Electronic copy available at: http://ssrn.com/abstract=1887888
65
AN EMPIRICAL ANALYSIS OF FAIR USE
DECISIONS UNDER THE UNIFORM
DOMAIN-NAME DISPUTE-RESOLUTION
POLICY
David A. Simon*
Abstract: Since its inception in 1999, the World Intellectual Property Organization
(WIPO) has resolved nearly 22,000 domain-name disputes under
the Uniform Domain-Name Dispute-Resolution Policy (UDRP)—a
mandatory arbitration policy implemented by the Internet Corporation for
Assigned Names and Numbers (ICANN). The UDRP allows the holder of a
legally protectable trademark to initiate proceedings to cancel a similar
domain name or have it transferred to the trademark owner. Domain-name
holders, though, have a number of defenses, including that they are using
their domain names in a noncommercial, fair manner. Although several
empirical studies have analyzed various aspects of the UDRP, none have
specifically examined this fair use defense. This study does what others have
not. It analyzes the fair use defense in decisions before WIPO. Using
WIPO’s online decision database, this study makes two important findings.
First, respondents from the United States are more likely than those from
other countries to succeed on a fair use defense. Second, arbitrators from
the United States are more likely than those from other countries to find
that a respondent’s use of a domain name was fair. This means that, under
the UDRP, respondents from the United States enjoy greater speech protections
than those from other countries, and that arbitrators from the
United States are more sympathetic to speech interests than arbitrators
from other countries. To improve the UDRP, I propose two revisions. First,
ICANN should adopt a choice of law provision stating that the law of the
respondent’s home country governs fair use disputes. Second, ICANN
should implement a panel assignment provision in fair use disputes that
requires arbitrators to share the nationalities of the litigants.
* © 2012, David A. Simon, Fellow, the Project on Law & Mind Sciences, Harvard Law
School; Ph.D. candidate, University of Cambridge; LL.M., Harvard Law School; J.D., Chicago-Kent
College of Law; B.A., University of Michigan. For helpful comments, suggestions,
and criticisms, I thank Graeme Dinwoodie, Konstantinos Komaitis, Edward Lee,
Jacqueline Lipton, Cedric Manara, Lynn LoPucki, Jason Luliano, Mark McKenna, Parina
Patel, Katherine Porter, Betsy Rosenblatt, Alan Simon, Michal Shur-Ofry, the participants
in the 2011 Intellectual Property Scholars Conference at DePaul University College of Law,
and the participants in the 2010 Empirical Analysis of Law Workshop at Harvard Law
School. Thanks also to the staff of the Boston College Law Review for their diligent work.
Electronic copy available at: http://ssrn.com/abstract=1887888
66 Boston College Law Review [Vol. 53:65
Introduction
Because domain names are valuable commodities,1 trademark
owners attempt to control domain names that incorporate their trademarks.
Sometimes they do this by preemptively “buying up” domain
names that disparage their companies or executives, or direct users to
websites that do so.2 But when another individual already owns such a
domain name, the trademark owner must try to wrestle it away from
the domain name owner.3 These are disputes over private censorship.
They raise questions about what people on the Internet can say and
how they can say it.
Although these disputes are amenable to traditional litigation,
trademark owners frequently pursue a quicker, cheaper option:4 they
file complaints under the Uniform Domain-Name Dispute-Resolution
Policy (UDRP)5 with a domain-name arbitration “provider”6 such as the
World Intellectual Property Organization (WIPO).7 Under the UDRP,
1 See infra notes 45–88 and accompanying text.
2 Bank of America recently bought domain names incorporating the name of its Chief
Executive Officer, Brian Moynihan, including <BrianMoynihanBlows.com>. Hayley Tsukayama,
Bank of America Buys Up Anti-BoA Domain Names, Post on Faster Forward by Rob Pegoraro,
Wash. Post (Dec. 23, 2010, 9:55 AM), http://voices.washingtonpost.com/fasterforward/
2010/12/bank_of_america_buys_up_anti-b.html.
3 See e.g., Gail Guarulhos Indústria e Comércio Ltda. v. Watson, No. D2006-0655 (WIPO
July 7, 2006), http://www.wipo.int/amc/en/domains/decisions/html/2006/d2006-0655.
html (finding fair use in a case in which the complainant’s business name included the word
“gail” and the respondent registered <gail.com> because his wife’s given name was Gail);
Wal-Mart Stores, Inc. v. MacLeod, No. D2000-0662 (WIPO Sept. 19, 2000), http://www.wipo.
int/amc/en/domains/decisions/html/2000/d2000-0662.html (relating that respondent who
registered <wal-martsucks.com> to criticize Wal-Mart lost UDRP dispute).
4 See infra notes 45–87 and accompanying text.
5 Uniform Domain Name Dispute Resolution Policy, Internet Corp. for Assigned Names
& Nos., http://www.icann.org/dndr/udrp/policy.htm (last modified Sept. 6, 2011) [hereinafter
UDRP]. The UDRP also contains rules of procedure. Rules for Uniform Domain
Name Dispute Resolution Policy (the “Rules”), Internet Corp. for Assigned Names & Nos.,
http://www.icann.org/dndr/udrp/uniform-rules.htm (last modified Sept. 6, 2011) [hereinafter
UDRP Rules].
