Return to Nicholas Johnson's Main Web Site www.nicholasjohnson.org
 

      Academia’s response was unanimous.3  Zeran was wrong.  Wrong on Congress’ intent.  Wrong on the basics of defamation.  And wrong in its application of public policy.  The most frequently voiced contention was that by exempting publishers from liability for third party content, the Zeran court’s interpretation of the CDA provides Internet publishers far greater protection from defamation liability than is available for ink and paper publishers.4  There is nothing inherent in the Internet, these commentators argue, that could justify the extent of this disparate treatment.5

      Despite the academic arguments to the contrary, it is the contention of this paper that the approach to the CDA taken by the Zeran court is correct

      Occasionally, there is something inherent in its use of the Internet that will set an online publisher apart from more traditional publishers.  Sometimes an online publisher will publish information that is accessed by tens of millions of users per day.  Sometimes an online publisher will allow users to post content to the Internet without any editorial review.  And sometimes, an online publisher will not have the resources to properly screen or monitor its Web service for defamation.  While Zeran might not have arrived at the proper statutory interpretation of the CDA, it does a good job of recognizing the special characteristics of online publication.

      Zeran’s approach to the CDA is particularly appropriate in the context of student-run online message boards.  Using a popular student run Web service at Grinnell College as an example, this paper shows that the reading of the CDA proposed by the opponents of Zeran results in a substantial and unwarranted chill to expressive speech.
 

      In the mid to late 1990’s, Grinnell College used a Vax computer system to provide e-mail and other computer services to its students, faculty, and staff.  Vax systems offer a feature called “plans.”  Plans were text files that were associated with a user’s Vax account.  These text files, called “.plan” files, could be edited by the owner of the account to which they were attached.  Any Vax user could enter a view command followed by another Vax user’s username and the Vax terminal would display that user’s .plan text file.6

      Plans were originally designed to allow employees of a company to view other employee’s “work plans.”  Within a few months of the introduction of Grinnell’s Vax system, Grinnell students began to use the .plan framework as a way to post personal information, social commentary, and other musings in a semi-public forum.  The use of .plan files became such a phenomenon that several student Vax administrators, known on campus as VaxGods, wrote Grinnell-specific software designed to enhance the social dimension of .plan files.7

      In late 1999 Grinnell administrators decided to move the College’s e-mail service away from the Vax to a replacement e-mail system that was entirely Web-based.  This new Web-based system did not provide services comparable to the old .plan framework.  As a result, the .plan system continued to be used by students until the Vax was finally shut down in the summer of 2000.8

      After the Vax system was shut down, many students feared they would not have a convenient online forum for socialization.  Several students began a discussion of ways to revive the .plan system.  After a few false starts, a computer science student developed a Web-based .plan system from scratch.  Originally, the student’s .plan system was intended only for her close friends.  However, it ultimately caught on and was opened to any student who expressed an interest.  Thus, Plans was born.  At its inception, Plans was hosted on the creator’s personal Webpage on the computers of the Computer Science department.9

      Plans was created in the model of many popular blog services.  Students who wanted to access and edit their plan were required to log in to the Plans homepage using their campus username and password.  After login, the user could then edit their plan in a text box on the Plans Webpage.  After completing their edits, a user’s plan automatically would be updated in the Plans directory.  This plan would then be viewable by all other Plans users.  The software and social expectations were such that Plans administrators were unable to review a particular posting before it was made viewable to the Plans community.10

      Plans did not allow for anonymous posting.  A user’s plan was prominently labeled with their username, which was often merely their last name and year of graduation.  Indeed, a particular plan was accessible only if a Plans browser knew that user’s username.  Plans also had an automated “audit trail” which kept track of when particular plans were updated.  The Plans software, however, did allow for anonymous browsing.  The Plans system incorporated a “guest” login procedure that would allow a person without a plan access to the Plans system.  However, guests did not have their own plan.  This guest login procedure was significantly limited in later versions of Plans.  In these later versions, a plan owner could designate their plan as not “guest viewable”.11

      Plans was initially designed to be used exclusively be Grinnell College students.  However, the system became so popular that student organizations, professors, administrators, trustees, and alumni requested their own plan.  In its last incarnation, Plans boasted 1509 unique users.  By way of reference, only about 1300 students attend Grinnell at any one time.  During April 2003, Plans received 3,212,787 unique operation requests -- an average of 107,095 requests per day.  As a result, Plans transferred 21.06 gigabytes of text files during this month -- an average of 719.03 megabytes per day. 12  An average plan contained 4,242.9 characters, which translates to roughly 1060 words or approximately 3.5 pages of double-spaced 12 point Times New Roman text.13  Suffice to say, Plans was an extremely popular service.

      Because Plans was entirely student created, it was also entirely student administered.  Student administrators registered Plans accounts, designed Plans specific Web tools, and made sure Plans usage was adequately monitored.  Other than the occasional comment regarding programming made by members of the computer science faculty, the college only provided Web hosting services for Plans.14

      From the outset, Plans suffered from a lax set of policies.15  However, because it was hosted on the College’s Web servers, Plans was required to follow the campus computing code.  Among other things, the campus computing code required that:

Computing resources may not be used to intimidate, threaten or harass individuals, or violate the college's policies concerning relationships between college constituencies. Such activities include, but are not limited to, using computing resources to store, print, or send obscene, slanderous, or threatening messages.16

It is improper and inappropriate to use the College's World Wide Web servers purposely to annoy, abuse, libel, threaten, or harass anyone, individually or collectively, or to violate state or federal laws.17

 At approximately 3:00 P.M. on July 21, 2003, the student Plans administrator was notified by the head of Grinnell’s Information Technology Service (the “ITS”) that the Plans system was going to be removed from the college’s Web servers at 9:00 P.M.19  At the time, the administrator did not give any definitive reason for this decision.20

      The response of the Plans community was swift.  In the span of a few days, at least two hundred e-mails were sent to the president of the college protesting Plans shutdown.21  In addition to student users, approximately seven hundred alumni signed a “no-pledge” petition stating that these alumni would forgo any financial contribution to the college the following year.22

 To this day, neither the President of the College nor the director of the ITS has definitively stated why Plans was shut down.  However, several themes are present in the President’s e-mail responses to Plans’ users’ protests.  First, the President seemed to believe that some individuals’ plans contained information that was either illegal or offensive, and was therefore in violation of the campus computing code.  Second, the President suggested that posts on Plans were often hostile.  Some plans, he argued, often unfairly singled out individual Plan users for ridicule.  This hostility, he argued, was antithetical to an educational institution that prides itself on acceptance of alternative viewpoints.  Finally, the president expressed concern that information posted on Plans was libelous and might expose the College to liability.  Each of these concerns, the president argued, were systemic.  Eliminating a single user’s plan would not solve these problems.23

      One of the more interesting lines of attack used by student protestors used was that the College was completely immune to liability for any defamation contained in an individual’s plan.  In support of this contention, the protesters cited Section 230 of the Communications Decency Act of 1996.24  The protestors argued that Section 230 of the CDA absolutely protects ISPs, or other owners of servers, from liability for any defamation that arises from third party content published on the Internet.  Consequently, they argued that the College could never be liable in a defamation action merely for hosting Plans.
 

