DISPUTE RESOLUTION IN CYBERSPACE:

ENGINEERING A VIRTUAL MAGISTRATE SYSTEM

David G. Post 1

Cyberspace Law Institute Working Paper # 2

Presented at the NCAIR Conference on On-Line Dispute Resolution

Washington, DC

May 22, 1996

Do we need a Virtual Magistrate? Does it address a problem that really needs solving, and does it do so in a way that suggests that it is likely to be successful? I take it that these questions are likely to be the focus of much of our attention during this Conference. By way of background and introduction to many of the issues that will be addressed here, I would like to take this opportunity to engage in a bit of legal systems engineering. I hope to present a simple model of dispute generation and resolution that, notwithstanding its simplicity, may help to define the boundaries of the problems to which the Virtual Magistrate is addressed, and to ask whether the features of the Virtual Magistrate are well-suited to the task of solving those problems.

The Problem Any systems engineering project should begin with as careful a specification as possible of the problem to be solved. The Virtual Magistrate has a particular and, perhaps, a somewhat unusual focus. 2 The Virtual Magistrate is not a general purpose dispute resolution system, designed to help any and all parties resolve any and all disputes they may have in the electronic sphere. Rather, it was specifically designed to address a set of disputes that occupy a discrete portion of “dispute space” and have the following generic form:

The Standard Dispute Form. One party -- whom I will call Complainant -- asserts that a second party (Actor) has posted a message or a file on a system under the control of another party (Sysop) containing “wrongful content” of some kind, e.g., material that infringes Complainant’s copyright or trademark rights, misappropriates trade secrets belonging to Complainant, is defamatory or fraudulent or inappropriate (obscene, lewd, or otherwise violative of system rules), and demands that the offending posting be removed from the system under Sysop’s control. 3

We might consider the issues raised in Religious Technology Center and Bridge

Publications, Inc., v. Netcom On-line Communication Services, Inc., et al. 4 as a example of the paradigmatic dispute of this kind. In its opinion, the court described the genesis of that dispute as follows:

Plaintiffs Religious Technology Center ("RTC") and Bridge Publications, Inc. ("BPI") hold copyrights in the unpublished and published works of L. Ron Hubbard, the late founder of the Church of Scientology ("the Church"). Defendant Dennis Erlich ("Erlich") is a former minister of Scientology turned vocal critic of the Church, whose pulpit is now the Usenet newsgroup alt.religion.scientology ("a.r.s."), an on-line forum for discussion and criticism of Scientology. Plaintiffs maintain that Erlich infringed their copyrights when he posted portions of their works on a.r.s. Erlich gained his access to the Internet through defendant Thomas Klemesrud's ("Klemesrud's") BBS "support.com." Klemesrud's BBS is not directly linked to the Internet, but gains its connection through the facilities of defendant Netcom On-Line Communications, Inc. ("Netcom"), one of the largest providers of Internet access in the United States.

After failing to convince Erlich to stop his postings, plaintiffs contacted defendants Klemesrud and Netcom. Klemesrud responded to plaintiffs' demands that Erlich be kept off his system by asking plaintiffs to prove that they owned the copyrights to the works posted by Erlich. However, plaintiffs refused Klemesrud's request as unreasonable. Netcom similarly refused plaintiffs' request that Erlich not be allowed to gain access to the Internet through its system. Netcom contended that it would be impossible to prescreen Erlich's postings and that to kick Erlich off the Internet meant kicking off the hundreds of users of Klemesrud's BBS. Consequently, plaintiffs named Klemesrud and Netcom in their suit against Erlich, although only on the copyright infringement claims.

Here, then, we have Complainant (Religious Technology Center), Actor (Dennis Erlich), and two Sysops 5 (Klemesrud and Netcom), and a charge that material appearing on the two online systems infringes Complainant’s copyright. 6

Complainant’s charge, of course, may be true or false as a matter of fact -- i.e., the Actor may or may not be responsible for the actions underlying the complaint -- and may be redressable or not as a matter of law -- i.e., there may or may not be any legal redress for the harm that Complainant has suffered. Notwithstanding this uncertainty, Sysop must choose some course of action. By hypothesis, the offending material is under Sysop’s control; because Sysop can, therefore, act to prevent any continuing injury to Complainant, even a decision to “do nothing” is, in a sense, an affirmative act that maintains the status quo (and perpetuates whatever harm may flow from Actor’s actions). 7 For expositional simplicity, I will discuss only two of Sysop’s options: Sysop can either delete the offending posting or “do nothing” (i.e., leave it unmodified). I will refer to these in what follows as the “Options,” and to the choice between the Options as the “Sysop’s Dilemma.”

