Berkman Center for Internet & Society
Change Password
Forgot Password
Feedback (not active)

Process Bad Faith Rights Choice of Law Overview

Overview of Domain Name Policy Development

by Diane Cabell

Conflicts between Trademark Law and Domain Names

Domain names were originally for the sole purpose of functioning as an address, like a street name. With the globalization and commercialization of the Internet, domain names have developed important significance as business identifiers.  They are now highly visible in "real space" (known as “meatspace”) as well.  Who hasn't seen or heard a "dot com" commercial on radio, television, in magazines and other media?  In this new role, they frequently conflict with trademarks and other traditional business identifiers.  Two factors contribute to this conflict.

First, domain names are global and must be unique - a particular string of characters may only link to one site.Trademarks, however, may overlap in different industries or different geographical locations.  Increasingly, though, cyberspace is being treated as a single market that cannot be subdivided into comparable regions.

Second, it is common practice for many Internet users to guess at domain names.  Thus domain names based on intuition become valuable corporate assets. The more commonplace the name, the more valuable the asset. Trademark law, on the other hand, severely restricts protection for generic words. Disputes between trademark owners and domain name holders are exacerbated by these differences.

* * *

Purpose and Function of the Domain Name System (DNS)

Purpose and Function of Domain Name System

The Domain Name System was designed in the 1980s to deal with the problem of growth.   Data flowing over the Internet that is intended for "Jane User" is actually addressed to a long number, known as an IP address.  These numbers must be unique so the routers (machines that forward Internet data from node to node) can properly target a destination.

There are many kinds of records that can be "mapped" (correlated) to an IP address and IP addresses can change frequently, for example when a user changes Internet access providers.   Imagine the convenience of having the telephone, fax and mobile phone numbers of each individual listed in one location under that individual's name.  In an electronic database, these numbers can be changed or added as needed, but the file name remains constant so one can always find all of the records for that person.  This is the reason that "names" were originally adopted for use on the Internet. Of course, they are also user-friendly and easy to remember.

Each name must be a unique "string" of characters; otherwise duplicate addresses would disrupt Internet traffic. RFC 920, RFC 1101, RFC 1034, and RFC 1035.To transfer data to a particular name, a telephone-book like directory is required so that routers could look up the IP address currently assigned to that name.  This list, in electronic form, had to be updated constantly.  As the list of users got longer and longer, routing speed slowed proportionally because it took longer to scan the entire list to find the correct IP address.  Imagine having to reissue the White Pages of the New York telephone book every time a new customer got a telephone.

The technical solution to this problem was the Domain Name System. Instead of a single list, a Yellow-Pages approach was designed, breaking the content into categories, called “domains.”  Domains are simply lists of lists, nested within each other. The URL for the Berkman Center home page, for example is “Cyber” appears on the list for the 3rd level domain ".law" which on the list for the 2nd level domain ".harvard", which is listed in the Top Level Domain (TLD) ".edu."   Updates in .law list are made locally by the law school.  Harvard University decides whether or not to add .med to the .harvard list, and the .edu administrators determine who appears in the category for colleges and universities.  Thus, changes are made on a discrete level. The .edu list does not have to be changed just because Harvard adds a new department; only the .harvard domain need be amended. Routers only need to find the .harvard domain to direct traffic to the Berkman Center. Additional routers direct traffic within each domain.

At the top level, the list is small. There are only 7 generic TLDs (gTLDs) and approximately 145 country code TLDs (ccTLDs). This very small database is called the Root. If one deleted the entry for .com (a gTLD), then all of the lower level lists nested within that domain would disappear with it. No one could reach a "dot com" address unless she used the specific IP address number. Authority over the Root is at the heart of the Domain Name System.

Understanding how the DNS works leads to the next question, “How has the DNS been administered?”

History of Domain Name Administration

For most of the history of the Internet, online use was limited primarily to educational and research institutions. The domain name directory was maintained by Jon Postel, an academic who worked under a research contract from the United States government. Dr. Postel operated through an entity known as the Internet Assigned Numbers Authority (IANA). The registration policy was "first come, first served" with no review of rights. As the work load of entering and deleting names became heavier, the task was contracted out to a private corporation, now known as Network Solutions, Inc. (NSI).

NSI continued Dr. Postel's policy but almost as soon as the Net was opened to commercial traffic, trademark disputes arose. The most egregious cases in these early days involved "cybersquatting" and "cyberpiracy." Cybersquatting generally referred to an individual who registered domain names identical to well-known marks in an attempt to resell it to the mark owner at a significant profit. See Panavision International, L.P. v. Toeppen.Cyberpirates are those who register the same or confusingly similar domain name in an attempt to lure traffic intended for the mark owner away to pirate's web site. In the absence of legislative definition, these terms are used interchangeably. Frequently unable to obtain service on remote infringers, or to identify the actual domain holder due to false registration data, trademark owners brought contributory infringement actions against NSI for wrongfully granting domain registrations to those without rights to use the mark.