      Because defamation is a state common law cause of action, the particular elements required to establish a prima faciae case of defamation vary by jurisdiction.  However, some general elements in defamation law are common to most jurisdictions.  First, the defamer must make a false or injurious statement concerning another.27  This statement must then be “published” to a third party.28  Finally, the publisher must be at fault in publishing the defamatory material.29  If the person who is defamed is a public figure,30 the defamer must know that the damaging statement is untrue or be reckless in ascertaining its truth.31  If the defamed person is not a public official, a showing of negligence will establish the appropriate fault.32

      Often, defamation causes of action are separated into the smaller subcategories of slander and libel.  The way in which a statement is published is what separates slander from libel.  Libel is publication of a defamatory written statement.33  Slander is publication of defamation through spoken words or gestures.34   In general, the term publication is loosely interpreted to involve any “communication intentionally or by a negligent act to one other than the person defamed.”35  Courts have interpreted publication to include, among other things, stenographic notes of an oral proceeding36 and dramatic pantomime.37

      Generally, any party who republishes defamatory material is just as liable as an original publisher of the defamation.38  Republication fault is often separated into two categories:  republisher fault and distributor fault.  A republisher is someone who takes an active editorial hand in republishing a defamatory statement.  To be liable for defamation, it must be shown that the republisher “had, or should have had, substantial reasons to question the accuracy of the articles or the bona fides of (the) reporter.”39  Essentially, the fault standard for a republisher is negligence.  A republisher also has a general duty to reasonably investigate anything republished.  If the publisher does not reasonably investigate what is republished, it is strictly liable for any defamation the publication may contain.40

      A distributor is someone who engages in no substantial editorial function in republishing material.  Distributors are traditionally understood to be “vendors” of publications rather than editors.  The classic example of a distributor is a bookstore or newsstand owner.  To be liable for defamation, a distributor must have “no knowledge of the libelous material and that there were no extraneous facts which should have put him on his guard.”41  Thus, the fault standard for a distributor is knowledge of the defamation.  A distributor of a third party publication has no duty to review any publication for defamation unless special facts exist that should have put the distributor on guard that it contained defamation.42

III. Defamation in the Information Age.

      The Internet has greatly increased the potential for the rapid and widespread distribution of defamation.  The Internet is the largest forum in which a statement can be published.  Virtually anyone can publish a Web page.  These pages are viewable by millions of people within seconds of their publication.  Individual Internet content providers can begin to host thousands of Web pages a minute.  If a Web site is structured to receive and display third party content, the Webmaster may well not take an active hand in editorializing that third party content.  And any one of these third party pages can contain defamatory statements.  As the Internet has grown in scope, speed, and importance, courts have struggled to develop a comprehensive framework that sufficiently addresses the problems posed by this new medium.  Some of the unique problems with Internet defamation are detailed below.

a. Anonymity.

      The Internet is structured to allow anonymous Web usage.  Web users are able to register anonymous e-mail accounts and post anonymous comments on Internet message boards.  While the anonymity of the Internet provides a powerful voice for the “persecuted, controversial and the simply embarrassed”,43 it also gives a sense of security to the potential defamer.  Anonymous defamers often seem to believe that if they do not attach their name or identity to a Web page or message board, they cannot be held liable for defamation.  While it is possible for a dedicated investigator to determine the actual identity of a message board defamer, the sense of security that the Web’s perceived anonymity provides can embolden a defamer or a potential defamer.44

      In at least one field, this perceived anonymity has emboldened potential defamers into creating a new type of Website.  Dissatisfied, disgruntled, and discharged employees often have much to say about their employer.  Traditionally, the only way to express these frustrations was to either vent around the “water cooler” or after work.  Bitch sites, which are work grievance message boards, arose as a way to express workplace concerns in a more public forum.45  Because many bitch sites allow for anonymous venting, disgruntled employees often feel that they will not be held accountable for their postings.  While this lack of accountability allows for increased protection for whistleblowers it also fosters an environment that is perfect for defamers who do not wish to be discovered.

      For an example of statements made on bitch sites, consider the following messages found on a board for grocery store employees:

“What's with Todd Crouch and all these alleged Women he meets at Bars? They end up as Department Managers at #13 at Cameron Village and we have to deal with them?????????”46

“Todd has intentionally, willingly, knowingly and maliciously violated Mark's Civil rights by discussing private, personal and confidential Personnel information about Mark's pending termination with an hourly non Store management Employee who has no "need to know" and subsequently subjected Mark to great public embarrassment. This is typical of how Todd and Bob operate. ILLEGALLY!”47

“I'm not gonna say who I am, but I can confirm that story. . . David was asked to go with Stuart and a few others to a new store opening in either Jacksonville or Boca. I forget which. David was riding with Stuart, who, according to the David, demanded oral sex in exchange for a ride back to Columbia.”48

b. World Wide Usage

      The Internet is accessible from virtually every corner of the globe.  The defamation laws of country where a foreign defamer is located are often considerably different than U.S. law.  Thus, a party who makes a statement that is clearly defamatory under U.S. law might believe that their statement is not defamatory under the laws of their home country.

 Consider defamation law in South Korea.  The defamation laws of South Korea are notoriously more plaintiff oriented than U.S. defamation law.  For example, South Korean laws have a less stringent fault standard for public official defamation than does U.S. law.  In the U.S., public official defamation liability requires that the publisher of the defamation know that their statement about the public official is defamatory or that they are reckless in ascertaining its defamatory nature.49  South Korea, on the other hand, treats defamation of public officials in the same way that it does defamation of other people; liability requires that the publisher of the defamation did not reasonably believe that the defamation was true.50  Thus, defamation of South Korean public officials requires a showing of negligence while defamation of U.S. public officials requires recklessness.  The disparity between these fault standards is problematic.  A resident of the United States could negligently make a statement about a South Korean official and still expect immunity from defamation liability.  However, in South Korean court, this resident would likely be liable for defamation.

c. Editorial Discretion / Screening

      Organizations that publish third party content on the Internet often take no editorial hand in this content.  Sometimes a hands-off approach is part of an Internet publisher’s business plan.51  One of the reasons that bitch sites are popular is because they provide a forum in which people can express their unedited views of the company being “bitched.”  Sometimes a hands-off approach is a matter of necessity.  On a popular Web message board, it is not uncommon to find thousands of posts per day.  The owner of the message board would not likely have the resources to vet each message either before or after it was posted.52

      Additionally, many publishers do not screen material before it becomes available on the publisher’s Website.  This is particularly true of Internet message boards.  Internet message boards are usually designed to allow messages to become viewable to the general public before the message board operator has the opportunity to authenticate the substance of the post or to verify the bona fides of poster.  Again, it is often the case that a Web publisher does not pre-screen posted information either out of a desire to encourage the posting of material that might not otherwise be posted or out of a sheer inability to screen a large number of posts.
 
d. Scope

      One of the factors that helps spur the popularity of Internet publications is that a publication’s readership is not tied to a physical location.  While it is certainly true that Internet publications have limited circulations, the potential market for an Internet publication is only limited by the number of people who use the Internet.  In effect, the broadness of the base to which an Internet publication can appeal is limitless.  When an online publication’s readership is only a click away, there is nothing to stop an incredible number of people from viewing a particular publication.  Thus, the potential for damage resulting from online defamation is potentially greater than for traditional publishers simply because more people are able to access online defamatory statements.53
 
e. Speed

      Information can be published and accessed on the Internet at blistering pace.  With the advent of easy to use Web publishing tools like Microsoft’s Pagemaker or Macromedia’s Dreamweaver, virtually anyone can design a rudimentary Web page in a few minutes.  Web hosting services are relatively inexpensive.  Accessing defamatory content only takes the click of a hyperlink.

