Legal/Technical Architectures of Cyberspace

Law of Defamation in Cyberspace

Team Members
Nick Eddy
Alan Kellman
Andrew Lau
Shao-Fei Moy
Executive summary by Andrew Lau
Oral presentation by Andrew Kellman

The popularization of cyberspace gives birth to new medium for defamatory speech to which current libel law does not easily apply. Traditional libel law seems lost amidst the realities of cyberspace, where large audiences are easily reached, publishing comes cheaply, and content can seem to last forever. A key problem is the inherent uncertainty as to who assumes the responsibility for content control. Worst of all, on the Internet, identifying the libeler is significantly harder than in real space. In our white paper we strive to analyze the shortcomings of existing libel law in the terms of a parameterized analysis and an economics analysis. In doing so we also hope to introduce possible architectures that address some of these concerns.

Libel law was originally engineered to remedy the damage to reputation by defamatory statements. Libel consists of a written defamatory (reputation-damaging) statement about an identifiable party, communicated to a third party. Under common law, the courts have established that one who takes a responsible part in the repetition of a libelous statement is just as liable as the original speaker. However, the courts also have long recognized a distinction between those who republish libelous statements and those who may merely pass on the libelous statement without meaningful scrutiny. This distinguishes newspapers and broadcasters, who have control of their content, from bookstores and libraries who do not. In cyberspace various protocols serve as media for potentially libelous statements.

For the purpose of examination in the scope of libel, cyberspace can be separated a few general areas. The first is online services such as America Online, CompuServe, and Prodigy. These companies retail access to their services and content. The other is the Internet, which is comprised of millions of computers inter-linked to provide its services. The primary distinction between the content on online services and content on the general Internet is that most Internet protocols distribute content and ownership all over the world, where no one party can completely control it. Online services, by contrast, have the possibility of control over most of the content offered by the service. Most online services augment their private services with access to Internet protocols.

All of these protocols suffer from the same general issues with respect to libel, especially on the Internet. For example, in general it is harder (if at all possible) under current technology to trace back to the creator of the content than it is in real space.

We can characterize every protocol in terms of five parameters in order to give a better insight into their impacts on defamatory speech. Through examination by these parameters, we are able to characterize and objectively assess the viability of different architectures against various protocols. The effectiveness of a defamatory statement is directly related to the audience reached. A defamatory statement does no damage if the audience is non-existent or unrelated to the libeled party.

A potential libeler's ability to damage reputation is partially dependent on the durability of the protocol with which the statement is transmitted. A protocol that allows a libelous message to have a long lifetime permits the statement to effect a greater audience. A long durability also makes it more difficult to remedy the continued damage of a defamatory statement. For example, messages on newsgroups of the Internet were once read and never seen again, impacting just the initial readers. However, today, companies such as Dejanews and Supernews archive and index the messages extending their lifetime and allowing the messages to effect a greater audience.

Another parameter that characterizes a protocol is the ease of publication on that media. Often, individuals are more likely to make defamatory statements if they are easily posted. Greater difficulty in posting tends to provide time for reflection. Conversely, the ease of publication also describes the ease with which an individual can post a reply or a retraction. The ability for a potential libeler to publish messages anonymously or in the identity of other users describes another parameter of a protocol. The ability to trace back to the author of a libelous message is critical to assigning a remedy to the libel committed. Additionally, users are more likely to defame if it is known that a protocol does not allow their identity to be disclosed. Even worse, forged identities may cause false blame to be assigned.

The last parameter describing a protocol is the amount of content control available, by which the harm of defamation may be mitigated through the removal of libelous messages. Under the current protocol of real space magazines, articles usually are submitted to rigorous editorial control. Furthermore, certain protocols may be calibrated to allow for varying levels of content control.

Traditional libel law assumes that publications that have widespread distribution and a long lifetime are hard to attain. A large audience requires substantial resources in maintaining the distribution chain. In contrast, protocols such as the WWW can reach large audiences with virtually no more cost than reaching a small audience.

Furthermore, in real space it is relatively hard to create professional publications that are believable whereas on the Internet producing a professional looking web site is not much harder than producing a simple one. Traditional libel law is often ineffective in cyberspace if the libeler can not be identified. With current Internet technologies, it is substantially harder to find a libeler than in real space. On the Internet, reaching a large audience is as easy anonymously as it is identifiably.

