Comments on ICANN Uniform Dispute Policy

A. Michael Froomkin
Professor of Law
University of Miami School of Law
Oct. 13, 1999 [Reformatted into HTML Oct. 14, 1999]

Please note: As an organizer of, a potential Provider under these Rules, I may suffer from a conflict of interest and my comments on matters relating to providers of arbitration services should be discounted appropriately.

A Catalog of Critical Process Failures; Progress on Substance;

More Work Needed

The process that has taken us to this point has been seriously deficient in a number of ways. Each of these failures individually would be sufficient to require that this proposal be re-evaluated. Taken as a group they admit no other possibility. In my opinion, and I do not say this lightly, no one aware of all the facts could claim could claim "consensus" has yet been demonstrated for this policy based on its history to date.

The irony is, however, that although,

  1. there remain very principled reasons to question the appropriateness (and indeed legality under anti-trust principles) of a Uniform Dispute Policy (UDP) that intrudes so deeply into the working of the market in an anti-consumer fashion; and
  2. the current draft does not achieve all the goals set out in ICANN Board Resolution 99.83 at paragraph 3; and
  3. there remain a small number of other critical flaws in the proposed UDP which should be removed before the policy goes into effect;
the current draft is, for all its substantial problems, unquestionably an improvement over its predecessors. Alas, this may say as much about the sad state of the predecessor drafts as about the virtues of the current document.

Process Failures

Working Group A

Working Group A, the ostensible source of the recommendation that has lead to the current proposal suffered from both formal and real failures.

As a formal matter, Working Group A was not properly constituted according to ICANN's own rules. A Working Group requires a representative from each constituency of the DNSO. It did not have one from each.

As a practical matter Working Group A was both manipulated and railroaded. I had both written (email) and oral assurances from the Chair that I would be included in the working group. Somehow, mysteriously, I was never formally included, and thus my dissents (while often referred to in substance) could be somehow be excluded from the claimed consensus. Nor was I not the only person excluded from this group.

The Working Group was further manipulated by being cut up into sub-groups. People were not always allowed access to the sub-group they wished to be in, thus reducing the amount of input and disenfranchising those points of view that the Chair chose to exclude.

The ultimate report was written single-handedly by the Chair, and posted for comment to the full working group for a brief period. It does not in my opinion represent a consensus of the group or of the community at large.

The pNC Endorsement

I have grave doubts as to the validity of the pNC endorsement of the WG-A report. At the time of the endorsement the pNC existed in a rump form, and it is not evident that it was appropriate for it to act on substantive matters. In any event, from the external evidence available (which may not be all the relevant facts), it appears the pNC failed in its duty to scrutinize the existence of a consensus in even the most elementary way. I am not aware of any consultation or investigation by the pNC on this subject. Consensus was not documented in any meaningful or adequate way beyond the then-members of the pNC.

The ICANN Board Resolution

The deficiencies of the Working Group and the pNC might perhaps be regarded as an academic issue, since the ICANN Board's actual action bore only an indirect relationship to what had actually been alleged to be agreed by the Working Group. Resolution 98.3 of the ICANN Board states that the "Registrars' Model Dispute Resolution Policy should be used as a starting point". This was, to me at least, a complete surprise, and I have been participating fairly intensively in this process for some time.

This document was not the product of any working group. It was not the product of the DNSO. It was prepared by one group of partisans, at their own expense, working largely in secret. The only notice that ICANN even contemplated such an action was, I gather, posted somewhere on the ICANN web page approximately seven days before the Santiago Board meeting. I certainly did not notice it at that time, and I doubt seriously whether many other concerned parties noticed it either. Prior to this decision, ICANN had been stating that the Registrars would be encouraged to come to agreement among themselves, as some sort of supplement to the ICANN processes, and to inform ICANN of it. I at least did not understand this to mean that ICANN proposed to use consensus among the Registrars as a substitute for consensus among a more diverse group of stakeholders and I suspect I was not alone in this understanding.

The question of provenance is an important issue because one of the most objectionable features of the current draft has its origins in the Board's resolution in Santiago, and indeed the entire document currently being proposed is materially different from previous proposals. (It also has implications for the legality of the proposal under competition law.) Furthermore, from the time that I was added to the "small drafting committee" part way through its deliberations and continuing until the end of its existence, members of that group were told that comments would only be accepted on the limited matters referred to the committee by the ICANN Board.