The Sysop’s Dilemma is, essentially, the problem that the Virtual Magistrate system engineers had in mind. The Virtual Magistrate, in effect, offers to take the choice between the Options out of Sysop’s hands, and to make that choice on the basis of an independent inquiry -- to determine, in the words of the Virtual Magistrate FAQ, “whether it would be reasonable for a system operator to delete or otherwise restrict access to [the] challenged file or posting.” 8

The Existing Alternatives. If this is the problem that the Virtual Magistrate is designed to address, we should first inquire into the need for a solution to this problem before we address whether the current Virtual Magistrate design is adequate to the task. If it ain’t broke, as the saying goes, don’t fix it -- what existing mechanisms can Sysops utilize to choose between the Options in the absence of the Virtual Magistrate, and in what ways (if any) are those mechanisms inadequate? 9

The existing mechanisms can be modeled as a three-step process of (1) Investigation, (2) Decision, and (possibly) (3) Litigation. Investigation entails obtaining information regarding both the veracity of the Complaint as a factual matter -- did Actor actually do the things that Complainant attributes to her? -- and the source of “law” that governs Sysop’s response -- in what sense is what Actor did “wrongful”? I have placed “law” in quotation marks simply to illustrate that the set of principles to which Sysop refers as a source of guidance in regard to its conduct in respect of any dispute in the Standard Form may include the output of what we generally call the “legal system” -- primarily statutes, actions of relevant administrative agencies, and reported court decisions from previously decided cases. In other cases, this may include only reference to Sysop’s standard “Terms and Conditions,” or to “common sense,” i.e., that complex set of principles, comprised of some formal law and much informal custom, that we all carry around in our heads and to which we refer when presented with claims of “wrongful” action.

The Investigation step is necessarily resource constrained, because obtaining information of this kind is costly and time-consuming. The extent of the resources that Sysop will devote to these tasks will undoubtedly vary greatly across Sysops. Large systems, such as America Online and Compuserve, may have a number of full-time staff persons devoted to this Complaint review effort, as well as lawyers who can be consulted about the applicable legal principles; other Sysops will devote fewer resources to this task, and some none at all. Investigation, in other words, may range from full-blown fact- finding and internal legal review, on the one hand, to reading the Complaint and attending to it momentarily.

In step 2, Sysop must decide which of the Options to select. As noted above, this is a necessary step; because Sysop is in control of the offending message and thereby possesses the power to remove it, even a decision that consists entirely of tossing the Complaint into the trash has consequences for the Complainant and the Actor. In other words, this is a form of dispute resolution however Sysop acts; the dispute described in the Standard Form has been “resolved,” however unsatisfactorily, even by the act of ignoring the Complaint entirely.

Sysop’s decision is, however, non-binding, in the sense that any of the parties may potentially move to step 3, escalating the dispute to an alternative dispute resolution system, the courts, to review Sysop’s decision and to render a binding judgement on it. 10 That is, the parties may deposit the dispute in a court of law and request that the court determine whether or not any particular action taken by Sysop was, or was not, wrongful and therefore the basis of liability. 11

But litigation, of course, is costly. This is hardly a novel observation, or one that is in any way unique to the cyberspace context. If people act rationally, we expect them as a general matter to pursue litigation only if their expected recovery exceeds their expected costs by some margin. 12 And given that there are, generally speaking, substantial fixed costs to pursing a litigation option, “small claims” -- in cyberspace just as in the non-virtual world -- are unlikely to be pursued to step 3. 13 The implications for our model are that Sysop’s resolution at the second-stage of the process will, at least for a substantial number of cases, be final. 14

Litigation is also slow. Again, this is a commonplace observation, but here there may be important implications for dispute resolution in cyberspace. Even in those instances where legal system oversight is called upon -- i.e., where one of the parties invokes her right to pursue a claim through litigation -- a substantial period will likely elapse between the time of Sysop’s initial notification of the Complaint and any resolution in court. During this period, at least, Sysop’s interim decision is controlling. And because one of the features of cyberspace is that information travels rapidly, the harm complained of can in many cases do a great deal of damage during even a short period of time, in comparison to analogues in the non-virtual world. Files with infringing material, messages with defamatory content, and the like can be disseminated widely long before the legal system can respond. This implies that Sysop’s decision, even where it is not final, is significant in terms of minimizing overall harm.