      The speed with which online defamation can be disseminated poses two major potential problems.  First, once defamation is posted on a Website other Websites can quickly link to the defamation, thereby facilitating its quick spread.  Second, defamation can be posted and accessed before anyone realizes that it is defamation.  Even if the defamatory content is later removed, the content might have been accessed by enough people to cause irreparable injury.  Furthermore, someone who viewed the original content could have made a copy of this content, which could then be reposted to another site.  Thus, removal of defamatory content from a Website rarely keeps the material from causing serious, continual damage.
 

      In early Internet defamation cases, courts were usually able to apply the traditional defamation framework in a way that achieved an equitable result.  If a party posted a defamatory statement to a Website, the posting party was liable for the defamation.  However, the traditional defamation framework began to break down when it was applied to Internet service providers that hosted third party content.  Courts struggled with several conceptual problems.  Are Internet content providers original publishers or republishers of third party defamatory content?  Or are these providers merely distributors of defamation?  The two most important answers to these questions were adopted in Cubby, Inc. v. CompuServe54 and Stratton Oakmont v. Prodigy.55

      In Cubby, the defendant CompuServe was a noted Internet content provider.  CompuServe allowed its users to access several specialized message forums created by third party companies.  One of these forums, Journalism Forum, contained a gossip board called Rumorville USA.  Cubby operated its own gossip database called Scuttlebut that was unrelated to either CompuServe or Rumorville.  Cubby claimed that numerous postings on Rumorville falsely reported that Scuttlebut reworded and reposted Rumorville articles.  These false reports, plaintiff argued, constituted defamation.  CompuServe did not pre-screen or take an active editorial hand in any content posted on Rumorville.  Furthermore, CompuServe had no actual knowledge of the defamatory nature of any Rumorville article.  CompuServe left the management of Rumorville and its Journalism Forum to an unrelated Internet service contractor.  Nonetheless, Cubby sued CompuServe because it republished Rumorville articles in its message forums.56

      In the resulting diversity libel action, Judge Leisure of the Southern District of New York held that an Internet content provider is not liable for the defamation of a third party unless the Web provider knew of the defamation or had knowledge of facts that might put the provider on guard.  Because it took no active editorial hand in Rumorville, Judge Leisure reasoned that CompuServe should be treated like a “public library, book store, or newsstand” for purposes of defamation law.  As such, CompuServe should be held to a distributor liability standard for determining fault.  As noted above, distributor liability requires that the distributor know of the defamatory nature of the statement but publish it anyway.  Because CompuServe did not have knowledge of the defamatory nature of the Rumorville posts, it could not be liable as a distributor of this content.  To require CompuServe to investigate all the third party content it provided would have been an undue burden on the service.57

      The most intriguing aspect of Cubby is that the court grounded its decision in the First Amendment.  Because they publish a wide variety of expressive content, the court reasoned that distributors of third party Internet content provide a useful service to the population at-large.  Imposing strict liability on a distributor for carrying defamatory information would result in fewer information distribution services.  Furthermore, the court argued that “if the contents of bookshops and periodical stands were restricted to material of which their proprietors had made an inspection, they might be depleted indeed.”58  Again, requiring inspection for defamation would result in a chilling effect upon information distribution services.  In reaching this conclusion, the court cited Smith v. California59, which held that a bookstore owner could not be required to investigate all of the material he provided for obscenity.  The Smith court reasoned that the “bookseller's burden would become the public's burden, for by restricting him the public's access to reading matter would be restricted.”60

      As in Cubby, the defendant is Prodigy was also a noted Internet service provider.  One of the services offered by Prodigy was an online financial services message board called “Money Talk”.  Money Talk was reputed to be the most read message board in the country.  In Money Talk, an anonymous user posted a statement saying that the plaintiff financial services corporation and its CEO had committed criminal acts in association with an initial public offering of stock.  Because Prodigy hosted Money Talk, plaintiff alleged that Prodigy was liable as a republisher of these defamatory statements.  Accordingly, Stratton Oakmont filed suit against Prodigy.61

      Judge Ain of New York’s Superior Court held that Prodigy was a publisher and not a distributor of its online message forums.  The court found significant that Prodigy retained substantial editorial control over its message forums.  Prodigy employed both automated screening software and human forum managers to ensure that its message boards complied with its computing policy.  The court argued that because Prodigy retained significant editorial control over the content of posts made on its message boards, it resembled less a public library, book store, or newsstand and more a traditional publisher.  Because Prodigy resembled a traditional publisher, it had a duty to scan for defamation in all of the posts in its message forums.  Because it did not adequately scan its forums for defamatory content, Prodigy could be liable for the anonymous defamatory statement in the Money Talk forum.  It should be noted that this holding is consistent with the Cubby decision.  As the Prodigy court pointed out “Prodigy's conscious choice, to gain the benefits of editorial control, has opened it up to a greater liability than CompuServe and other computer networks that make no such choice.”62

      Some commentators argued that the Prodigy court’s decision arrived at an inappropriate result.  The Prodigy court imposed a higher duty of care on a company that developed strong anti-defamation policies than would have imposed on a company that did not screen any of its content.  This increased liability standard could act as an incentive for an Internet service provider to refuse to screen or edit its message boards.  Public policy dictates that online publishers instead be rewarded, or at least not penalized, for attempting to limit objectionable content it provides to the public.63

      The Prodigy court did not consider this concern substantial.  The court recognized that the legal standard it adopted might encourage a paucity of editorial overview.  The court reasoned, however, that market forces would encourage editorial overview.  Thus, the heightened duty to investigate content for defamation would be offset by user interest in pre-screening and editorializing.  Prodigy itself, for example, benefited substantially from its reputation as an Internet service provider that was active in limiting access to objectionable content.64
 

 Congress responded to Cubby and Prodigy by enacting the Communications Decency Act of 1996.  This act was intended both to overrule Prodigy and to remove the disincentive to screen content that this decision caused.  The act was also intended to allow Internet content providers to restrict their customers’ access to content that these services considered offensive.  In particular, the CDA was meant to encourage Internet content providers to develop internal measures to restrict customer access to online adult content.