The small drafting committee, for all its virtues, was composed entirely of North Americans who communicated erratically with ICANN counsel, himself a US lawyer. I am especially concerned that non-US viewpoints may not have been heard, and also about the effects of some of the proposals on non-US registrants. This is a true concern rather than a conclusion that something is wrong--I do not actually know what effect these rules are going to have in large parts of the world, and I fear no one else involved in the drafting process does either. It may be that the effects of the rules are no worse outside the US than in it; but I think considerable further thought needs to be given to this issue by non-US trained lawyers before they can be approved. (I discuss specifics below.)

Thus, it is critical to note that this current, rather short, fourteen-day comment period represents the only opportunity that most non-registrar stakeholders have ever had to comment on the major features of the actual text and substance of the proposed UDP. I do not consider this to be adequate to determine consensus.

Substantive Issues

General Considerations

My comments on the substance of UDP policy and rules are based on the following assumption, which I take to be shared by ICANN, although not by all commentators: the purpose of the UDP policy is to provide a quick, cheap, just, and generally lightweight process to resolve clear cases of abusive registration. Easy cases are those where facts are either not in dispute, or where there is dispute it is easy to tell where the truth is. (Cyber-arbitrators are not well positioned to make difficult credibility determinations in a paper process.) Easy cases are those where the law is relatively clear, and in most cases those where the relevant law is the same in all jurisdictions relevant to the dispute.

I take it that it is not, and could not be, the objective of the UDP to provide a global trademark court that would be able to resolve every dispute; hard cases belong in a court. This is especially the case since in the most complex matters what constitutes a "legitimate interest" in a domain name may be a difficult question of national law. A major goal of the UDP is thus deterrence of clearly abusive practices. This should benefit trademark holders who will benefit from global jurisdiction, low fees, and a relatively quick process, and (one hopes) will provide some benefit to registrants, although exactly what this in the current proposal is somewhat hard to discern.

Critical Issues

A few major issues cry out for re-assessment or further discussion and explanation.

Competition Law. As a preliminary matter, I believe ICANN owes the relevant stakeholders and the international community in general an explanation as to why it believes that this proposal complies with US anti-trust law and EU competition law principles. It is not in keeping with the nature of this process to put out some set of rules and dare people and governments to sue. All stakeholders have an interest in stability, and stability is not achieved by rules that appear vulnerable to accusations of illegality. Although I am not an anti-trust specialist, in view of the fact that there are already registrars who are competing by offering consumers more favorable terms than those to be mandated by the UDP and that the consequence of this proposal will to require them (on pain of losing the ability to register in the gTLDs) to cease to compete on quality of service, and in view of the fact that the text on which the current document is based was primarily drafted by a lawyer paid by an association of registrars, the case for a claim of conspiracy in restraint of trade between ICANN and the registrars (or abuse of a dominant position by ICANN) would appear to be at the very least to be plausible. It may be that ICANN has commissioned a legal opinion as to why this UDP proposal is in fact legal despite appearances to the contrary. If so, I think at least the broad outlines of that case should be disclosed. If ICANN has not secured such an opinion, I believe it would be irresponsible for ICANN to subject itself and the registrars to the potential liability, and that the potential for this type of lawsuit or enforcement action would do more to create instability in the Internet than almost anything else ICANN could do with the UDP.

"Parity" of Appeal. Perhaps the single greatest improvement in the current draft is that for the first time it creates a mechanism which guarantees that a losing registrant will have some means, however difficult, of securing court review of a decision against him. The fact that this right is now certain for the first time should not blind one, however, to the extraordinarily cramped nature of that right, the very difficult conditions under which it may have to be exercised, and the possible adverse consequences for a registrant.

The process proposed works as follows. After a registrant loses, implementation of a domain name transfer is stayed for ten business days. If the registrant files a complaint in the "Mutual Jurisdiction" in that period the transfer is halted. For all new registrants and for all re-registrants the Mutual Jurisdiction will be the location of the Registrar, not where the registrant is located.

