message already posted. It is similarly difficult to prevent someone somehow adjudged to be obstreperous from posting to the groups — it’s too easy to post anonymously or with shifting identities, defying even the ad hoc filtering that readers of messages employ in an attempt to shun certain contributors’ messages. This distributed architecture isn’t necessary to public messaging, and there are other, more centralized architectures that lend themselves to some kind of control. Put the messages in one place, accessible only to those who’ve passed some kind of test — or at least not flunked it — and one has achieved a technical basis for de jure control to buttress the de facto control achieved through norms developed among people who cared about their standing in each other’s eyes. America Online ("AOL"), CompuServe, and even the mom-and-pop bulletin board systems operated off of individuals’ home computers offer such architectures for public messaging.

A typical CompuServe "forum," for example, has a topic — dogs and cats, or wine, or national issues, or Macintosh games. The only people able to access such a forum are CompuServe subscribers, people who have identified themselves to the company with a name, an address, a credit card. CompuServe, too, once had a quite natural filtering of forum members because, like all other telecommunications-based services, it required unusual equipment and above-average interest and skill to use. But it saw the effects of the great migration of the VCR-owning public to the Net just as the newsgroups did — in fact, much of the migration was made possible precisely because CompuServe, AOL, and their siblings made it easy for those in the technological mainstream to sign up for a networked account.7 CompuServe forums avoided the newsgroups’ fate because the forums have sysops. Sysops can, in effect, police their respective areas. If they see a public message they don’t like, they can erase it, or retitle it, or move it to subsections reserved for off-topic conversations. If they see subscribers they don’t like they can deny them access to their respective areas. Such "lockouts" can be effected even if the subscribers aren’t currently logged in; any further attempts to read or participate in public discussions in an area to which one has been denied access are met by form-letter rejection notices written by the area’s sysop.8

The means by which CompuServe and AOL choose sysops for their forums are obscure. Some sysops appear to be experts at their forums’ topics — the legal forum is run by a lawyer, the medical forum by a doctor. Others are not experts, but preside over discussions anyway. One pattern is clear: the sysops chosen by these large companies aren’t the companies’ employees. They are, almost without exception, independent contractors, scattered across the United States and the world, perhaps intentionally at arm’s length from their corporate hosts. Most are paid little and many work for free. Virtually all are drawn from the ranks of the paying forum membership. The forums they sysop (it’s also a verb) are not mere franchises, independently owned and operated, with each detail centrally prescribed. To a CompuServe subscriber it may look at first blush as if each topical forum is simply another standard CompuServe area; in reality, the sameness of the operating software — the technical means by which messages are posted and read — from forum to forum masks the fact that most are fiefdoms run by intense individuals without any particular direction from the company.

The policies these sysops create and impose vary wildly. The "rules" are functions of the respective topics covered by the forums, the experience level of the membership, and the individual quirks of the sysops. Some sysops are utterly hands-off — indeed, some haven’t been heard from in any way for months or years. The forums of these absentee landlords run on autopilot, and the content of such forums predictably devolves to that of the unmoderated newsgroups of the Internet: advertisements, squabbles, transients. Other sysops are hands-on philosopher kings and queens, attempting to lead by example (posting, say, plenty of helpful public messages answering topical questions posed by visitors) and using moral suasion to deal with recalcitrant troublemakers (defined solely in the eye of a sysop) in an ad hoc fashion. Still others have a legalesque method of wielding power, publishing contract-like lists of rules and policies, issuing formal "warnings" upon perceived breaches, and finally rendering punishment in the form of a lockout if a requisite number of warnings accumulates on one subscriber. Those sysops who do enforce their policies or whims soon learn the power and efficiency of threats over action, a mere hand on the holster, and some of the most controlled forums turn out to be ones in which the sysops never actually go so far as to invoke their formal powers.