 The CDA accomplishes these goals in two ways.  Section 230(c)(2) of the CDA provides that “no provider or user of an interactive computer service shall be held liable on account of . . . any action voluntarily taken in good faith to restrict access to . . . material that the provider or user considers to be obscene.”65  This provision is intended to allow Internet service providers to restrict access to content that they deem objectionable.  The hope was that this provision would dampen fears that service providers might have over being held to a higher standard of defamation liability for implementing automated adult Website screening services.  Note that this provision overrules Prodigy.  In Prodigy, the defendant service provider was held to a higher standard of liability than other service providers because it undertook a good faith effort to limit access to offensive material.  Because this section is intended to overrule Prodigy, this portion of the CDA can be read to preclude considering a service provider’s amount of editorial control over published third party content as a factor in assessing defamation liability.

 Section 230(c) (1) of the CDA provides that “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.”  Because publication is one of the requirements of defamation liability, this provision seems to completely preclude defamation liability for the republication of third party content.  Indeed, many commentators believe that this section was the result of lobbying by large content providers like AOL that desired complete immunity from defamation liability.66  These content providers were under significant market pressure to screen out pornographic material but did not also want to incur a higher fault standard for defamation liability than service providers who did no screening.67

      Despite its deceptively simple wording, §230(c) (1) of the CDA has become the single most contentious statute in Internet defamation law.  Most academic commentators believe that the statute precludes only republisher liability and not distributor liability.  Most courts read the CDA to precluding, for good or ill, all Internet service provider liability for publishing the defamatory statements of others.  Two cases are emblematic of these two different approaches taken to the CDA: Zeran v. America Online68 and Barrett v. Rosenthal.69

a. The CDA Precludes Distributor Immunity:  Zeran v. America Online

      In Zeran, an anonymous user of the online message forums at America Online, a large Internet service provider, posted a message naming the plaintiff as a source of tasteless t-shirts that made light of the bombing of the Alfred P. Murrah Federal Building in Oklahoma City.  The posting listed Zeran’s home telephone number as a contact for ordering the t-shirts.  Zeran received a large volume of calls at this numbering response to the advertisement.  Most of these callers left angry and threatening complaints on Zeran’s answering machine.  Zeran even received a few death threats.  On the same day that the initial posting was made on AOL’s message board, Zeran contacted AOL managers and requested that the offending material be taken down.  AOL agreed to remove the offending posts, but numerous similar posts were then made in different forums.  All of these posts came from the same user account.  Despite requests from Zeran, this account was never suspended.  At the height of the incident, Zeran received an angry call every two minutes.  He received so many death threats that he had to enlist local police and FBI in round the clock surveillance of his home.  Eventually, after an Oklahoma City newspaper reported that the sale of the offensive t-shirts was a hoax, threatening calls subsided to approximately fifteen per day.  In the resulting suit the plaintiff claimed that because AOL knew of the defamatory posts and took no action, they were liable for defamation as distributors of the third party defamation.70

      The Zeran court held that Section 230(c) (1) precludes an online publisher of third party content from original publisher liability, republisher liability, and distributor liability.  The practical effect of this holding is that no Internet content provider can ever be liable for publishing third party defamatory content.  This immunity exists whether or not the service provider had actual knowledge of the defamatory nature of the third party content.71

      The court reasoned that Section 230(c) (1) has only one reading:  content providers are not publishers of defamatory statements made by third parties.  If a content provider is not a publisher, the provider can never be an original publisher of the defamatory content.  The court then reasoned that establishing the required fault in a republisher or distributor context merely resulted in an inference of original publication.  To support this contention, the court cited Section 577 of the Restatement (Second) of Torts.  This provision proves that a publisher is “one who intentionally and unreasonably fails to remove defamatory matter that he knows to be exhibited on land or chattels in his possession or under his control is subject to liability for its continued publication.”72  The court read this provision to imply that “different standards of liability may be applied within the larger publisher category, depending on the specific type of publisher concerned.”73  In essence, although republishers and distributors are conceptually different from original publishers, republisher and distributor liability are merely subsets of original publisher liability.  Because original publisher liability is unambiguously precluded by Section 230(c) (1), distributor liability and republisher liability are also precluded.

      The court then argued that §230(c) (1) was enacted “to maintain the robust nature of Internet communication and, accordingly, to keep government interference in the medium to a minimum.”74  The court felt that online message boards offered a unique forum for the exercise of First Amendment rights.  If the CDA did not pre-empt distributor liability, a content provider would need to make a “careful yet rapid investigation of the circumstances surrounding the posted information, a legal judgment concerning the information's defamatory character, and an on-the-spot editorial decision whether to risk liability by allowing the continued publication of that information.”75  While traditional distributors like bookstores and libraries would be able to make such decisions, online content providers, which can publish thousands of unique new posts a day, would likely find this process cost prohibitive.  The court concluded that “faced with potential liability for each message republished by their services, interactive computer service providers might choose to severely restrict the number and type of messages posted. Congress considered the weight of the speech interests implicated and chose to immunize service providers to avoid any such restrictive effect.”76

      Additionally, the Zeran court thought that imposing a distributor liability would likely result in the curtailment of speech only arguably offensive. Under a distributor liability scheme, a distributor notified of defamatory third party material must proceed in one of two ways.  The distributor could make a quick and detailed investigation into the offending material and make a snap legal judgment about the defamatory nature of the content.  Or it could simply remove the material.  Because of the expense and uncertainty inherent a fact based investigation of allegedly defamatory content, most content providers would choose to simply remove the offending material.  This removal without review would be unduly chilling to Internet speech.77

      Finally, the court reasoned that imposing noticed based distributor liability would likely impinge upon a content provider’s ability to scan its content for offensive material.  Any attempt to screen for offensive material might result in actual notice of content that was potentially defamatory.  Upon finding potentially defamatory content, the content provider would be required to undertake a quick and costly investigation of the content.  Even if no actual defamatory material was found during screening, a content provider might worry that a court might later find that their screening was negligent.  The obvious solution to these two dilemmas would be to forgo screening in the first place.  This disincentive to screen would result in a conflict with the CDA’s express intention to support content providers’ attempts at self-regulation.78

b. The CDA Does Not Preclude Distributor Liability:  Barrett v. Rosenthal

      Ilena Rosenthal, the defendant in Barrett, was the director of the Humantics Foundation for Women.  This foundation promotes alternative healing and homeopathic medicines.  Ms. Rosenthal was a contributor to several online Usenet forums that discussed homeopathic healing.  On these Usenet forums, she frequently posted e-mail messages she received from third parties.  One such message accused a leading opponent of homeopathic medicine of illegal acts.  The message read:
"Polevoy, police reports show, STALKED Canadian Radio Personality Christine McPhee, until, terrified, she called in police. He followed her around, affecting disguises, for months--then further terrified her by e-mailing her the details of his stalking actions. Police agencies felt it necessary to assign armed uniformed officers to protect McPhee from Polevoy. Reports show that McPhee was not the only female Polevoy stalked."79