The combination of the short time period to file a complaint, combined with the location of the Mutual Jurisdiction raises a number of issues:

  1. First, the short period to file a complaint means that a registrant will have to hire a lawyer to represent him before the arbitration provider, since it is unlikely to be possible to secure counsel quickly enough once the decision has gone against him. As a result, claims that the UDP policy somehow benefits consumers because it will save them money on lawyers are undermined. This serious unfairness would be ameliorated by extending the "freeze" time period to 30, or even 20 days.
  2. There is a notice issue, especially for re-registrants who are currently registered with NSI. Under US law at present there is no in rem jurisdiction in Virginia over domain names. Will people re-registering their domain names be told of the legal consequences in clear terms? The policy fails to address this, and it potentially disadvantages non-US re-registrants the most as they are the ones who will have the most difficulty finding US counsel in a hurry, and who are the people who potentially will be paying two sets of lawyers, one at home who speaks their language for the arbitration, and then another in Virginia. Somehow I suspect that unless someone requires it to do so, NSI is not going to warn its foreign customers that they might be substantially better of transferring their registrations to a competitor.
  3. Until there is a gTLD registrar in every country, there is a fairness issue in requiring some people to give up their home forum to register a domain name. One hopes this will be a transitory problem, but until it is no longer a problem one could fairly expect an international body like ICANN to use a transitory rule that recognized the fairness issue. The current UDP does not address it.
  4. The most serious potential long-run issue, however, has to do with the consequences of imposing a 10-day filing rule on parties who register with a non-US registrar and whose "Mutual Jurisdiction" is therefore a non-US court. The US and many European jurisdictions have relatively liberal rules of pleading. In the US, for example, an initial complaint (in federal court at least) can be very skimpy, even mildly inaccurate in places, and can be amended once as of right with no or minimal consequences to the plaintiff. A rule that says, rush to the courthouse, file in a hurry and amend at leisure is not unjust in such a regime. Pleading rules were not always so liberal, however; the filing of an inaccurate complaint (and any complaint that needed amendment potentially was "inaccurate"!) was once considered fraud on the court. In some regimes pleadings could not amended, but had to be withdrawn and re-filed (which under the UDP policy would be cause for enforcement of the arbitrator's decision even if the complaint were re-filed the same day); under some of the older forms of action some pleadings could not be withdrawn without prejudice. I personally have no idea how many countries in the world use liberal pleading rules. I would be surprised to learn that they all do. Indeed, I read in Jorge A. Vargas, Mexican Law: A Treatise for Legal Practitioners and International Investors § 20.2 (1998), that in Mexico, "Once drafted, a complaint cannot be refiled." I would not be surprised to learn that many other countries have the same illiberal rule. The UDP puts all nationals of such countries to an unreasonable choice: they must either give up their home forum by registering with a foreign registrar, or do a great deal of advance legal work before they even know the result of the arbitration if they want to have a meaningful chance of appeal. (Note that the same does not apply to a losing challenger, who can wait as long as he likes before trying for a second bite at the legal apple, subject only to laches or the statute of limitations.) This matter requires a degree of study it simply has not received.
Tarnishment. Policy ¶ 4(c)(iii) contains a provision which states that a legitimate noncommercial use of a domain name will nonetheless be denied protection under paragraph 4(c)(iii) if the panel determines that the registrant has an intent to "tarnish" the complainant's trademark. This is wholly inappropriate, undermines the a very substantial part of the free-speech enhancing value of 4(c)(iii), and will have especially pernicious effects outside the US.

Within the US, there is a substantial risk that this language will be used to deny protection to legitimate criticism sites. A site designed to attack a company's labor practices, its environmental record ("MegaCo kills the earth") might be considered to show an intent (recall-it's mere intent, not even actual tarnishment) to tarnish a mark. This would go far beyond US law in protecting trademark holders. The picture is even more disturbing if one considers the various meanings of tarnishment in other legal systems. At various times and places, even gentle criticism of corporations such as comparative price and quality advertisement have been held to be tarnishment. It seems certain that the effect of this language will be to blunt the protection of free speech that noncommercial users are entitled to. The inclusion of this tarnishment language is particularly inappropriate given that it was inserted at the eleventh hour, by either the registrar draft or the ICANN board itself. It does not belong in this document.