What do forum policies tend to contain? Some sysops ban any form of profanity (yes, the ban usually covers the odd attempt to declaw an otherwise-objectionable word by substituting asterisks for vowels) and delete any message that contains it. Some ban "personal attacks," which, to those familiar with the concept of flaming, happen with astonishing frequency over virtually all of the Net’s many-to-many applications.9 Some ban commercial advertising, or require that it be limited to a particular subsection that members can easily filter out.

This description may make sysops sound like little Napoleons and Red Queens running around looking for wayward people to discipline for failing to conform their contributions to the requirements of whatever law-giving muse the sysop may draw upon. It’s partly true — and to be sure, some sysops are absolutely obsessed with and corrupted by their ability to govern others’ behavior. But in the context of a bazaar of disparate forums such as those offered within CompuServe or AOL, members oppressed by such governance in one forum can readily vote with their feet and choose to patronize another.10 Even those who stay might stay to complain. No one likes to lead a community that’s empty, or that comprises a few sullen and only semi-willing participants.

Thus those sysops sufficiently egotistical to abuse their essentially absolute power are also the ones sufficiently egotistical to require lots of affirmation in the form of a large and happy membership gleaned by giving the members what they want and refraining from giving them what they don’t want. Even if no higher law or authority were to bind sysops (certainly the CompuServes and AOLs that sponsor such forums have no interest in getting involved in forum governance disputes), the forces of norms and markets are strong, and in the long run elicit behaviors by sysops that tend to forestall the deterioration found in the unmoderated newsgroups without creating petty sysop tyrants.

The result is a cadre of sysops who specialize in encouraging useful discussion and contributing to it themselves: 24-hour-a-day talk show hosts who put their audiences so much at ease that the electronic exchange is driven almost entirely by the audience without spoon-feeding. An entire generation of sysops — thousands of them across the major information services — has developed this specialty, safeguarding quite remarkable online areas where one can ask almost any question or raise any issue and expect a series of insightful answers from strangers within hours or days. In these venues guards are down; the rotes of typical interactions seem stiff and confining in comparison. For example, an automobile forum somehow has car dealers confiding the tricks of their trade to wide-eyed consumers ("It’s true, I don’t really talk to anyone when I go back to ‘consult with my manager.’"); a travel forum might host flight attendants explaining how to get bumped into first class without losing one’s frequent flyer miles; and a handy forum attracts plumbers and electricians craving the opportunity to coach a homeowner whom, for all they know, by day they just snubbed when she phoned their offices for advice. Participants in these forums, often encouraged by the example of a sysop who seems to work only for the love of it, crave the opportunity to contribute and demonstrate their knowledge just as much as they seek to tap everyone else for advice.11

When intervention by a sysop takes place, it often takes place thoughtfully, with notions like due process or tolerance of distasteful speech generated on the fly or cobbled together from the Western cultural and legal landscape that still looms large on today’s Internet.12 As these communities have matured and grown the sysops have grown with them, fashioning new policies as old ones are tested and found wanting. For example, a sysop of a computer-related forum at one point agonized over whether to bind herself philosophically to the First Amendment. A member had posted several messages with anti-gay rhetoric, and the sysop’s original purist stance allowing all messages was withering away as others left the forum in disgust. Some fellow sysops, when asked for advice on the situation, hewed to the idea that even off-topic political speech shouldn’t be censored, no matter how offensive, while others saw a responsibility to build a community in which norms were backed up by rules. Whether the sysop’s chosen resolution was "right" — she ultimately insisted that the member stop making such posts in part because there were other forums in which to express such views, and the messages at issue had nothing to do with computers — it was arrived at thoughtfully, with an appreciation that the decision was not merely an administrative or ministerial choice, and without an appeal to corporate regulations or a passing of the buck that occurs so frequently in more "professionalized," bureaucratic environments.

Such thoughtful sysop leadership facilitates what I see as the transformative promise of an online community: a chance for individuals, in the real world increasingly slotted into specialized occupations with narrowly defined relationships, to find a manageable way to find, greet, and interact with other people along a variety of topics and in a variety of ways that, sadly, the physical world seems to make difficult or deny altogether. It is a far cry from a Net where the expectation is that the content will be produced by Lucasfilm or Disney — entertaining, perhaps, but about as interactive and relationship-forming as a garage door opener.