Clearly, applying traditional law to cyberspace is an exercise in avoiding ambiguity. Seeking to avoid burdensome restrictions on this new medium, congress finessed the traditional distinctions between publisher and distributor by provisions of the CDA which prevent ISPs from being considered publishers of content provided by third parties. While this avoided potentially chilling restrictions on speech, it left libel plaintiffs without recourse against the most potent disseminators of defamatory speech. Our goal therefore is to explore various architectures that take alternative solutions to the traditional law to solve the problem of libel on the Internet. One proposal would be to assign all legal liability to the Internet Service Provider. Internet Service Providers would then be responsible for controlling all content authored by their users. Under this liability regime, ISPs are given a role similar to publishers in real space. With this design, we assume that ISPs are in the best position to inspect content and prevent libel. This architecture brings up the question as to who acts as the role of the ISP held liable? Traditional ISPs, backbone providers, or individual servers? Does the liability fall equally among all ISPs?

Another architecture we propose in our white paper involves users digitally signing their content. Unlike the previous architecture we propose to use social norms as a motivator instead of strict legal liability. Under this "Z" proposal (as it was inspired by Jonathan Zittrain), authors are made liable for their content if they decide to digitally sign their messages. The presumed credibility of signed documents would help shift the norms towards a majority signing their content. With this architecture liability is easily assigned for signed libelous documents and unsigned libelous documents, lacking credibility, are less likely to damage reputations. This architecture has the added benefit of allowing those that do not wish to digitally sign their documents to continue doing so. Again this architecture raises a plethora of issues. For example, whether signing by identity is necessary or if general trace-ability is enough? Who will provide the infrastructure for digital authentication?

The last architecture integrates the previous two by assigning the default liability to ISPs but allowing users to digitally sign their content to assume liability. With this combination, ISPs are able to partition the burden of liability between themselves and the users by instituting policies that require their users to sign. The liability can be pushed down to lowest levels or floated up to higher layers. This architecture has advantages over the previous two by allowing liability to be dynamically adjusted to accommodate the changing protocols and dynamics of the Internet. Furthermore, in contrast with the Z proposal, this architecture adds a layer of liability by implicating ISPs when content is not signed, instead of leaving the harm unactionable.

One way to examine the effectiveness of each of the architectures over traditional libel law, is to examine the way each architecture handles parameters not addressed by traditional libel law.

Under the initial architecture of strict ISP liability many of the problems of traditional libel law go away. This architecture resolves the issues created by the durability, large audience, and ease of publication of cyberspace. Under this regime, ISPs are forced to control all the content authored by the user thus virtually removing all chances of libel. This architecture, however, does not force authors to identify themselves. Instead, by making ISPs filter content, an anonymous post is unlikely to cause harm. The "Z" proposal, like strict ISP liability, attempts to reduce the amount of defamation on the Internet better than traditional libel law does. The "Z" proposal affects the durability and the audience reached of the Internet in similar ways. By shifting the norms, the public dismisses unsigned content and focuses on signed content. Libelous signed content can be traced back to the libeler and therefore liability can be assigned to him or her. This architecture directly addresses the ease of publication. By adding legal liability to the publisher, it becomes harder for a potential libeler to publish to a large audience. The publisher must actively decide if they wish to legally stand behind every piece of content posted. This architecture also eliminates the problem of anonymity in traditional libel law since anonymous content is deemed as noise.

Like the two previous architectures, "Z" with ISP liability addresses the parameters not acknowledged by traditional libel law. This architecture deals with infinite durability and large audiences by forcing at least one party to be liable for each piece of content on the Internet. Publication is made harder by requiring users to either decide to stand behind their content by signing or to relinquish editorial control to the ISPs. This architecture moves the role of content controller dynamically between the ISPs and the users.

A different, yet similar, approach to analyzing the law of defamation in real and cyberspace is an economic analysis. While the parameterization analysis above focuses chiefly on real space parameters and their cyberspace counterparts, the economic analysis instead investigates the private and public costs of defamatory speech.