 After informing the defendant that the accusations in the e-mail were false, Dr. Polevoy requested that the defendant remove her posts from the Usenet forums.  The defendant refused and instead posted several messages claiming that Dr. Polevoy and his associates were “quacks.”  Drs. Polevoy and Steven Barrett filed suit against the defendant for defamation.  The California trial court, in part, held that Section 230 of the CDA precluded defendant liability for publishing the third party e-mails on the Usenet forums.  The California Court of Appeals reversed holding that the CDA does not preclude distributor liability.80

 The Barrett court disagreed with the central premise of Zeran.  The court believed that distributor liability was not a subset of publisher liability.  Recognizing that it is true that “defamation requires a publication, and that every repetition is a publication, whether it is effectuated by a primary publisher or by a distributor,”81 the Barrett court argued that Congress intended to limit the CDA to publishers rather than distributors.  The Court reasoned that the drafters of the CDA knew of the distinction between distributors.  With this distinction in mind, if  Congress had “intended section 230 to immunize providers and users not merely from primary publisher liability but also from distributor liability it would have made this clear, as, for example, by adding the word ‘distributor,’ and not merely barring liability ‘as the publisher or speaker’ of information provided by another.”82  In essence, the Barrett court argued that Congress’s decision to use only the word “publisher” evidences an intent to leave common law “distributor” liability intact.

 The Barrett court then turned to the policies that underlie Zeran.  The Barrett court did not believe that imposing notice based distributor liability on an online publisher would be necessarily chilling to free speech.  First, the court noted that “market forces exert enormous influence on the character of information transmitted over the Internet, and the excessive removal of Internet postings, or a type of postings, without any inquiry as to whether they are actually defamatory would not likely be tolerated by the market.”83

      The Barrett court next noted that “it is very hard--indeed, ‘almost impossible’--for plaintiffs to succeed in defamation actions.”84  While the court did not explicitly explain why the difficulty of suit is relevant to a discussion of the CDA, the court seems to suggest that because it is near impossible for plaintiffs to succeed in defamation actions, Internet content providers do not take cognizance of the specific legal requirements associated with publisher or distributor liability.  These content providers instead tend to aver to the general risk that a plaintiff could not prove that the content of the third party message was defamatory.  Thus, Internet content providers will not modify their behavior on the basis of the legal regime adopted.

      The Barrett court also pointed out that, as a practical matter, many online content providers cannot control their distribution of defamatory material.  E-mail providers, for example, are not, as a technical matter, able to limit their transmission of defamatory e-mails. Other online distributors can control the content they distribute.  The Barrett court speculated that the Zeran court was concerned mainly with protecting providers that could not control their distribution of defamatory material.  However, in order to arrive at a plausible statutory interpretation of the CDA, the Zeran court was forced to exempt from liability those service providers that did have the ability to screen and review their distribution of defamatory material.  The Barrett court concluded that “it is one thing to grant an Internet intermediary immunity on the basis of a factual analysis of the degree of control, if any, it exerts over content, but quite another to grant immunity without regard to that critical factor.”85

      Finally, the Barrett court argued that granting complete immunity to content providers will cause what it calls a “clever defamer” problem.  The court argued that if a publisher is not liable for publishing third party content they know to be defamatory, a clever defamer will choose to distribute the defamation of others rather than create original defamation.  For example, someone wanting to defame a judge might decide to solicit Internet articles from others that claim the judge takes bribes.  Even though the publisher knows that the article is false and defamatory, they can never be held liable for the Internet republication.  Furthermore, if the article defaming the official is submitted by an anonymous source, the official will not be able to successfully bring suit against any party.86

      The Barrett court did not address what could have been a strong argument.  The court could have concluded that distributor liability is not a subset of publisher liability as a matter of California common law.  Remember that because defamation is a creature of state common law, its formulation can and does differ between the states.  The Restatement, which was one of the authorities used by the Zeran court, merely represents the most common formulation of defamation law.  If California law treated distributor liability and publisher liability as different theories of liability then the CDA would have no practical effect on California defamation law.  While it would be true that the CDA would preclude treating a distributor as a publisher, this preclusion would be meaningless.  Not being a publisher would have no bearing on distributor liability.

c. Zeran is Wrong, Barrett is Right:  The Academic Response

 Academic commentators generally agree with the conclusions of the Barrett court.  Most commentators agree that Congress’s purpose in enacting the CDA was to “encourage the development of technologies that maximize user control over what information is received' over the Internet, and (2) 'remove disincentives [for ISPs to] develop[ ] and utiliz[e] ... blocking and filtering technologies.”87  If the CDA precludes all defamation liability for third party content, a content provider would have little legal incentive to develop to develop any screening mechanism.  That the content provider is immune from suit would likely mean that they would have no fear of defamation liability and no reason to develop defamation screening procedures.
 Commentators also argue that if Section 230(c)(1) was intended to be an absolute bar to third party defamation suits, Section 230(c)(2) would be superfluous.  Remember that Section 230(c)(2) exempts from liability any content provider that undertakes a good faith effort to limit access to material they deem objectionable.  This section was intended to remove some of the disincentives imposed by Prodigy on the development of technologies that screen third party content.  If Section 230(c)(1) was really intended to be a complete bar to suit, why would Congress “have bothered to put a safe harbor in the CDA for ISPs that may incur liability as a result of monitoring content?”88  It is possible that Section 230(c)(2) was included in the CDA merely as an illustration of one type of defamation immunity.  However, commentators argue that this reading of the CDA is implausible.

      The shutdown of Plans provides an excellent opportunity to put the rationales in Zeran and Barrett to the test.  Before analyzing the Grinnell Plans shutdown under these cases, the peculiarities of Plans needs to be emphasized.  Unlike AOL and CompuServe, Plans was a non-profit Internet content provider.  Unlike Prodigy, it was not popular because it screened for offensive conduct.  And most importantly, Plans was sponsored by an academic institution.
 
      a.  Colleges and the CDA
 
      Academic online message boards differ substantially from more traditional online message boards.  Some of these differences are incredibly important to any discussion of online defamation.  The most important differences are detailed below.

      Many Internet services provided by a college are designed and administered by students.  Student programmers are peculiar animals.  On the one hand, they are members of a cheap and often incredibly motivated labor pool.  On the other hand, they often use student programming as an opportunity to learn effective Web development skills.  Because student programming is often a learning process, student programmers can and do make programming, design, and administration errors.  Indeed, from an educational perspective, these mistakes are the point of student programming.  If a student makes a mistake in an educational setting it is less likely that the student may make the same mistake in a non-educational setting.  Unfortunately, a design error that allows or encourages defamation is not held to a lower standard of liability just because it is made as part of an educational experience.