IV. From Friendly Gatherings Back to Just Sending Notes, Pointing, and Clicking: The Latest Threat to Online Communities


Lately the promise of online communities has dimmed. The essence of the strong but fragile open newsgroup communities that was saved by the moderated communities of forums on membership-based networks is increasingly threatened — by law. Sysops evolved to maintain some form of order in communities that were growing too quickly to allow innate neighborliness to govern. The Net grew yet further, though. It was as if a small hamlet had expanded first enough to require a citizen-mayor, and then into a megalopolis whose attendant civic risks were too much for that mayor to bear. A gnawing worry about personal financial liability for private torts transformed the easygoing hobbyist sysop into a risk calculator whose numbers demanded that he or she seek another, less legally unpredictable pursuit. For example, one sysop encountered a series of messages in his forum in which a new user sought advice on a particular software package. Another member of the forum recommended against purchase of the product because he’d had such a hard time with customer service; another darkly warned that the software company had engaged in an unspecified "dirty tricks" campaign a few years ago and that he no longer bought its products as a result. A representative of the company informed the sysop that the company was prepared to sue him over both messages unless they were deleted immediately, and even then was considering suing for whatever damage to the company’s reputation had already been done. This request was not merely an opportunity for self-reflection on the part of the sysop to decide the philosophical issue of whether it was generally fair to allow users to provide unsubstantiated negative views of software products. Instead the sysop had to weigh the likelihood that the company would actually sue, and in a faraway venue at that. Whether the company had a valid cause of action was, of course, unclear, and the hundreds of dollars of legal fees required to simply research the issue were beyond the sysop’s means.

Companies interested in defending their products aren’t the only ones who threaten suit. Individuals upset at their treatment at the hands of others, along with users who are locked out of forums because of their behavior (perhaps at the prompting of another user), have also threatened suit, presumably on as slim a reed as "breach of contract" or, state action issues aside, denial of First Amendment rights. (One user in North Carolina told a sysop in New York that "Southern justice" awaited him if he refused to reconsider locking the user out for repeated anti-Cuban remarks.)

These threats, idle or not, present a novel problem. Current private legal regimes generally contemplate action between sophisticated players. Apart from divorce, parties to civil cases are typically companies. When individuals are involved, they are represented by attorneys on contingency or, if defendants in automobile or other personal injury actions, by insurance companies. Non-wealthy individuals traditionally don’t sue each other, unless perhaps for spite, and there are too many other simpler ways one can express displeasure with another’s nonpecuniary transgressions. Libel, slander, intentional interference with business practices — these nonphysical torts are in practice limited to businesses; it’s just too difficult in the physical world for an individual not expressly in the journalism trade (and hired or backed by a publisher of some kind) to get a defamatory article in front of many pairs of eyes. A total free-lancer would have to be risk-assuming enough to have his or her untrue, nasty letter to the editor about a next door neighbor actually published by the paper — which, wanting to avoid liability itself, would put its relatively sophisticated knowledge of libel to work to help filter such material out. The sorts of speech-based harms for which individuals would legally pursue other individuals don’t happen in the real world because in the real world individuals without institutional sponsorship or collaboration cannot speak to the world at large. Not so on the Net.

Precisely because some of the targets of complaints about speech in online forums are individuals who have uttered it, or sysops who have allowed it, there are procedural and substantive differences between a claim of speech-based harm as we know it traditionally and such a claim as we see it online — differences that might counsel substantially different treatment under law.13 Procedurally, a person or company disparaged on the Net can quite possibly sue wherever he, she, or it is located without regard to where the original offender and sysop are. After all, a message posted by a user typing in one location is available to subscribers checking into the forum from anywhere. Answering a lawsuit several states, or a country, away — even to simply demand a dismissal or change of venue — is prohibitively expensive and downright terrifying for the average sysop, not to mention the average forum participant. For the sysop who wishes to manage a forum in conformity with legal obligations, it is near-impossible to become familiar with the laws of every potential or likely jurisdiction. Even when both plaintiff and defendant are in the same location and working from the same set of rules, civil procedure that has developed to provide full process to sophisticated parties only serves to helplessly confuse unassisted individuals. Distance, variety, and the innate complexity of the procedure itself arbitrarily create an extra incentive, known to both parties, for the unsophisticated defendant in such a case to buckle.