Weaknesses of the Existing Alternatives. Is this existing system ill- designed? If this process, in the majority of cases, terminates after Sysop’s decision in step 2, is that a problem that needs rectifying at all (let alone through the specific design of the Virtual Magistrate project), or merely an interesting feature of this new environment? If Sysops consistently make the “right” decisions when confronted with Complaints, then perhaps the system ain’t broke at all, and needs no special remedial procedures. How confident are we that they do so?

To answer that question in any comprehensive manner, of course, one would have to have some working definition of what is the “right” or the “wrong” answer in any particular case. I will elide that most difficult of questions, and state simply that I believe we can all agree that the decision-making process should, ideally, give “appropriate” (however defined) weight to the interests (however defined) of both the participants in the dispute and all non-participants whose interests may be affected by the dispute. However one might operationalize those definitions in regard to determining what weight is “appropriate” and what interests “count” in this calculus, I would suggest that there are two important systematic biases that might cause our confidence in the Sysops’ decision-making abilities to be less than rock-solid. This is not, I hasten to add, because I assume that most Sysops are somehow venal and have no desire to reach the “right” decision, balancing the interests of Complainant, Actor, and the community as a whole. But I believe that they are likely to be unable, or unwilling, to do so in many cases.

The first problem is, simply, inadequacy of information. As noted above, the stage 1 information-gathering process is costly. Cyberspace is an environment where taking on the role of Sysop is relatively easy; a fully functional BBS can be set up with a minimal capital investment, and listservers or Usenet groups can be implemented at virtually no cost at all. 15 The category of “Sysops” is thus likely to contain large numbers of individuals or entities with few, if any, resources to devote to the stage 1 investigative task. However one might choose to define the characteristics of the “right” decision in any case, it is likely that many Sysops will be unable to devote the optimal amount of investigative resources to the task of determining what that decision might be.

The second problem is a bit more complex and perhaps more troubling. Let us return to the facts of the Netcom case. In a preliminary ruling in that case, Judge Whyte refused to dismiss charges of contributory copyright infringement against the Sysop, Netcom, and held that Netcom may be liable for contributory infringement if it “knew or should have known” about the infringing postings. The Church of Scientology had, in fact, notified Netcom by letter that, at least in their view, infringing material had been transmitted over Netcom’s system; this, the judge held, creates a factual question . . .

“as to whether Netcom knew or should have known that Erlich [the defendant who originated the postings] had infringed plaintiffs' copyrights . . . .Because Netcom was arguably participating in Erlich's public distribution of plaintiffs' works, there is a genuine issue as to whether Netcom knew of any infringement by Erlich before it was too late to do anything about it. If plaintiffs can prove the knowledge element, Netcom will be liable for contributory infringement since its failure to simply cancel Erlich's infringing message and thereby stop an infringing copy from being distributed worldwide constitutes substantial participation in Erlich's public distribution of the message. 16

In other words, upon receipt of the Scientologist’s letter -- the Complaint, in my parlance -- the Sysop faces the following risk calculus in choosing between the two Options (i.e., whether to delete, or leave unmodified, the challenged postings): On the one hand, there is a risk of liability if the “leave unmodified” option is chosen, if (a) the Complainant seeks court review of Sysop’s decision and (b) Complainant’s claim of injury turns out to be a valid one. On the other hand, the risk of liability is far less (and may be non-existent) if Sysop chooses the “delete” option; neither Actor, nor the larger online community, is likely to have a legally-enforceable right against Sysop that is violated by deletion of the challenged posting. 17 If Sysops are risk-averse, 18 the rational course of action in such circumstances is to avoid the expense of extensive investigation and decision-making procedures and to move immediately to delete the offending messages. Thus, there is a bias operating to skew Sysop’s decision in Complainant’s favor. 19