      Plans initial approach to defamation illustrates nicely the problems of student programming.  Plans was initially designed to facilitate communication for a relatively small number of friends.  Because Plans usage was limited to a small group of people who knew each other, Plans developers thought that Plans would be self-regulating for defamation. 89  That is, Plans developers assumed that because all Plans users were friends, a user would not be inclined to defame another user out of fear of a negative social and reputational backlash.  Strong defamation policies were unneeded.  But as Plans began to provide its services to a wider variety of people with looser bonds and affiliations, defamation became much less self-regulating.  The self-regulating nature of plans changed from a virtue into a liability.  When it finally became apparent that Plans needed to take its defamation policy in a new direction, it was too late.  Plans had already been shut down.

      Another distinct characteristic of an academic message board is that it is tied to a physical location and a physical community.  Grinnell Plans was initially tied to the Grinnell College campus.  Because Plans was tied to the campus, its use cycle tended to coincide with events on campus.  For example, after the suicide of two students in the spring of 2003, Plans experienced a dramatic upswing in usage.  Many students claimed that venting their frustrations on Plans was a kind of emotional catharsis in a difficult time.90  An artificially high usage of a message board is important to a discussion defamation liability because it could place an unforeseen burden on the message board’s administrator.91  In the face of this unforeseen burden, an administrator either might not be able to deal with the volume of complaints it normally receives or might have to spend less time evaluating the merits of each complaint.  Indeed, during a dramatic spike in usage and a resulting deluge of complaints, an academic message board administrator might be more willing to merely remove posts that were the subject of complaints rather than investigate its merits.

      As online message boards become less tied to a physical, every-day reality, they tend to become less effective at self-regulating defamation.  Plans is again illustrative.  Plans began as an online component of a real world group of friends. It was only after Plans users began to graduate and move away from the College that Plans became less tied to a physical community.  As Plans became less of a Grinnell, IA online community and more of a Grinnell College online community, the natural impetus to engage in polite discourse began to fade.  Why bother attempting to soothe the feelings of a Plans user if, with the click of a link, you could ignore that person forever?

      Finally, in an academic setting, wealth plays an important limiting factor in defamation monitoring.  While some universities benefit from large endowments and generous operating budgets, most find that budgeting is the single most important factor limiting the services they offer.  Cheap student Web development is often the only alternative to forgoing an otherwise useful service.  However, as mentioned earlier, using student software developers and administrators can pose significant problems.  Moreover, for some wealthy academic institutions, professional online message board development and administration, while feasible, is not a high priority.

      Despite the fact that Grinnell’s endowment is greater than one billion dollars,92 Plans received no official college funding.  The College only supported Plans by hosting the service on computers in the computer science department.  Plans was entirely student developed, administered, and supported.  The Plans administrators were volunteers.  The only real benefit to Plans administrators was that they gained valuable experience in Web development and administration.
 
b. Plans Liability and the CDA.

      Was Grinnell College liable for defamation posted an individual’s plan?  Under Zeran, the College is clearly not liable for publishing third party defamation.  Because the college provided Web space for Plans, it would normally be considered either a republisher or distributor of an individual Plan.  However, because the CDA precludes the inference of publication, the College would not be a publisher of the defamatory Plan.  Under the Zeran approach, republishers and distributors are merely special kinds of publishers.  And because Plans could not be a publisher it could also not be a distributor or a republisher.  Thus, Grinnell could never be liable for hosting a defamatory Plan.

      Under Barrett, distributors, republishers, and publishers are subject to different kinds of defamation liability.  Plans would probably be considered a republisher or a distributor of a defamatory post rather than an original publisher.  For the College to be liable as a distributor or republisher of a defamatory post, a plaintiff would have to show that Grinnell had some knowledge of the defamatory nature of the post.  Additionally, if a Plans user complained that a post was defamatory, the College would incur an obligation to investigate and possibly remove the offending plan.
 
      c.  Zeran and Policy

 Recall that one of the central concerns underlying Zeran was that imposing notice based distributor liability on online message boards would result in a chilling effect on speech.  The argument was that if, after receiving a complaint alleging defamation, an online message board’s administrator was presented with the option of either removing the offending message or undertaking the expense and difficulty of an investigation and analysis of the message, the administrator would likely choose to simply remove the offending message.  Thus, content would be removed from the message board regardless of whether or not it was actually defamatory.  This result would be unduly chilling on the exercise of speech.

 It is possible that this is precisely what happened when Plans was shut down.  Remember that one of the biggest problems with Plans was that it did not have a detailed policy concerning defamation.  It is possible that Plans was shut down because one too many allegations of defamation made their way to the College President and the Director of ITS.  Rather than take the time to investigate whether or not the offending post was actually defamatory, the College administrators might have decided to remove the service from campus computers altogether.  Additionally, it is also possible that because Plans did not have a strong policy framework for dealing with defamation, the campus administrators decided that developing and implementing its own defamation framework would not be cost effective.  So instead of developing a system that worked to limit defamation, the College might have instead chosen to shut Plans down.  This decision to limit a forum for speech would have been caused by the costs associated with developing a policy that fulfilled the obligations of a notice based liability scheme.

 Zeran was also concerned that requiring investigation of alleged defamation could be cost prohibitive for many content providers, especially when its generates a great number of complaints.  Again, this was certainly true for Plans.  Plans, in its heyday, was administered by one volunteer student.  In addition to administering Plans, this student had a heavy course load and a couple of part time jobs.  At most, the student was able to spend a couple of hours per day administering plans.  A substantial amount of this time was spent either developing new functionality for Plans or registering new accounts.  Requiring this student to thoroughly investigate the numerous complaints that were made about 1500 Plans would have been impractical.  Additionally, if given a choice between investigating each complaint and merely deleting an offending post, it is possible that the student would not have the time and resources to do the investigation.  The student administrator would likely choose the path of least resistance -- deleting the post without investigation.  Again, this choice would have an undue chilling effect upon Plans speech.

 Zeran correctly anticipated that in an online academic message board context, imposing notice based distributor liability would result in increased costs that would, in turn, likely result in a chill on expressive speech.  Although, not explicitly part of the Zeran rationale, one of the most common arguments for allowing absolute defamation immunity for online Web services is that the Web is the ultimate free-speech laboratory.93   Proponents of this theory posit that one of the reasons that defamation exists as a common law cause of action is that traditionally, responding to and discrediting a libelous statement was slow, cumbersome, and expensive.  In a wired world, responding and discrediting libel is usually quick and relatively inexpensive.  Thus, those defamed are generally able to respond to and vitiate the impact of the defamatory statement.