Reverse Domain Name Hijacking The current UDP contains no meaningful punishment for reverse domain name hijacking (RDNH) other than the potential slap on the wrist of being chastised in an decision. Nor, other than the aspects of the policy that might make it hard for RDNH to succeed (primarily the defenses listed in Policy 4c), does it contain provisions calculated to deter RDNH. In this respect, therefore, the proposed policy does not even pretend to comply with the Board's Santiago resolution.

Although this is not a small problem, I believe that this lapse will have to be forgiven so long as the other problems with the Policy and the Rules are corrected because I believe this problem cannot be solved without causing worse problems. It is generally agreed that the UDP process should not give the arbitrators authority to levy fines. Yet without this authority there is essentially no sanction which could be applied against a reverse domain name hijacker other than excluding them from any further UDP process-something that might be contractually awkward.

Criticisms of the UDP which seek to further weaken the various protections offered to registrants, e.g. amendments advanced by INTA, should be read with this failure to address RDNH in mind. Although as I explain below those amendments are misguided on their own terms, they are doubly inappropriate given that they would worsen the already one-sided nature of the UDP.

Rules for Computation of Time. The rules for computation of time are extremely hard to follow and, to the extent I can follow them, either unfair or poorly thought out. I think they should be re-written from scratch to flow in a logical order.

We begin with Rule 4(c) which states, "The date of commencement of the administrative proceeding shall be the date on which the complaint is forwarded by the Provider to the Respondent" and Rule 2(f) which states that "All time periods calculated under these Rules shall begin to run on the earliest date that the communication is deemed to have been made in accordance with Paragraph 2(e)."

The key parts of Rule 2(e) state that "Except as otherwise provided in these Rules...all communications provided for under these Rules shall be deemed to have been made.... if via the Internet, on the date that the communication was transmitted, provided that the date of transmission is verifiable." [Note that this refers only to transmittal, not receipt.]

Rule 2(a) does not appear to be a rule "otherwise" providing in terms of rule 2(e) because it does not speak of computation of time (this may be a drafting error?). It speaks only of a duty to achieve actual notice....eventually. Thus, if email to a defendant bounces, time starts to run anyway while the Provider continues to attempt to achieve actual notice by some other means. Similarly if an email is sent but not read (e.g. if the recipient is on vacation), time starts to run anyway.

I cannot say this clearly enough: A policy that starts the clock running when an email is sent, and does not require either ACTUAL notice in advance of starting the running of time or something substantial to show that efforts have been made to achieve notice is immoral. Merely initiating communications via email is not adequate notice. Repeat: Merely sending email is not adequate notice. Running a procedure of this sort without actual notice or things reasonably calculated to achieve notice prior to the commencement of the proceedings must surely be contrary to the public policy of all civilized nations. It will also produce results that are unenforceable in the courts of all civilized nations. It is also unwise. The idea that people go on vacation for two to four weeks and might not check their e-mail is not new to this process, but nothing has been done to take account of the issue.

At a very minimum, Rule 2(e)(iii) should have added to it, "...further provided that the date of receipt must verifiable in the case of the initial communication from the Provider to the Respondent." Alternately, the word "transmission" could be changed to "receipt" (perhaps this is what was intended?).

Other Issues

Policy ¶ 2 (b) ("By applying to register ... you hereby represent and warrant to us that ... to your knowledge, the registration ... will not infringe upon on otherwise violate the rights of any third party.") This language broadly follows the WIPO recommendation. I continue to believe that this language is counter-productive and unreasonable. One might argue that this requirement is pointless since it is not at all clear whether a mere registration of a domain name can, without use, infringe the intellectual property of another. Indeed, in the US, a person cannot violate the Lanham Act or the Dilution Act without commercial use of another's mark. The requirement may seem even more innocuous when one considers that many lawyers I have spoken to take the view that in all but the most obvious cases of unfair competition the only way to know whether a proposed trademark, much less a domain name registration, is an infringement is to see what a competent court says. Whatever one may think of this type of legal realism, it demonstrates that many lawyers will have no fear advising people to register whatever they want and let the courts sort it out later; the rule will therefore stop only the highly scrupulous (and those without legal advice) without in any way daunting the ethically dubious.