A basic tenet of the law is that liability should be assigned to the party who is the least-cost-avoider of a particular harm. In fact liability should be assigned to the party best able to ensure the maximum benefit, while avoiding the harm at least cost. In this context harm is the defamatory speech and the magnitude of the harm is measured partly by the audience the speech reaches. The total cost of the speech is the mitigated harm plus the sum of the cost of the mitigation, publication, and research necessary to ensure accuracy of the speech. The total benefit of the speech is the value to the consumer (what they are willing to pay). The overall benefit to society is the total benefit minus the total cost.

Real space defamation law has molded itself around these costs and benefits and designed rules which maximize the overall benefit to society. The law has taken a two-pronged approach to designing these rules. The first prong is the least-cost-avoider. In real space this has amounted to assigning liability to the party best able to mitigate or eliminate the harm. The plaintiff is usually not in the best position to avoid the harm, since he is typically unaware that a defamatory statement will be published until he reads it himself. Typically the distributor is responsible for distributing a large number of publications. It would be a great cost for the distributor to read all of the publications. Rather it seems obvious that the publisher, with resources and intimate association with the editing and authoring of the text, is in the best position to make the accuracy determination. Therefore the publisher is assigned liability.

This dynamic changes in cyberspace as the publisher may no longer be the least-cost-avoider. There are two reasons for rethinking this balance. First, publishers, distributors, and the plaintiff all have different abilities in cyberspace. Second, there is a new potential class of people to assist in mitigating harm - the Internet public. Authors can now act as their own publishers at low cost, and can reach large (and often targeted) audiences without the traditional publishing costs. Aside from allowing ease of distribution to all, it also places more individual speakers in the market with very shallow pockets. Distributors (whether they be ISPs or others) now have electronic versions of the content they distribute. These can be manipulated and searched to quickly and efficiently find offending material. Finally, the Internet community has immediate and low (or no) cost access to "publications." A PICS-like rating system may allow the public to assist in avoiding defamatory speech. This modification of roles changes the cost balance and may shift the least-cost-avoider away from the author-publisher to the ISP-distributor, the plaintiff, or the Internet community.

The second prong is external benefits analysis. This mode of analysis has traditionally determined the standards of proof and the damages for defamation actions. However, in cyberspace, because the public may play a larger role in preventing harm, external benefits may actually influence action of those who may eliminate harm. Much of the content on the Internet is low cost or free, providing smaller economic incentives for the authors and distributors. Additionally, the ability to respond is now vested more in the general public so as to begin to blur the distinction between public and private figures. The economic mode of analysis can be directly applied with respect to the Internet. First, however, we must determine who is actually the least-cost-avoider. It is unlikely that the ISP will be able to provide the content control and editing that is associated with a traditional publisher. Rather, an ISP would most likely have to use some sort of automated filtering or flagging system. This would be extremely inefficient given present technology, although one could imagine that such a system might eventually prove useful.

At the same time, however, ISPs are more likely to have greater resources than individual authors have. There is a tension between the knowledge of the author (how much research he did, what he believes the accuracy of the information to be) and the great resources of the ISP, which are difficult to apply individually to every author. The Internet public may also be able to mitigate the harm. If a PICS-like system were implemented to rate the reliability of certain sites, defamatory speech could quickly be labeled as such and ignored. This would be the equivalent of preventing the harm. The PICS-like approach would be inexpensive since the costs would be distributed over the entire cyberspace community. However, this type of rating system is likely to be slow compared to author and ISP (automated) censorship.

Clearly, applying traditional libel law to cyberspace is extremely difficult as the dynamics make the assignment of efficient content control very difficult. As the popularity of the Internet grows, so do these problems. The solution therefore is to not attempt to directly attempt to map real space parameters to cyberspace but instead locate the key issues and build around them. Architectures should not attempt to take away from the parameters of the Internet that make it so popular. Rather, they should accommodate the parameters of cyberspace for the purpose of minimizing defamatory speech. Existing architectures, however, have not provided a clear template for accommodating current and future protocols. In the mean time, potential plaintiffs and defendants will instead have to take a cautionary approach to the hazards of this explosive medium.

Return to conference page

Send comments about this site to

Last modified: December 2 1998, 10:34 PM