 The “ultimate laboratory” theory fails miserably in a Plans context.  While it is certainly true that defamed Plans users were able to respond to and discredit defamatory messages, defamation on Plans was still fairly prevalent.  Plans administrators, for example, were periodically confronted by Plans users who simply refused to remove or modify defamatory content.  Furthermore, what self-regulation of defamation there was on Plans was probably spurred by the fact that Plans was not anonymous.  Because Plans was not anonymous, it is possible that defamation was limited so that the defamer would not be held accountable for the defamatory content.

c. Barrett and Policy

      One of the arguments underlying the Barrett decision is that adopting a notice based distributor liability schema would pose no substantive chilling effect on speech.  Faced with the choice of removing a particular post or undertaking an investigation, market forces would require a content provider to investigate.  As the Barrett court writes, “the excessive removal of Internet postings . . . without any inquiry as to whether they are actually defamatory would not likely be tolerated by the market.”94

      After the Plans shutdown, tremendous market forces were in operation to keep Plans open.  After the decision to shut Plans down was made, Plans users sent several hundred messages to the President of the College.  Almost two hundred of these messages are archived on the Web.95  The messages almost universally express outrage that Plans was shut down without student consultation and demand that the College continue hosting the service.  In addition to writing letters to the President of the College, more than seven hundred students, alumni, and community members signed a petition stating that they would refuse to donate any money to the college in the coming year if Plans was not reinstated.  Again, remember that Grinnell only has a total enrollment of approximately 1300 students.  These substantial political and economic pressures are the functional equivalent of the market forces that the Barrett court believed would prevent the flippant removal of messages in response to complaints.

      Despite these substantial market pressures, College administrators decided not to reinstate Plans.  By refusing to reinstate Plans after experiencing these pressures, the College administrators must have determined that the damage done by the fallout over Plans would be less than the damage done if Plans was allowed to continue to exist on Grinnell’s Web servers. Thus, these substantial market pressures were not enough to outweigh the potential liability that the College could be subjected to for defamation.

      Zeran opponents also argue that if blanket immunity were given to online content providers, these providers would have no impetus to screen messages for defamation.  In the absence of a motivation to screen messages, the amount of online defamatory content would increase exponentially.  Remember, however, that one of the reasons that the President of the College gave for shutting down Plans was that some of the Plans postings were overly hostile.  The hostility of these messages, which is something entirely different from defamation, was something with which the College did not want to be associated.  Even if the College were completely immune from liability, the concern over message hostility would have probably been impetus enough to establish some screening or message review system.

      The one argument that the Barrett court makes that seems appropriate in a Plans context is its “clever defamer” argument.  Remember that Barrett was concerned that if a content provider had absolute immunity from defamation contained in information supplied by third parties, a clever defamer would be tempted to merely solicit the defamatory statements of third parties.  Certainly, this was a possibility in a Plans context.  Plans users could have posted defamatory statements from third parties in order to avoid any possible defamation liability.  Additionally, it is also possible that because Plans was not anonymous, the “clever defamer” problem would have been greater than if Plans were anonymous.  If a defamer was not able to hide behind their own anonymity, it is possible that they would attempt to hide their desire to defame behind another person’s defamatory statement.
 

      Much of what seems to be driving the criticism of Zeran’s approach to the CDA is a fear that providing absolute immunity to Internet content providers for publishing third party defamation will result in a harm without a remedy.  If Internet content providers are absolutely barred from distributor liability, someone who is defamed online will be unable to recover damages if their defamer is anonymous.  Thus, the anonymity of an original defamer seems to be the real reason that some commentators believe the Zeran court’s reading of the CDA is unfair.

      There are few palatable ways to address the problems associated with anonymity that do not also chill speech.  The simplest option is to require online content providers to verify the actual identity of the suppliers of their third party content.  For example, Internet message board administrators could be required to establish and periodically verify the identity Web posters.  If the message board’s administrators could not verify the identity of the Web poster, they would be required to keep this user from using their service.

      This solution, which I label “post logging” is objectionable on several grounds.  First and foremost, post logging might itself be an inappropriate chill on free speech.  Remember that one of the reasons that anonymity is often considered to support expressive speech is that it gives “persecuted, controversial and the simply embarrassed.”96  Requiring post logging might deter the expressive speech of those Internet users who would feel too embarrassed to express themselves in the absence of anonymity.

      Additionally, post logging would again result in substantial administrative costs.  Faced with the choice of investigating and verifying a suspect message board member’s identity or simply removing the member from the service, an Internet message board user would likely take the path of least resistance – removing the member.  As noted above, encouraging this type of decision would likely result in an undue chill on speech.
 

      Whether or not Zeran arrives at the proper interpretation of the CDA, it arrives at the proper result when applied to poorly funded, student run, online college message boards.  A notice based liability scheme imposes too high of an obligation on the administrators of this type of system.  Faced with the option of either removing or investigating posts that are the subject of complaints, resource deprived academic message boards will often choose the path of least resistance – remove the posts.  Grinnell College’s Plans system was, as shown by the passionate protest after its removal, a useful forum for expression and speech.  The shut down of this vibrant online community is an example of just how costly this chill on expressive speech can be.

1 129 F.3d 327 (4th Cir. 1997)

2 Communications Decency Act of 1996, 47 U.S.C.§230.

3 See, e.g., Sewali K. Patel, Immunizing Internet Service Providers From Third Party Internet Defamation Claims:  How Far Should Courts Go?, 55 Vand. L.Rev. 647 (2002);  Susan Freiwald, Comparative Institutional Analysis in Cyberspace:  The Case of Intermediary Liability for Defamation, 14 Harv. J.L. & Tech. 569 (2001);  Brian McManus, Rethinking Defamation Liability for Internet Service Providers, 35 Suffolk U. L.Rev. 647 (2001);  Michael Spencer, Defamatory Email and Employer Liability:  Why Razing Zeran v. America Online is a Good Thing, 6 Rich. J.L. & Tech. 25 (2000);  Annemarie Pantazis, Note, Zeran v. America Online, Inc.:  Insulating Internet Service Providers From Defamation Liability, 34 Wake Forest L.Rev. 531 (1999);  David Sheridan, Zeran v. AOL and the Effect of Section 230 of the Communications Decency Act Upon Liability for Defamation on the Internet, 61 Alb. L.Rev. 147 (1997);  David Wiener, Publication of Statements Posted on Electronic Bulletin Boards:  Is There Any Liability left After Zeran?, 39 Santa Clara L.Rev. 905 (1999).

4 Patel, supra note 2, at 686.

5 Id.

6 For the purposes of this paper, “.plan” will refer to individuals’ Vax based text files.  “Plans” will refer to Grinnell College’s Web-based plans system in the abstract.  The term “plans” will be used to refer to an individual’s plan on Grinnell’s Plans system.

7 See generally Plans Faq available at http://www.cs.grinnell.edu/~heckr/Finger/plans/faq.html.  See also David Archer, Making Plans, Scarlet and Black, Volume 119, Number 16, Feb. 7, 2003, available at http://web.grinnell.edu/sandb/archives/volume119/16/features/1.html.

8 Id.

9 Id.

10 Id.

11 Id.

12 For a statistical analysis of Plans usage in April of 2003, see http://www.cs.grin.edu/~caseevan/plans_report-april.html.

13 For statistics on the average length of a Plan see http://www.math.grinnell.edu/~kensler/planlove2/. My calculation of the page length of a Plan assumes an average word length of 4 and that an average page of text contains 300 words.