Wholly apart from procedure, a defendant must face the fact that the equities of causes of action like libel, slander, and copyright infringement have developed in a context of sophisticated players, usually corporate ones, who can internalize the laws’ nuances, plan rationally around them, and simply adjust their understandings of the costs of doing business accordingly. Disputes arising from casual conversations don’t fit that model. Acquaintances telling tall tales to each other at a local pub or cocktail party, unless they’re high-anxiety libel lawyers on break, do not engage in the corporate counsel’s calculus — even as they foment the lies and baseless opinions typical to such gatherings. The individual participants in online forums don’t either — even as their lies and baseless opinions, spoken similarly off-the-cuff, reach thousands of strangers instead of a table of friends.

It is unclear that, as a matter of policy, we would want to hold individuals’ rantings — no matter how much exposure they get — to the same standards we hold those of professional publishers. Indeed, identical words conveyed in different media by different entities might have quite different meanings or impacts. A comment encountered without attribution on the Net is simply not taken as seriously as, and if false therefore does less damage than, the lead of a front-page story in the New York Times or even the National Enquirer — even if the Net lie enjoys greater circulation than the Times’s and the Enquirer’s put together. To try to swaddle every thought that zings across the Net in the protective anti-libel gear that has grown around news and views purveyed across media run by rational, efficient corporate entities is, indeed, to unplug the hobbyists from yet another frontier and send in the suburban terraformers. Such a transformation is as unnecessary as it is unsightly. The jungle of words that is the many-to-many Net need not be particularly dangerous even if left unmowed by corporate counsel working in tandem with earnest regulators. While it is not realistically within the ambit of an individual Internet message author or sysop to know a whole set of rules and standards about what can and cannot be said before legal action in any of hundreds of jurisdictions is possible (or at least not laughable), it is eminently possible to beware inaccuracy and falsity as a reader. The USDA stamp of inspection on our food makes us less likely to expect it to be downright dangerous, and we can afford to be, and are, more careless in purchasing pretty much anything we find among the cold cuts at the supermarket because of it. Words cannot readily be as life-threatening as rancid meat, and words from one’s peers can be treated with a healthy skepticism — particularly if it is understood that they are not to be subject to the same standards of truth and accountability as those written or uttered by a newspaper or television station. We cannot all be competent, responsible meat packers, but the fundamental values of our society are not at stake when smaller, less reputable food makers are driven out at the margins through the reality or fear of public or private law responding to consumer complaints. We can trust meat packing, like amusement park operating, airplane transport, and steel milling, to professional others, in particular to the superpeople that are corporations. But why would we presume speech to be properly so cabined, particularly when we have found, almost by accident, a venue of nearly unlimited bandwidth through which we can communicate productively with each other before we even know each other?

The danger of an exodus of individuals formerly participating in a speech enterprise, driven out by law insensitive to individuals’ reasonable limits, arose in the briefs and oral argument of Reno v. ACLU,14 the Supreme Court case challenging the Communications Decency Act15 ("CDA" or the "Act"). Defenders of the CDA argued that reams of FCC interpretive cases defining terms such as "patently offensive" and "indecent" clarify the meaning of otherwise troublingly vague provisions of the Act.16 But the CDA would impose criminal penalties on individuals for their indiscretions on the Internet — a far cry from FCC regulations imposing administrative sanctions on radio and television stations well-accustomed to happily litigating their way through the FCC and D.C. Circuit mazes.