Although NSI routinely avoided liability in these contests, it began revising the registration contract signed by domain holders to impose a domain dispute policy that reflected these issues. This dispute policy was revised with some frequency. By 1999 it had several characteristics that were found unsatisfactory by both domain holders and trademark owners. It could only be used by owners of federal or national marks (ignoring design marks, common law marks, state registrations and foreign personal name rights). It could only be utilized against a domain name that was identical to the mark unlike the broader standard used in trademark law where infringement is found if the two names are "confusingly similar." And it did not reflect principles of fair use such as free speech or product criticism. NSI Flawed Domain Name Policy.

The Clinton administration also recognized that the Internet had become such a broadly distributed network, with ownership of its constituent parts so widely scattered and subject the laws of so many jurisdictions, that it was no longer appropriate for one single government to be coordinating its functions and decided to withdraw from the role of technical coordinator to permit the private sector to assume this responsibility. There was also intense pressure to break up NSI’s monopoly position in the lucrative com-org-net market. In November 1998, the Department Of Commerce entered into a Memorandum of Understanding (MoU) with a private corporation established by Dr. Postel called the Internet Corporation for Assigned Names & Numbers giving ICANN oversight responsibility for domain names, numbers, protocol parameters and the Root.ICANN has two years (ending September 2000) to develop methods and procedures for technical coordination in a manner that will ensure bottom-up coordination, market competition, global representation and stable operations.

Introducing competition, however, meant the possibility of conflicting domain dispute policies, so the DOC contract specifically required that ICANN make recommendations regarding trademark/domain name policies and particularly those proposed by the World Intellectual Property Organization (WIPO) concerning: (i) the development of a uniform approach to resolving trademark/domain name disputes involving cyberpiracy; (ii) a process for protecting famous trademarks in the generic top level domains; (iii) the effects of adding new gTLDs and related dispute resolution procedures on trademark and intellectual property holders; and recommendations made by other independent organizations concerning trademark/domain name issues. [MoU, Section V.C.9.d]

ICANN’S Dispute Policy Development

ICANN’s policy on domain disputes evolved fairly rapidly. The WIPO proposals, which had been circulating for several years, were presented to ICANN in May 1999.They included the establishment of a uniform, mandatory domain name dispute policy for resolution of cybersquatting disputes and a system of panels to arbitrate them. Many elements were adopted in a model agreement proposed for voluntary adoption by the half-dozen newly-accredited registrars and NSI.The models were presented for public comment during the summer whichreported them to be unfairly burdensome on non-commercial and fair users, so the Board insisted on the following amendments:

a. In determining whether a domain name has been registered in bad faith, consideration should be given to the following factors in addition to those stated in the WIPO, DNSO, and registrars’ policies:

(1) whether the domain name holder is making a legitimate noncommercial or fair use of the mark, without intent to misleadingly divert consumers for commercial gain or to tarnish the mark;
(2) whether the domain name holder (including individuals, businesses, and other organizations) is commonly known by the domain name, even if the holder has acquired no trademark or service mark rights; and
(3) whether, in seeking payment for transfer of the domain name, the domain name holder has limited its request for payment to its out-of-pocket costs.

b. There should be a general parity between the appeal rights of complainants and domain name holders.
c. The dispute policy should seek to define and minimize reverse domain name hijacking.

The models were then subjected to redrafting by staff and a small number of DNSO activists.  The resulting ICANN proposals, along with a Staff Report, were posted in September for 2 weeks of public comment.  The final versions, Uniform Domain Name Dispute Resolution Policy (the Policy) and Rules for Uniform Domain Name Dispute Resolution Policy (the Rules), were accepted by ICANN at its Annual Meeting in Los Angeles in November.  The UDRP became effective for all com-org-net registrars by January 2000.See ICANN, Uniform Domain-Name Dispute-Resolution Policy for a list of ICANN's historical and current policy documents.

ICANN still has much work ahead in regard to trademark issues.In July of 2000, the Board is expected to take action on the questions of famous marks (whether they should receive special registration protection) and the addition of new gTLDS (whether they should be chartered for special purposes and how they should be administered). The corporate by-laws and external agreements with the registry require that domain name policy decisions be made through a bottom-up consensus process, going first to ICANN's Domain Name Supporting Organization for review and recommendation. DNSO Working Groups are open to participation by the public. See ICANN By-laws. To follow the development of these issues, see ICANN, Consideration of New Generic Top-Level Domains, and scroll down to the subscription links.