Can the Virtual Magistrate Alleviate this Problem? The Virtual Magistrate can therefore best be viewed not as an alternative to litigation, but as an alternative to existing procedures whereby Sysops themselves make reviewable, but oftentimes de facto final, decisions about the merits of Complaints. That existing process, I have suggested, is problematical because Sysops are likely to reach decisions that are either uninformed or biased towards one of the parties, decisions that are unlikely to be optimal when viewed from the perspective of all participants and all persons affected by those disputes. Our question, then, is whether the Virtual Magistrate, as currently designed, is an improvement on the existing decision-making system? That question itself has two important components. First, we might ask ourselves the following hypothetical question: If we owned cyberspace -- if our sole objective were to maximize the overall welfare of all of the participants in online communications, and if we were in a position to mandate the use of the Virtual Magistrate procedures in lieu of existing procedures -- would we do so? Is there any reason to think, in other words, that the Virtual Magistrate is more likely to reach the “correct” result than Sysops? And second, if it is indeed an improvement, how can we explain the apparent hesitancy on the part of the online community to utilize the Virtual Magistrate’s services? I offer the following observations and questions as a springboard to what I hope will be additional discussion during and subsequent to this Conference about these important and difficult questions.

In reference to our first question -- would we impose the Virtual Magistrate if we were the owners of cyberspace? -- if I am correct that many Sysop decisions are flawed because of inadequate or unreliable information about either the relevant facts involved in the dispute or the relevant sources of law governing the resolution of the dispute, the Virtual Magistrate may hold promise in many cases as an alternative information gathering mechanism. This is likely to be of particular significance for smaller Sysops, who are likely to have fewer resources to devote to the information-gathering task. This leads to my first suggestion: the Virtual Magistrate services might most profitably be directed towards -- marketed to, if you will -- the community of small-system Sysops who are likely to be most in need of this information-gathering service.

I have suggested that systematic bias, and not merely imperfect information, skews Sysop decision-making, systematic bias that is a function of aversion to the risk of being held subject to externally-imposed liability rules. Any decision-maker who, like the Virtual Magistrate, demands a waiver of liability prior to hearing a Complaint 20 can avoid that bias and afford to act as a risk-neutral decision-maker, precisely because it is not taking on any of the decision-making risk (at least as measured by the likelihood of liability being imposed as a consequence of any particular decision). But as the owners of cyberspace, we clearly should not be persuaded that the absence of this liability-imposed bias by itself guarantees higher-quality decision-making. Indeed, one could argue just the opposite, that the absence of any risk of liability increases the likelihood of irresponsible decision- making. After all, we impose liability on certain parties in certain circumstances for the precise reason that we intend to “skew their decision-making” in some fashion, i.e., away from the disfavored course of action. What other information, then, would we require about the Virtual Magistrate before we were persuaded that the Virtual Magistrate’s decision-making process is a better one than the existing alternatives?

This question has far too broad a sweep for me to answer it here in any remotely comprehensive manner. I would suggest, however, that the information we now have about the Virtual Magistrate may not be persuasive in this regard, and that perhaps the Virtual Magistrate system engineers need to do some additional work along these lines. If we owned cyberspace, presumably we would ask: as to process, are the Virtual Magistrate procedures fair (or, at least, fairer than existing alternatives provided by Sysops)? And as to substance, will the results of Virtual Magistrate decision-making be just? The information currently available about the Virtual Magistrate appears, at least to this eye, to be long on process but short on substance. We have a reasonable amount of information about the way that Complaints will be handled by the Virtual Magistrate, 21 but very little information about the substantive underpinnings of any Virtual Magistrate decision-making. 22

This, I daresay, was not unintentional. Perhaps it is in the very nature of decision- making in this new and uncertain context -- with the difficult jurisdictional and choice-of- law quandaries that it presents, and the general uncertainties that attend to any attempts to transpose traditional legal principles in the digital online environment -- that make it difficult to point to a specific and pre-existing body of rules that any decision-maker is going to apply. But without such a body of rules to be applied in given disputes, how can we, as the owners of cyberspace, have any confidence in this process?