Attacking the CDA at oral argument, Bruce Ennis raised the prospect that the Act’s indecency provisions, even tempered by FCC interpretations, would quite easily throttle what many-to-many or individual-to-many communication exists on the Net.17 Even speech that the Act did not prohibit would be stifled; individuals would be reluctant to bet that their own legal interpretations of what was and wasn’t covered would be vindicated were they arrested. Chief Justice Rehnquist and Justice Scalia acknowledged such concerns, and said, in essence, "tough luck."18 They noted that the complaint that individuals are being silenced by onerous media regulation is an old one: the hobbyists experimenting with radio in the 1920’s may have been quite mortified to encounter a sudden raft of airwaves regulation, the sheer burden of which made them shelve their broadcast antennae and leave the field to companies sufficiently large to treat broadcast as a commercial enterprise whose regulatory and legal costs could be passed along like any other. Such regulation had been upheld nonetheless. Their comments suggested that it’s within Congress’s purview to decide that the content of the Net is best generated only by commercial enterprises, just as the lion’s share of movies, television, and radio is today. Congress might find a Net that is 90% "public access channel" by volume and only 10% Disney a bit strange, and have few qualms about enacting policies that favor Disney’s professional entertainment over the public’s amateur discourse.

Whether the CDA’s provisions about indecency stand or fall (and whether or not the constitutional point urged by Chief Justice Rehnquist and Justice Scalia is vindicated), the specter of private tort liability is driving the amateur sysop to take up other pursuits. (The effect may not be as pronounced upon individual forum members who as a practical matter aren’t as identifiable as sysops and may not be similarly appealing targets.) Ironically, for a law described as a threat to the freewheeling Net, one provision of the CDA seems to reduce the legal liabilities to which individual sysops might be exposed. Section 230(c) of the CDA, narrowly titled "Protection for ‘good samaritan’ blocking and screening of offensive material," contains the following quite broad language under subsection (1), "Treatment of publisher or speaker": "No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider."19 While section 230 by its terms is to have no effect on intellectual property law,20 and does not apply to prevent the enforcement of state law "consistent with this section,"21 it does state that "[n]o cause of action may be brought and no liability may be imposed under any State or local law that is inconsistent with this section."22

The Joint Explanatory Statement of the Committee of Conference that accompanied the CDA seems to suggest that despite the narrow title of section 230(c), Congress did not write its contents broadly by accident, but fully intended the Act to provide some degree of immunity from state law for sysops attacked for poorly managing the dynamic exchange of messages written by others.23 Indeed, the conference report states that the section was intended to overrule Stratton Oakmont, Inc. v. Prodigy Services Co.,24 in which the Prodigy online service, through the sysop of its "Money Talk" forum, was found to be the publisher of certain statements by a forum member labeling a particular company’s officer "criminal" and describing the company’s stock offering as a fraud; liability was found because the sysop had allowed those statements to remain in the forum after the member had posted them, while exercising control over other messages.25 Particularly in the wake of Stratton Oakmont, some sysops have sought to intervene in their forums as little as possible: better to suggest that control of members’ behavior wasn’t possible at all than to exercise some control in general, someday not to the satisfaction of a particular plaintiff. The CDA appears to remove this disincentive to hands-on management.

A case wholeheartedly adopting this interpretation, and moving somewhat alarmingly beyond it, is Zeran v. American [sic] Online, Inc.26 There, in a nightmare scenario not wholly unfamiliar to some sysops, an AOL user pretended to be the plaintiff — himself not an AOL subscriber — and posted a barrage of messages falsely offering t-shirts with offensive slogans relating to the 1995 Oklahoma City terrorist bombing. The offer included the plaintiff’s phone number. The result was a series of threats and other harassment of the plaintiff by readers outraged by the offer and convinced of its authenticity.27 AOL was apparently slow to react, deleting repeat messages about the bogus t-shirt offer in a piecemeal fashion, failing to lock out the account posting them, and refusing to announce a "retraction."28 The district court held that section 230 of the CDA preempted the claim against AOL even if it did not in principle preclude all state law causes of action concerning interactive computer services.29 The court suggested that online services — and presumably sysops, too — are immune from liability for defamatory statements not authored by them or their agents, even when they know or have reason to know that the statements are false and they can readily delete them from view.30