See The Process for a complete review of the UDRP procedures.

* * *

Comparison of Trademark Law to the UDRP

The principles of trademark and name rights are well-established and benefit from a long history of judicial interpretation. The UDRP is new; its terms are not well-defined and their interpretation is not necessarily a matter of stare decisis.

Trademark law allows many different merchants to use the same word on their goods or services as long as the consumer isn’t likely to be confused as to the source or origin of these items. Confusion may not exist where there are different marketing territories, different types of products, or different label appearance. A domain name is less flexible. It crosses all geographic boundaries, regardless of the domain holder's intent to limit sales to a local territory. It is text-based and has no other attributes (such as a logo design) that can help distinguish it.

Trademark law is territorial, it differs from jurisdiction-to-jurisdiction; the UDRP is intended for global application and does not specify which local law, if any, should be applied.

Trademark law applies to commercial enterprise, whereas the bulk of Internet use is not-for-profit. Trademark rights are enforced by the majesty of government; the UDRP is enforced by contractual agreement.

Trademark registration practices in many jurisdictions require complex searches to determine possible conflicts. The ease of domain registration and web site posting has permitted thousands of people to publish online; few of them realize the need to consult trademark counsel before doing so. Conflicting opinions are being issued by UDRP panelists, but unlike judicial systems, there is no single court of final review that can resolve differences among the dispute resolution providers.

The Uniform Domain-Name Dispute-Resolution Policy was designed to provide a quick and relatively inexpensive procedure that could be used by mark owners around the world against those who register domain names in bad faith and without rights or legitimate interest. Trademark litigation, on the other hand, is a slow, expensive, deliberative process requiring rigorous standards of proof with application limited to the local jurisdiction.

Lastly, the burdens of proof are quite different. Under the UDRP, five elements must be established:

(1) that the Complainant has trademark or service mark rights in the name,
(2) the domain name is identical or confusingly similar to the mark,
(3) that there was both registration and use of the domain name,
(4) such registration and use was in bad faith, and
(5) the registration and use was without rights or legitimate interest.

See the Sections on“Bad Faith and Rights or Legitimate Interest for further discussion of those terms.

In trademark law, the most common elements are

(1) the existence of senior trademark or service mark rights,
(2) commercial use of the name, and
(3) that the use is likely to confuse consumers as to the source or origin of the goods or services being offered by the infringer.

Trademark may find either the use or the registration of a name to be infringing. The UDRP requires both.

It is hoped that many of these problems will be resolved by the Internet community as experience with the new system reveals its shortcomings. One of the many benefits of the UDRP is the willingness of its administrator, the Internet Corporation for Assigned Names and Numbers (ICANN), to hold open fora where such complaints can be voiced, and where working groups can be formed to advise on improvements.

* * *

Limitations of the UDRP

There is no question that the UDRP provides a benefit to trademark owners and domain holders in foreign jurisdictions who wish to assert their rights to domain names registered in the US. They will no longer have to hire expensive foreign counsel to lead them through our legal system. Rights under local law have a better chance of being properly recognized if the panelist is from the same region. However, there are many unresolved issues in this new process that may cause some serious growing pains.

One shortcomings is that the UDRP does not apply to approximately 150 ccTLDs (.uk, .nz, etc.). Although 80% of registrations are currently made in the applicable com-org-net domains, that may well change as Internet use increases in all parts of the globe. Cybersquatters are likely to seek other ports from which to launch their unfair practices.

The UDRP is also limited in application to cybersquatters, as defined in Section 4 of the Policy. Traditional trademark infringement actions looked at whether there was a likelihood of confusion in the minds of consumers as to the source or origin of particular goods or services. Under the UDRP, the standards are the domain holder's "bad faith" and absence of "rights or legitimate interests." See Bad Faith and Rights and Legitimate Interest for discussion of those terms. The legislative history indicates a clear intent that it not become a substitute for the courts for resolving routine cases of trademark infringement. Nonetheless, cases are heard even where both of the parties are engaged in legitimate commercial enterprise. Shelley Harrison v. Coopers Consulting, Inc. (, FA-0121; Fiber-Shield Industries, Inc. v. Fiber Shield LTD (, FA0092054.