My colleague Lisa Bernstein has in fact suggested that much successful commercial arbitration occurs in a context where arbitrators are bound to a known set of rules -- the typical context being a trade association, where inter-member disputes are handled with reference to published codes of conduct or other default rules. 23 But this need not always be the case for arbitration to be successful. Two alternative models have proven successful. In many close-knit commercial communities, arbitrators, chosen from among the members of those communities, can apply the unwritten “trade custom” applicable in the particular community. 24 This model, one suspects, will be difficult to apply to the kinds of disputes that the Virtual Magistrate expects to resolve. 25 But there is also so-called “wise-man” arbitration, arbitration in which the disputants agree to submit their dispute to persons specifically chosen to be knowledgeable and who are widely respected by the persons affected by the decision-making process. 26 One way, this model suggests, that we can persuade the owners of cyberspace that the Virtual Magistrate will be better than existing alternatives is to expend additional energy to defining the relevant characteristics of the “wise men” (and, needless to say, wise women) who will be serving as magistrates and to recruit, perhaps, a number of such individuals (and publicize their willingness to serve as magistrates).

We might, on the other hand, take some small measure of confidence from the public nature of the Virtual Magistrate decision-making process. 27 The alternative process -- under which decision-making authority resides with Sysops -- is effectively private; Sysops need not, and in my experience generally do not, offer any explanation of their reasoning in particular cases. The relative transparency of the Virtual Magistrate process is likely to improve the quality of decision-making, at least insofar as it can facilitate the development of a better-functioning market for dispute resolution procedures. That is, as the owners of cyberspace, we may accept the general principle that we are not in a position to determine whether the Virtual Magistrate, the Sysops, the courts, or some other party is likely to be the best qualified decision-maker in regard to the Standard Disputes. We might choose therefore to leave that decision to the users themselves, and attempt to facilitate a well-functioning market for dispute resolution procedures. Markets need information to function; the current market is likely to be ill-functioning precisely because Sysop decision-making procedures are largely invisible to users, who therefore cannot base their choice among online systems on the basis of the quality of those procedures.

The Virtual Magistrate, however, is likely to continue to suffer from a variant of the long-standing engineering dilemma, the chicken-and-egg problem: Until the Virtual Magistrate has had the opportunity to demonstrate the quality of its decision-making it may be difficult to persuasively contend that it is an improvement over existing systems, but without such a persuasive case potential participants may be unwilling to allow it to exercise decision-making authority with actual disputes and therefore to demonstrate the quality of its decision-making. This leads to another suggestion: perhaps the Virtual Magistrate could hold a number of “mock trials” as a means of both illustrating the way in which its procedures will operate with “real” disputes and also of demonstrating the quality of its decision-making?

Of course, even if we were to assume that the quality of Virtual Magistrate decision-making were an improvement on the current system, that does not guarantee that it will be utilized. We don’t own cyberspace, and cannot force disputants to use particular forms of dispute-resolution. How do the private interests of the relevant parties align with the public interest in better dispute resolution? Is it in the interest of any of the parties to invoke this mechanism ex post (i.e., after a dispute has arisen)?

This is likely to be a relatively serious problem for the Virtual Magistrate, for two reasons. First, it is a voluntary system, requiring all parties -- Sysop, Actor and Complainant -- to consent to the submission of a dispute. Therefore, it must be in the interest of all parties before the Virtual Magistrate can participate in the process. That will generally be difficult to achieve in the tort context; in Lisa Bernstein’s words again,

“In the tort context, the parties rarely have the opportunity to bargain before the dispute arises, and, because delay almost always favors the defendant, it is unlikely that they will agree to use private ADR after a dispute has arisen.” 28

This might suggest that the Virtual Magistrate loosen its requirements with respect to the “unanimous consent to jurisdiction” rule, and/or that it explore ways to bind potential participants ex ante, i.e., before disputes have arisen, by contractual means.

In addition, while the public nature of the Virtual Magistrate proceedings may produce a public benefit (as discussed above), it may also be viewed as a cost to many potential participants. Most successful arbitration programs are conducted in secret, and this may be regarded as a significant feature in determining whether individual disputants participate or not. 29 This suggests that we might consider some mechanisms (involving, perhaps, pseudonymous participation in Virtual Magistrate procedures) that would allow the reporting of cases without compromising the privacy interests of the participants.