The result in Zeran may look like a perfect life jacket for those sysops fearing liability for the discussions that take place in their forums. It’s actually a poor fit. Under Zeran, the scant provisions of the CDA addressing sysop and information service liability at once go too far and fall short of the sort of groundswell necessary to preserve some part of the Internet for individual development and management. Even the most inexperienced sysops, if only slightly apt, know to discourage members from posting telephone numbers publicly, and they would be quick to react, especially if alerted, to an incident like the one that formed the basis of Zeran. Indeed, the facts suggest an absence of any sysop participation in the controversy — apparently the entire matter was handled centrally by AOL. If the facts are as described in the opinion, AOL badly bungled the situation, and it is difficult to understand why AOL was unable or unwilling to discover and disclose the identity of the person masquerading as the plaintiff.

Most sysops would likely be willing to accept some legal responsibility for the quality of their work. An "actual malice" standard borrowed from New York Times Co. v. Sullivan,31 for example, wouldn’t have them answering to a tribunal when some forum members had a dispute over the merits of computer software or called each other names, but it would have them called to account for sheer indifference to an obvious problem like that facing the hapless plaintiff in Zeran. Sysops do not benefit from a small area of overbroad immunity carved from an unlikely statutory source any more than they do from overly nuanced legal discussions grounded in whether they are "distributors" or "publishers." (They are clearly neither; the right legal category is "sysops.") Rather, sysops would warm to a legal regime speaking clearly to more situations: procedurally developed to deal with the problem of faraway venues and substantively developed to guarantee relatively easily understood and attained standards of conduct that would apply across all states, and for which behavior in a case like Zeran (and not much else) would fall short. Icing on the cake would be a federal policy refusing to honor foreign judgments against sysop defendants based on applications of speech-based law that asked more of sysops than the United States standard did.32 With luck, Zeran will trigger an examination of such possibilities, and not just a backlash against even those sysops who would have quickly nipped the situation in Zeran in the bud.


V. Conclusion


For all we know, the hobbyist broadcasters of the 1920’s were destined to be drowned out by, or perhaps become, Westinghouse and its ilk even without a push by the regulations of the newly-chartered Federal Radio Commission. It may be that the idyllic vision of citizen-sysops leading an engaged, physically far-flung circle of diverse people to connection with each other and the interests close to their hearts is also history, regardless of what type of legal regime is applied to sysops. Not only thanks to the dampening effects of law has there been a distinct trend towards the corporatization of the online spaces communities have been known to inhabit; perhaps the maturation of any new field entails a fading of the pioneering spirit that infuses those who originally inhabit it. But the opportunity for experimentation with dispute resolution, law-making, and substantive questions at the intersection of law and society normally reserved to constitutional scholars and philosophers makes online communities as special as they are fragile. The fewer pushes down the gangplank the better.


* Senior Fellow and Executive Director, Harvard Law School Center for Law and Information Technology; Sysop of CompuServe’s Sysop Forum. J.D., 1995, Harvard Law School; M.P.A., 1995, John F. Kennedy School of Government; B.S. (Cognitive Science and Artificial Intelligence), 1991, Yale University. E-mail: I thank Christine Jolls, Chris Kelly, Larry Lessig, and Charles Nesson for helpful comments and inspirations, and Molly Shaffer for that and patient editing.

1. MUDs augment the immediate, textual character of chat with software designed for long-term, repeated interactions. One can "create" objects, for example, still solely comprising textual descriptions and behaviors, which remain from one MUD session to the next.