I am especially concerned that the words "interfere with" requires too broad a representation by the registrant-one that goes beyond any existing legal requirement. If Alice and Bob both have registered the same trademark in different sectors and/or jurisdictions, and Alice is aware of Bob's registration, how can Alice give a representation in good faith that her registration, which prevents Bob from registering his mark in that gTLD, "the registration of the domain name does not interfere with or infringe the intellectual property rights of another party"? There will be some names, like "united," that in theory no one with a trademark containing the word could register because everyone understands that one person's registration "interferes" with the holder of a trademark in a different sector's ability to register that term. Furthermore If Bob is ignorant of Alice's registration (the proposal creates no duty to investigate), he can make the representation, but there seems no sound reason to reward Bob's ignorance at the expense of an Alice too honest to make the representation.

Policy ¶ 9. Contrary to the first sentence of this paragraph, the Policy should accurately reflect what I understand to be the reality that the Registrar cannot in fact modify the Policy at will. The document should be amended to accurately state that the registrar retains the right to modify the rules when so instructed by ICANN. The language could be patterned on Rules ¶21 which require the express written consent of ICANN.

Rules ¶ 3(b). I am unclear why the rules require complaints to be submitted in hard copy. I would have thought that it might be in everyone's interest to encourage or at least permit all-electronic filings when possible. Furthermore, does time begin when electronic or hard copy is submitted, e.g. for the purposes or Rule 4(a)? Finally, the rules as drafted appear to make it very difficult, perhaps impossible, for a Provider to use a web-based forms approach to structuring filings, although I would have thought that this would be in everyone's interest as it would make the process easier, reduce costs, and minimize accidental filing errors. (A similar question applies to the Response, and to Rules ¶ 5 generally.)

Rules ¶¶ 12 & 13. I am unclear whether and how these paragraphs apply to a Panel that wishes to view a web site for itself. Is that encouraged? Allowed? Discouraged? (I think it should be allowed; but more importantly, whatever the rule is, it should be explicit so that all cases are treated alike.)

Rules ¶¶ 15(e). I can understand the policy reason for word limits on pleadings. I do not really see why one is needed on opinions. Panels are not paid by the word, but by the case, and they will thus have ample incentive to keep decisions short as it is.

Open Issues

I do not have an opinion on absolute word limits for pleadings. Whatever number is picked should be reviewed periodically in light of experience.

I disagree with the ICANN staff report on the selection of Providers and Panelists.

First, I think this is an important issue that requires more discussion than this 14-day comment period allows. This is wholly new material, and there has been absolutely no opportunity to either form or gauge community consensus on this important topic.

As to the substance, whatever procedure is selected should have the following properties:

As a matter of principle I do not believe that ICANN should have any role whatsoever in the selection of arbitration providers. The selection and vetting of arbitrators is in no conceivable form a technical issue relating to either names, numbers, or the functioning of the Internet and is thus outside ICANN's proper jurisdiction. There is nothing in the structural composition of ICANN, nor in the composition of the current or any imaginable future ICANN Board, which suggests any special competence in this area.

Nevertheless, I understand that ICANN intends to take on this function. If it does so, it will presumably seek to ensure that the providers are competent, and will monitor them for partiality. As a result, if ICANN is doing an adequate monitoring job, the danger of forum shopping can be drastically reduced.

I think therefore that if ICANN is going to take a role in approving Providers, a fairly simple and fair system that would ultimately make it easier to take account of national and linguistic differences would be as follows:

ICANN would require every Registrar to choose at least N providers from the ICANN-approved list and offer those providers to potential complainants. The value of N could start small when there are few providers and be adjusted with experience, but it would always be larger than one. Registrars would of course be free to offer more than N if they wished, so long as all the choices were from the ICANN-approved list.

Complainants would choose a Provider from the sub-list of Providers maintained by the registrar who had registered the disputed domain name. The Provider would choose the arbitrators.

I believe that the Registrar is well placed to promote price and service competition since it is the registrar's customers who will have to deal with the Provider, and may be paying half of the Provider's fees. The chance that providers will collude with registrars to provide biased services seems to me to be fairly remote. If I am wrong about this, it should be come evident fairly quickly when a particular provider's decisions get overturned in court more frequently than the others, or when a particular provider's decisions seem more pro-plaintiff or more pro-respondent in a statistically significant manner. If ICANN retains a role in certifying Providers as eligible for the Registrar's list, it can always decertify a rogue Provider.