Finally, there is the question of Sysop incentives to utilize the Virtual Magistrate’s services. As sketched out above, the existing dispute-resolution system may in fact serve Sysop’s interests reasonably well, at least insofar as it provides Sysops with a means to minimize their potential liability. It is unlikely to be in the Sysop’s interest to utilize the Virtual Magistrate unless the Virtual Magistrate can provide equal or greater protection from potential exposure to liability. It may well be argued -- in fact, I would be prepared to argue -- that use of a neutral decision-making procedure like the Virtual Magistrate constitutes the sort of action that should immunize Sysops from contributory liability for failing to delete allegedly offending postings, and the Virtual Magistrate system engineers and the Cyberspace Law Institute may wish to consider formally preparing that argument for public discussion and debate and possible use in court. But however persuasive that argument may be, it is likely to have limited force unless and until it is successfully utilized as a shield against imposition of liability. This leads to my final suggestion, that perhaps the Virtual Magistrate will be of greatest utility for those Sysops who can not so easily control their own exposure to liability by means of choosing the “delete” option, e.g., public library or university system administrators who are subject to the First Amendment and therefore to claims brought by aggrieved Actors.

Footnotes

1 Visiting Associate Professor of Law, Georgetown University Law Center and Co-Director, Cyberspace Law Institute. Email: david.post@counsel.com. Thanks to Dawn Nunziato for helping to refine some of the thoughts expressed in this paper, as well as to the participants in the Tooldisp discussion group (organized by Tom Bruce at Cornell University Law School) and the Cyberia-l discussion group for their many thought-provoking comments.

2 Information about the Virtual Magistrate project is derived from the information provided at the Virtual Magistrate World Wide Web site, in particular the Virtual Magistrate Rules and Virtual Magistrate Frequently Asked Questions, available online at http://vmag.law.vill.edu:8080/, as well as the Cyberspace Law Institute Background Paper available online at www.cli.org.

3 See, for example, the following description taken from the Virtual Magistrate FAQ:

What Will the Virtual Magistrate Decide?

The Virtual Magistrate Project accepts complaints about messages, postings, or files allegedly involving copyright or trademark infringement, misappropriation of trade secrets, defamation, fraud, deceptive trade practices, inappropriate (obscene, lewd, or otherwise violative of system rules) materials, invasion of privacy, and other wrongful content.

The Virtual Magistrate will decide whether it would be reasonable for a system operator to delete or otherwise restrict access to a challenged file or posting. Other cases may call for decisions about the disclosure of the identity of an individual to a person other than the government. In extreme cases, the Virtual Magistrate may rule on whether it is appropriate for a system operator to deny a person access to an online system.

The Virtual Magistrate Project will not decide questions about billing or financial obligations as between users and system operators.”

4 ND Cal, No. C-95-20091 RMW, Nov 25, 1995 (Order denying defendant’s motion for summary judgment).

5 Generally speaking, a Sysop, for my purposes, can be viewed as any party with control over a potentially offending message; this covers everyone from the operator of giant online systems such as America Online or Compuserve, to moderators of Usenet newsgroups, to proprietors of small BBS systems.

6 More information about the disputes between the Church of Scientology and a number of online participants can be found at Ron Newman's “Scientology v the Net” home page at

http://www.cybercom.net/~rnewman/scientology/home.html, and in Post, “Flame On: The State of Nature and the First Internet War,” Reason, April 1996, pp. 28-33 (available online at http://www.reasonmag.com/reason/9604/Fe.POST.text.html).

Many of the other well-known legal disputes regarding online activities can be cast in the Standard Dispute Form as well. See, e.g., Cubby Inc. v. Compuserve, Inc., 776 F. Supp.. 135 (SDNY 1991); Playboy Enterprises v. Frena, 839 F. Supp.. 1552 (MD FL 1993); Stratton Oakmont, Inc. v. Prodigy Services, Inc., ___ NYS2d ___ (May 26, 1995).

7 Sysop’s position, in a sense, is analogous to a defendant in a typical interpleader action under FRCP 22 who has some “asset” under its control and who faces conflicting claims to that asset.