2. Robert Putnam has documented the decline in civic engagement and blamed, primarily, television. See Robert D. Putnam, The Strange Death of Civic America, INDEPENDENT (London), Mar. 11, 1996, at 13; see also Robert D. Putnam, Bowling Alone: America’s Declining Social Capital, 6 J. DEMOCRACY 65 (1995). Although the Internet may seem to be television’s evil twin — one sits even closer to the screen and does so even more oblivious to one’s local friends and surroundings — community-building Internet use should be thought of as civic society’s ally against the common foes of alienation and apathy. Communities formed through computer networks can and often do spill over into real space, and may preserve some of the core characteristics so valued within now-neglected bowling alleys, community centers, and PTA meetings.

3. The technical means by which newsgroups "happen" are amazingly and counterintuitively decentralized. There is no central location on which a newsgroup’s contents are stored and then distributed to everyone; rather, lots of little hubs ("servers") pass newsgroup messages back and forth among each other, and subscribers connect to one of the various hubs to retrieve the messages. Whoever controls an individual hub could exercise some control over its contents, but that control would only affect those subscribers using that hub. The only common way to exercise control over a newsgroup is to establish it as "moderated," meaning that all contributions are pre-screened by a designated party. In some instances, despite e-mail’s optimization for one-to-one communication, e-mail distribution lists are used to approximate the architecture of a moderated or unmoderated newsgroup.

4. This isn’t to lionize or idealize the culture that prevailed in the good old days before the Net’s floodgates opened, but rather to emphasize simply that there were cultures that had character and flavor, and that their cores were remarkably consistent and durable despite a lack of formal leadership and centralized power: a form of anarchy without chaos.

5. See, e.g., Peter H. Lewis, Sneering at a Virtual Lynch Mob, N.Y. TIMES, May 11, 1994, at D7 (describing the unapologetic foray into advertising of attorneys Laurence Canter and Martha Siegel).

6. It seems an instance of a phenomenon Douglas Hofstadter calls "reverberant doubt." Simplifying somewhat, imagine twenty individuals sitting in a series of cubicles in which each person has a button; if no one presses the button, everyone gets, say, a thousand dollars. If anyone presses the button, a mere hundred dollars is guaranteed to the presser(s) and nothing goes to the non-pressers. Add enough people to the room and people become uncertain enough about others’ motives to start pressing buttons — even though they get even less than they would if they, and everyone else, simply refrained from pressing. See DOUGLAS R. HOFSTADTER, METAMAGICAL THEMAS: Questing for the Essence of Mind and Pattern 752-53 (1985).

7. Such accounts today allow subscribers access both to the Net at large — web pages, newsgroups, chat rooms — and to "proprietary" content like forums, centrally managed and run on the provider’s own computers.

8. Like any lock or security system, sysops’ tools aren’t perfect, but they generally work. There remains some ability for a determined participant to evade a sysop’s sanctions, but it’s not trivial to exercise: for example, one can post a new copy of a previously-deleted message, only to have it deleted again by the sysop, or obtain a new account, only to have it locked out as well.

9. John Seabrook thought he had it bad when he got into the testy e-mail exchange that inspired his anxious essay in the New Yorker. See John Seabrook, My First Flame, NEW YORKER, June 6, 1994, at 70. However stinging a nasty private message may seem, it’s far more provocative when done in a public forum, in front of one’s friends or associates.

10. There are some barriers to a quick departure — one may already have made friends in a forum which has turned out to have a distasteful sysop; the objectionable forum may be the only one within the umbrella service’s bazaar of forums that caters to the topic of particular interest to the member; one’s pride and personal sense of belonging in that forum’s community may make it difficult to leave it. These barriers somewhat insulate a wayward sysop from the embarrassment of a mass exodus, if not from the constructive carping of those who don’t leave. See generally ALBERT O. HIRSCHMAN, EXIT, VOICE, AND LOYALTY: RESPONSES TO DECLINE IN FIRMS, ORGANIZATIONS, AND STATES (1970).

11. Robert Ellickson describes similar informal favor-trading in his account of ranching neighbors in Shasta County, California, who happily ignored their legal rights and relationships in favor of an informal and unspoken system of favors granted and owed. See ROBERT C. ELLICKSON, ORDER WITHOUT LAW: HOW NEIGHBORS SETTLE DISPUTES (1991).