Requiring Registrars to offer more than one Provider further encourages competition, since complainants will be able to select on the basis of the factors that they care about (price, service, quality, whatever). Respondents lose out on this, since they do not get a choice, but their interests will be protected by the Registrars whose customers they are, and by their choice of Registrars in the first place.

I do not in any way share the ICANN staff report's interest in option (d), which I expect would be an administrative nightmare.

Responses to INTA

I would like to conclude with a few comments relating to the INTA submission because I fear that the extensive citation of irrelevant legal authority contained in that document may lend it an appearance of gravitas that it does not deserve.

The INTA submission is suffused with a fundamental misunderstanding of the purpose of the UDP. The UDP is not designed to enact a trademark agenda in cyberspace that trademark interests have been unable to get from legislatures. Nor, as I suggested above, is the UDP intended (nor should it be intended) to provide an arbitration forum for every trademark related claim pertaining to a domain name registration that might conceivably be litigated or even won in court. Rather, it is and should be designed to provide rapid resolution of easy - obvious - cases, and only those. By providing for quick and cheap resolution of cases of clear abuse, legitimate interests of trademark owners can be protected at significantly reduced cost without infringing the legitimate rights of registrants.

INTA's fundamental confusion is that it wishes to transform a set of rules designed to deal with cybersquatting and grossly unfair competition - cases of clear abuse - into rules that cover all trademark complaints.

Unless otherwise stated below, I oppose all of the INTA suggestions. My comments will focus on those that I fear may sound most plausible, but silence on others should not be read as agreement.

1. 4(a)(iii). INTA proposes that 'and' be changed to 'or' in order, it says, to make warehousing actionable. The reason proposed for this change demonstrates why it should not be adopted. US law, at least, does not create a right of action again mere warehousing of a domain name. Registration of a domain name is not a violation of a trademark. Courts consistently require registration plus something, either a pattern of registration for resale, or some sort of shakedown.

2. 4(b)(i). INTA objects to reimbursement for out of pocket costs beyond those relating to the registration of the domain name itself. This represents the thinking of traditional businesses, and ignores the circumstances of the typical online startup. In an online startup, you start with the domain name, then you print the stationary, the business cards, get the fictitious name, and so on. Unlike the bricks and mortar business, in a legitimate Internet business the online naming is the occasion for a burst of significant expenditure. It is thus proper to allow recoupment of these expenses.

3. 4(c). INTA's comments here are based on its misunderstanding as to the objectives of the procedure. Complex fact intensive cases are by their nature not going to be the sort of clear case appropriate for resolution in the extremely lightweight procedure contemplated in the UDP. Any fact-rich case and especially one where there are complex credibility issues belongs in court or in a more traditional arbitration where parties can be heard directly. The standard of proof, and the conclusive defenses, to which INTA objects are essential in order to ensure that the UDP applies only to the cases of clear abusive conduct for which it is appropriate.

(Incidentally, INTA's example regarding Donald Duck may be poorly chosen. By its own terms, paragraph 4(c)(i) applies only the to BONA FIDE offering of goods and services. It is doubtful that the provision of counterfeit goods would be considered a bona fide offering of goods or services.)

INTA's suggestion that 4(c)(ii) be stricken is especially outrageous. First, it ignores the evident fact that this provision is designed to deal with the "Pokey" problem (see also the problem). Second, it is supported by extensive citation to irrelevant authority. It is correct that under US law one does not have an absolute right to use one's own name for commercial purposes in the face of a pre-existing trademark registration for the same category of goods. Many cases do so hold, and INTA has helpfully cited a heap of them. These cases are however completely inapposite to cases where the registrant is making non-commercial use of his name.

Of course, 4(c)(ii) is not limited to non-commercial uses. Cases where there is commercial use of one's own name or of a name one has used commercially for a long time fall into the category of cases where both parties appear to have a legitimate claim of some sort (e.g. the Prince case) and it is for the courts to determine who if anyone is an infringer. Given that the parties to the dispute may be in different jurisdictions, these will often be complex matters ill-suited for arbitration. The point is that whatever these cases are, even when they are real disputes based on meritorious claims, they are not likely to be "abusive registrations". Not every trademark dispute is a cybersquatting case, and 4(c)(ii) properly aids in restricting the reach of the UDP to the more narrow category.