8 Virtual Magistrate FAQ, supra note 2.

9 We might, additionally, ask whether this is an “important” problem in the sense of there being non-trivial numbers of disputes of this kind. Although I know of no specific data on this question, I believe that we are safe in assuming that there are, indeed, large numbers of such disputes; as with all assumptions made in this paper, I welcome discussion of its validity.

10 I ignore a possible additional step, which we might term “internal escalation,” in which Complainant or Actor asks Sysop to reconsider its initial decision, and which consists, essentially, of repeated iteration of steps 1 and 2 (Investigation and Decision) of this process.

11 Of course, it may not always be true that any party can invoke this step. Rules of standing, for example, may limit a party’s ability to invoke the litigation option. In all cases, however, ether Complainant or Actor will be sufficiently injured by Sysop’s decision to be able to state a claim in court seeking relief, and Sysop may, additionally, be able to invoke procedures such as interpleader or declaratory judgment to obtain a court decision regarding the proper course of action.

12 And, if they are risk averse, that margin may have to be substantial. See, generally, the standard economic model of the litigation process presented in e.g., Shavell, Suit, Settlement, and Trial: A Theoretical Analysis Under Alternative Methods for the Allocation of Legal Costs, 11 J. Leg. Stud. 55 (1982); Posner, An Economic Approach to Legal Procedure and Judicial Administration, 2 J. Leg. Stud. 399 (1973).

13 As an aside, I would suggest, notwithstanding the rather extensive publicity that the cases cited in note 6 have engendered, that an interesting feature of cyberspace is precisely how rarely litigation has been invoked in regard to cyberspace disputes.

14 Here, again, I do not suggest that cyberspace is any different than the non-virtual world in this regard. My drycleaner, generally speaking, makes “final” and de facto “binding” determinations regarding complaints of inadequate service because the claims presented to him are rarely, if ever, escalated to court. Drycleaning “law” -- the rules governing the resolution of disputes about drycleaning services -- is thus unlikely to pay much regard to, say, the principles of the Uniform Commercial Code, for unless Complainant has a credible threat to litigate, it is unclear why the drycleaner would look to the set of rules that a court would apply for his own rules of decision.

15 See generally Perritt, Law and the Information Superhighway, pp. 163-170 (1996) (“it is in Internet architectures where intermediaries really flourish . . .”); Johnson & Post, Law and Borders: The Rise of Law in Cyberspace, Stanford Law Rev. (forthcoming, 1996) (discussing the role of intermediary “law-making” n cyberspace); Elkin-Koren, Copyright Law and Social Dialogue on the Information Superhighway: The Case Against Copyright Liability of Bulletin Board Operators, 13 Cardozo Arts & Ent. L.J. 346, 399-401 (1994) (discussing the absence of fixed hierarchies among users and “intermediaries”).

16 (emphasis added)

17 This is a more complex question than I can do justice to in this brief treatment. The generic situation is as follows. Complainant’s interest in having the posting removed will generally be based upon an interest in remaining free from some injury that the posting is allegedly causing, an interest generally protected under tort law. Complainant’s claim against Sysop will then, generally speaking, be in the nature of an action for contributory malfeasance. Actor’s interest (and the larger community’s interest) in retention of the offending posting, on the other hand, is a more generalized interest in the free exchange of information and ideas. If Sysop is a private entity, as will often be the case, neither Actor nor the public can vindicate this interest under the First Amendment. Actor may have a claim against Sysop sounding in contract, based upon whatever contractual agreement is in place governing Actor’s access to the system under Sysop’s control; Sysop can, however, relatively easily guard against any liability based upon such claims (e.g., by tailoring the Terms and Conditions to eliminate liability) and, in any event, Actor’s damages are likely to be minuscule in relation to Complainant’s alleged harm.

18 Sysops of smaller systems are likely to be particularly risk-averse, given their inability to spread the risk of liability over large numbers of users.