12. A fascinating challenge to this forum common law development process arises from increasing international cross-representation among forum sysops and members. Presumably the ability to select the communities whose membership standards incline to one’s tastes will ensure that general consensus about a sysop’s decisions can remain among members. So far one of the most visible manifestations of the internationalization of forum management is the relish with which sysops outside the United States tell First Amendment-invoking members that they’re invoking a "local ordinance."

13. To be sure, information service providers like CompuServe receive a share of complaints about their members’ speech. Among service providers, sysops, and the users who contribute to forums, however, sysops are especially vulnerable targets even if not the most frequent ones. Sysops are constant, identifiable, and professional-seeming presences in their forums. They are thus more appealing targets than the random offending users whose speech they monitor, even as they cannot afford the same legal resources that the truly professional service providers can.

14. No. 96-511 (U.S. argued Mar. 19, 1997).

15. See Telecommunications Act of 1996, Pub. L. No. 104-104, tit. 5, 110 Stat. 56, 133-43 (1996) (to be codified in scattered sections of 18 U.S.C. & 47 U.S.C.).

16. See Brief for Appellant, Reno v. ACLU, No. 96-511 (U.S. argued Mar. 19, 1997), 1997 WL 32931 at *44.

17. See Oral Argument, Reno v. ACLU (U.S. Mar. 19, 1997), 1997 WL 136253 at *34-36 (claiming that the Act threatens the most interactive Internet applications).

18. See id. at *39-41.

19. Telecommunications Act § 509, 110 Stat. at 138 (to be codified at 47 U.S.C. §230(c)).

20. See id. (to be codified at 47 U.S.C. § (d)(2)).

21. Id. (to be codified at 47 U.S.C. § (d)(3)).

22. Id.

23. See H.R. CONF. REP. NO. 104-458, 113, at 194 (1996), reprinted in 1996 U.S.C.C.A.N. 124, 207-08; S. CONF. REP. 104-230, 113, at 194 (1996).

24. 1995 WL 323710 (N.Y. Sup. Ct. May 24, 1995); see H.R. CONF. REP. NO. 104-458, 113, at 194; S. CONF. REP. 104-230, 113, at 194.

25. See Stratton Oakmont, 1995 WL 323710.

26. No. CIV.A. 96-952-A, 1997 WL 135703 (E.D. Va., Mar. 21, 1997).

27. See id. at *1.

28. See id. at *1-2.

29. See id. at *5, *8.

30. An interesting counterpoint to this case may be provided by the eventual outcome of a separate lawsuit by Zeran against KRXO, an Oklahoma radio station whose disc jockey noticed the messages, believed them, and proceeded to urge listeners to call Zeran’s number to register their outrage. See id. at *2. There is no reason to believe that the CDA would insulate the radio station from liability. An outcome in that case different from the outcome in Zeran itself might rightfully reflect — so far as it represented differential treatment between media — the fact that radio stations are generally corporate, one-to-many, communications facilities; forums and newsgroups, even when linked to corporate service providers like AOL, are platforms for the speech of multiple individuals. While the romantic individualist might even urge policies that encourage radio and television’s return to their pre-corporatization days, the convergence of media through the Internet suggests that if the individualist line can be held there, new opportunities to use aural and visual media will arise, unaffected by the legal regimes governing traditional radio and television.

31. See 376 U.S. 254, 279-80 (1964).

32. Cf. Matusevitch v. Telnikoff, 877 F. Supp. 1 (D.D.C. 1995) (refusing to enforce British libel judgment); Bachchan v. India Abroad Publications Inc., 585 N.Y.S.2d 661 (Sup. Ct. 1992) (same); Abdullah v. Sheridan Square Press, Inc., No. 93 Civ. 2515 (LLS), 1994 WL 419847 (S.D.N.Y. May 4, 1994) (refusing to entertain cause of action brought under British defamation law).