The leeway granted to the dispute panelists to interpret both the language of the UDRP and the applicable law is apt to lead to inconsistent conclusions, particularly where there is no guideline on choice of law. In Compagnie de Saint Gobain v. Com-Union Corp. (, D2000-0020, there were contacts with France (the Complainant), the US (Respondents) and Mexico (the host server) but the Panelist does not specifically identify any locality as controlling. See the Section on Choice of Law. Those who are familiar with trademark law may apply it usefully, but those who are not may impose principles that are not normally associated with trademark rights. In one case, bad faith was presumed in the absence of any use at all because no legitimate use could be conceived by the Panelist. Telstra Corporation Limited v. Nuclear Marshmallows (, D2000-0003. The case involved Australian parties, and perhaps Australian law does not recognize product criticism or fair use rights, but the opinion does not reveal if this is the reason. There is a real possibility that trademark rights will be expanded by the way "bad faith" and "rights and legitimate interests" are defined in these proceedings, which was not at all the intention of the policy drafters. In attempting to use language flexible enough for global utility, the diverse, but unique name rights in each of multiple jurisdictions may now be bundled into one package of hyper-rights available exclusively in cyberspace.

There is no evidentiary review in the UDRP. Facts, such as trademark rights, are simply alleged. They may be supported by attached exhibits (photocopies of trademark registration certificates, copies of web page results from online searches of the USPTO trademark database, copies of advertisements, product labels or letterhead, etc.) However, none of these documents are required to be authenticated. The role of publication may be uniquely critical in this regard. Panelists are already citing prior opinions as reference:

"Although entitled to consider principles of law deemed applicable, the Panel finds it unnecessary to do so in any depth. The jurisprudence which is being rapidly developed by a wide variety of Panelists worldwide under the ICANN Policy provides a fruitful source of precedent." 3636275 Canada, dba eResolution v. (, D2000-0110.

See also Telstra, supra and Ellenbogen v. Pearson (, D2000-0001. But without access to the pleadings, or careful pronouncement of the jurisdictional law being applied, this seems a dangerous process. In a court of law, the proof of claims is based on a rigorous set of standards both as to admissibility and to relevance. This is not the case in the UDRP which was intended to be a fast and relatively inexpensive process. In contrast to the decisions of administrative proceedings under the UDRP, arbitration procedures are generally not published. See the Section on Precedent.

The procedure may only be initiated by trademark owners. There is no corresponding ability of the domain holder to bring the dispute before a provider at his own convenience and before his preferred Provider. A respondent is disproportionately disadvantaged in selecting 3-member panels: the complainant has already had its choice of the host provider and thus the choice of the panel president. The complainant will also get to nominate 3 candidates for a second member of the panel. The respondent may also recommend 3 candidates for the third panelist, but has no way of determining whether or not her nominees are either available or eligible to serve. If none of them can stand, then the choice defaults to the host provider which was selected in the first place by the complainant. To the extent that clear differences among the Providers emerge, if they do, then gaming of these choices will proliferate.

Sometimes there is not enough comment in the opinion to determine the specific basis for the decision. Cunard Line Limited v. Champion Travel, Inc. (, NAF FA0092052. The disputants are also prohibited by virtue of the Complaint and registration contracts from suing any of the parties responsible for the dispute procedure. An action in equity, or an appeal for review by ICANN, may be the only (and expensive) remedy for poor performance or failure to follow procedures on the part of individual Panelists.

Other benefits expected from the UDRP are speed and lower cost. The process may be quick (less than 60 days) but injunctive relief from a court may be even quicker if the potential harm is great. The relatively low cost of the UDRP is likely to be an advantage to most domain holders and mark owners. However, the 10-day limit on an appeal to a court of mutual jurisdiction for review really necessitates having counsel involved from the beginning. Finding an attorney to bring such a suit, and preparing for it on such short notice, is going to be difficult for a Respondent, especially one who is not using the domain for commercial purposes and may not have revenue sufficient to justify protecting a $35 domain registration.

Clients who are facing multiple cybersquatters may find it much less expensive to file in a US court. Under the new US Anti-cybersquatting Consumer Protection Act, a single proceeding in rem in Herndon, Virginia can consolidate all possible defendants in one case, including those in the ccTLDs not covered by the UDRP. The UDRP only allows consolidation against one Respondent. Furthermore, the UDRP requires allegation of both registration and use in bad faith and may be available as a remedy in fewer situations than the ACPA which permits action under registration or use.

There appear to be substantial dangers lurking in the shoals of the UDRP procedures. Scholars of legal history have already noted that the UDRP may soon follow the path of other, earlier attempts to provide user-friendly alternatives to complex court processes. Arbitration, for example, is becoming an increasingly formalized activity, and the forerunners of the alternative dispute resolution, the courts of equity, have long since evolved administrative procedures so arcane that they have simply been merged into courts of law.

For a review and analysis of the opinions issued from January through March, see M. Scott Donahey, The ICANN Cybersquatting Decisions.

Wendy Seltzer, a Berkman Center Fellow, has created a keyword search engine for all published UDRP opinions.

Last updated on 20 April 2000

Please send all inquiries to:

Chris Babbitt
Head Teaching Fellow



Berkman Center for Internet & Society