14 See Plans Faq, supra note 7.

15 Sam Rebelsky, Grinnell College Computer Science Professor, explains the Plans defamation policy, or lack thereof, at http://www.cs.grinnell.edu/~rebelsky/Rants/Plans/why#policies.

16 See Grinnell College Academic Computer Use Policies of August 15, 2000 available at http://web.grinnell.edu/ComputerServices/policy.html.

17 Id.

18 See supra note 15.

19 See Professor Jon Stone, The Great Plan Outrage, available at http://www.cs.grinnell.edu/~stone/plans-archive/plans-outage.xhtml.

20 Id.

21 Some of these e-mails are cataloged.  They are available at http://www.cs.grinnell.edu/~stone/plans-archive/plans-comments.xhtml.

22 The online petition has subsequently been removed from the Web.  For a description of its scope see Professor Sam Rebelsky, Note to Campus Professors Re: The Dismantling of Plans, Self-Governance, and Faculty Governance, available at http://www.math.grinnell.edu/~rebelsky/Rants/Plans/faculty-memo.html.

23 A sampling of President Russell Osgood’s e-mail responses are online available at http://www.cs.grinnell.edu/~stone/plans-archive/plans-outage.xhtml,
http://www.cs.grinnell.edu/~stone/plans-archive/plans-comments-07-22.xhtml,
and http://www.math.grinnell.edu/~fuller/plan-emails/.

24 See supra note 2.

25 Restatement (Second) of Torts § 577 comment. b.

26 Id at § 621.

27 Id at § 558(a).

28 Id at § 558(b).

29 Id at § 558(c).

30 “The characterization of public official has been said by the Court to apply at least to those governmental employees who have, or appear to the public to have, substantial responsibility for or control over the conduct of governmental affairs; and it has added that the ‘position must be one which would invite public scrutiny of the person holding it, entirely apart from the scrutiny and discussion occasioned by the particular charges in controversy.’” Id. at §580A comment b.

31 Id at § 580A.

32 Id at § 580B.

33 Id at § 568(1).

34 Id at § 568(2).

35 Id at  § 577.

36 Frankson v. Design Space Intern., 380 N.W.2d 560, 566 (Minn. App. 1986).

37 Tumbarella v. Kroger Co., 271 N.W.2d 284 (Mich. App. 1978).

38 Restatement (Second) of Torts § 578.

39 Karaduman v. Newsday, 416 N.E. 557, 556 (N.Y. 1980).

40 Id.

41 Blabanoff v. Fossani, 81 N.Y.S.2d 372 (1948).

42 See id. and Restatement (Second) of Torts §580.

43  Orit Goldring and Antonia L. Hamblin, Think Before You Click: Online Anonymity Does Not Make Defamation Legal, 20 HOFLELJ 383 (2003). This article presents a perspective on why bitch sites have become pervasive.  The article also emphasizes that while the Web posting might seem entirely anonymous, the patient investigator can find out the identity of most bitch site posters.

44 Id.

45 Id.

46 Available at http://pub86.ezboard.com/femployeemessageboardfrm1.showMessage?topicID=113.topic.

47 Available at http://pub86.ezboard.com/femployeemessageboardfrm1.showMessage?topicID=109.topic.

48 Available at http://pub86.ezboard.com/femployeemessageboardfrm1.showMessage?topicID=191.topic.

49 See supra note 31.

50 Kyu Ho Youm, Freedom of Expression and the Law: Rights and Responsibilities in South Korea, 38 STJIL 123 (Winter 2002).

51 See Goldring and Hamblin, supra note 43.

52 America Online, for example, does not vet all of the messages found on its online forums.  See Zeran, supra note 1.

53 See generally, James French and Rafael Zahralddin, The Difficulty of Enforcing Laws in the Extraterritorial Internet, 1-FALL NEXJOP 99 (Fall 1996).

54 776 F.Supp. 135 (S.D.N.Y.,1991)

55 N.Y.Sup.Ct. May 24, 1995, 1995 WL 323710, 23 Media L.Rep. 1794.

56 Cubby, supra note 54, at 137-139.

57 Id, 140 - 142.

58 Id, 141

59 361 U.S. 147, 152-53, 80 S.Ct. 215, 218-19, 4 L.Ed.2d 205 (1959)

60 Cubby, supra note 54, at 139-140.

61 Prodigy, supra note 55, at 1 - 2.

62 Id. at 5.

63 R. Hayes Johnson, Jr., Note, Defamation in Cyberspace: A Court Takes a Wrong Turn on the Information Superhighway in Stratton Oakmont, Inc. v. Prodigy Services Co., 49 ARK. L. REV. 589, 622 (1996)

64 Prodigy, supra note 55, at 5.

65 The CDA defines “Interactive Computer Service” very broadly.  An “Interactive computer service” includes “any information service, system, or access software provider that provides or enables computer access by multiple users to a computer server, including specifically a service or system that provides access to the Internet and such systems operated or services offered by libraries or educational institutions.” 47 U.S.C. § 230(f)(2).  This definition not only covers traditional Internet service providers like Prodigy or CompuServe, but also any Web hosting service that provides access to third party content.

66 Devon Ishii Peterson, Comment, Child Pornography on the Internet: The Effect of Section 230 of the Communications Decency Act of 1996 on Tory Recovery for Victims Against Internet Service Providers, 24 UHILR 763 (Summer 2002).

67 Robert W. Hamilton, Liability for Third-Party Content on the Internet, 8 Seton Hall Const. L.J. 733, 743 (1988).

68 See supra note 1.

69 9 Cal.Rptr.3d 142 (Cal App. 2004)

70 Zeran, supra note 1, at 328 - 330.

71 Id at 334.

72 Restatement (Second) of Torts § 770.

73 Zeran, supra note 1, at 331.

74 Id at 330.

75 Id at 333.

76 Id.

77 Id at 333.

78 Id at 331.

79 Barrett, supra note 69, at 145.

80 Id, 145 - 147.

81 Id at 156.

82 Id at 156.

83 Id at 162.

84 Id at 163.
 

85 Id at 164.

86 Id at 151.

87 Patel, supra note 3.

88 Id at 685.

89 Plans Faq, supra note 7.

90 See http://www.cs.grinnell.edu/~stone/plans-archive/plans-comments.xhtml#herold; http://www.cs.grinnell.edu/~stone/plans-archive/plans-comments-07-21.xhtml#jarvisd;
http://www.cs.grinnell.edu/~stone/plans-archive/plans-comments-07-22.xhtml#fagan

91 It should be noted that national online message boards often see artificially increased use cycles.  However, these use cycles tend to result from some event of extreme national significance.   These events seem to occur less often than do events in small communities.

92 See http://www.infoplease.com/ipa/A0112636.html.

93 Noah Levine, Note, Establishing Legal Accountability for Anonymous Communication in Cyberspace, 96 Colum. L. Rev. 1526 (1996).

94 Barrett, supra note 69, at 62.

95 See http://www.cs.grinnell.edu/~stone/plans-archive/plans-comments.xhtml

96 Goldring and Hamblin, supra note 43.