19 This can be described in terms of what statisticians refer to as Type I and Type II errors. A Type I error is the designation of a hypothesis as false when it is in fact true. A Type II error is the designation of a hypothesis as true when it is in fact false. In regard to my dispute-resolution model, the “hypothesis” is that Actor’s posting is wrongful (using whatever sources of law Sysop believes is applicable to such a claim). Sysop commits a Type I error when he leaves up a posting that is in fact wrongful (thereby implicitly deciding that the “hypothesis” is false when it is in fact true). Sysop commits a Type II error when he deletes a posting that is not wrongful, i.e., when he decides that the hypothesis is correct when it is not. My point in the text is that Sysop’s decision-making is skewed in that he is far more concerned about Type I errors (because of the liability attached to Type I errors) than Type II errors.

This bias is troubling ex post, i.e., after a Complaint has been delivered to Sysop, in that it suggests that the interests of others affected by Sysop’s decision will not receive due consideration; but it is also troubling ex ante, in the sense that this may systemically lead to an increase in spurious Complaints alleging harm that is in fact not present.

20 See Virtual Magistrate Basic Rules, available online at http://vmag.law.vill.edu:8080/docs/vmagrules.html, which provides, under “Liability,” the following:

“Any complainant and any voluntary participant agree by virtue of their participation that the American Arbitration Association, Cyberspace Law Institute, National Center for Automated Information Research, Villanova Center for Information Law and Policy, and the Virtual Magistrate Program and any of its participants including Magistrates shall not be liable to any party for any act or omission in connection with any arbitration conducted under these procedures.”

21 See id. (setting forth rules under which complaints are to be filed, proceedings held, etc.). The presence of the American Arbitration Association, a well-known and highly reputable provider of dispute resolution services, in the Virtual Magistrate project is surely a useful signal to potential consumers that the Virtual Magistrate procedures, at least, will be well thought out and fair.

22 See Virtual Magistrate FAQ, supra note 2, which notes in regard to the substantive law to be applied only that “The Net already has its own ‘netiquette’ but these rules are unwritten, unclear, and changing along with the Net. The Virtual Magistrate will help to identify and develop rules for the Net on a case-by-case basis. Ultimately, codes of conduct may emerge to provide clearer guidance to all. If people cooperate with the Virtual Magistrate to achieve this goal, then everyone will benefit.”

23 See Bernstein, Understanding the Limits of Court-Connected ADR: A Critique of Federal Court-Annexed Arbitration Programs, 141 U. Pa. L. Rev. 2169, 2244-46 (1993).

24 Id., at 2247.

25 This is not to say that there aren’t close-knit communities in cyberspace -- indeed, it is one of its defining attributes. But arbitration in those communities can successfully proceed in the absence of written rules only to the extent that the arbitrator is himself or herself familiar with, and prepared to apply, the community’s unwritten rules. Query, then, whether the Sysop for such communities does not possess real advantages over outsiders (including the Virtual Magistrate) as a decision-maker in that context.

26 See Bernstein, supra note 23, at 2245 n. 256, describing “wise man” arbitration as follows:

“For example, in transactions between strangers, ‘good faith,’ ‘best efforts,’ and other clauses designed to give the parties flexibility, often have little effect on parties’ actions due to the uncertainty and expense of enforcing them, yet their effectiveness can be enhanced by designating a ‘wise man’ whose opinion both parties respect to decide issues arising under these clauses.”

27 See Virtual Magistrate FAQ, supra note 2:

“The Virtual Magistrate Project is committed to public availability of information about its decisions and activities. The Virtual Magistrate Web site will make information about the operations, rules, and decisions of the Virtual Magistrate Project publicly available. The Villanova Center for Information Law and Policy will maintain a public online repository of Virtual Magistrate complaints, decisions, and documents. The Villanova Center will also maintain electronic discussion groups for magistrates, participants, and other interested parties. The address is http://vmag.law.vill.edu:8080/.

When a complaint is filed, the Virtual Magistrate will not make it public until a decision is reached, hopefully within 72 hours. The parties to a complaint are not prohibited from discussing their participation during this period. All decisions will be made public.”

28 See Bernstein, supra note 23, at 2198. See also Shavell, Alternative Dispute Resolution: An Economic Analysis, 24 J. Leg. Stud. 1 (1995) (discussing the general difficulties of achieving agreement to engage in “ex post ADR.”

29 See Bernstein, supra note 23, at 2240.

Engineering the Virtual Magistrate David G. Post Page -16-