6 There are four approved providers currently operating. In addition to WIPO, the
three other providers are the Asian Domain Name Dispute Resolution Centre, the National
Arbitration Forum, and the Czech Arbitration Court Arbitration Center for Internet
Disputes. List of Approved Dispute Resolution Providers, Internet Corp. for Assigned Names
& Nos., http://www.icann.org/en/dndr/udrp/approved-providers.htm (last modified
June 9, 2011) [hereinafter List of Providers]. Since providers started hearing UDRP disputes,
two have gone defunct. These providers are CPR: International Institute for Conflict
Prevention and Resolution and eResolution. Compare List of Providers, supra (listing current
and former providers), with UDRP Providers, UDRPinfo.com, http://www.udrpinfo.com/
prov.php?b=22 (last visited Dec. 20, 2011) (listing providers as of 2002).
7 WIPO is an agency of the United Nations. What Is WIPO?, World Intellectual
Prop. Org., http://www.wipo.int/about-wipo/en/what_is_wipo.html (last visited Dec. 20,
2012] An Empirical Analysis of Fair Use Decisions 67
an alleged mark owner (“the complainant”)8 can attempt to cancel the
domain name or transfer it from its owner (“the respondent”)9 to the
trademark holder.10 After filing, a panel of either one or three arbitrators
is chosen to resolve the dispute.11 A respondent, however, can assert
certain defenses by showing that she has “rights or legitimate interests”
in the domain name.12 One of these defenses allows the
respondent to “make a legitimate noncommercial or fair use of the
domain name, without intent for commercial gain to misleadingly divert
consumers or to tarnish the trademark or service mark at issue.”13
This defense—which I call “the fair use defense” —is a safety-valve for
sometimes critical speech.14
But how, exactly, does the defense work in practice? More specifically,
do panels apply the fair use provision uniformly to all respondents?
This Article presents empirical evidence that panels—U.S. panels
in particular—apply the fair use defense more favorably to U.S.
respondents than to other respondents. My analysis of nearly one thousand
UDRP decisions rendered by WIPO between 1999 and 2010 sup-
2011). Created in 1967 by the United Nations member states, WIPO’s mandate is to “promote
the protection of [Intellectual Property] throughout the world.” Id. It is one of four
currently operating UDRP providers. See List of Providers, supra note 6.
8 The UDRP defines complainant as “the party initiating a complaint concerning a
domain-name registration.” UDRP Rules, supra note 5, § 1.
9 The UDRP defines respondent as “the holder of a domain name registration against
which a complaint is initiated.” Id. § 1.
10 Id. § 3 (noting that upon receipt of a UDRP arbitration, the provider will cancel,
transfer, or change the domain name).
11 Arbitrators are drawn from an international pool of various ICANN- and WIPOapproved
individuals. See WIPO Domain Name Panelists, World Intellectual Prop. Org.,
http://www.wipo.int/amc/en/domains/panel/panelists.html (last visited Dec. 20, 2011).
Thus, arbitrators of various nationalities can and do sit on the same panel. When the majority
of arbitrators on a three-member panel are from the United States, I refer to the
panel as a “U.S. panel.” The same is true for a one-member panel with an arbitrator from
the United States. When, however, the majority of arbitrators on a panel are from a country
other than the United States, I refer to the panel as “non-U.S.,” “other,” or “foreign.”
The same title applies to single-member panels of non-U.S. arbitrators.
12 Compare UDRP, supra note 5, § 4(a)(ii) (stating as an element of the complaint that
the complainant must show that the respondent “ha[s] no rights or legitimate interests in
respect of the domain name . . . .”), with id. § 4(c) (“Any of the following circumstances, in
particular but without limitation, if found by the Panel to be proved based on its evaluation
of all evidence presented, shall demonstrate your rights or legitimate interests to the
domain name for purposes of Paragraph 4(a)(ii) . . . .”).
13 Id. § 4(c)(iii).
14 See, e.g., Michael Froomkin, ICANN’s “Uniform Dispute Resolution Policy” —Causes and
(Partial) Cures, 67 Brook. L. Rev. 605, 642–43 (2002) (arguing that WIPO’s “rights and
legitimate interests” defense seemed to quell initial worry about whether criticism, parody,
and other speech interests would be protected by the UDRP).
68 Boston College Law Review [Vol. 53:65
ports this conclusion. The purpose of my analysis was to determine
whether the nationality of respondents or panels influenced the outcome
of fair use cases. First, I analyzed whether respondents from the
United States won more often than those from other countries. The
data show that U.S. respondents won 35% of the time while other respondents
won only 17% of the time. Second, I analyzed whether U.S.
panels were more likely to find fair use than non-U.S. panels. My data
show that U.S. panels found fair use 36% of the time. Other panels, by
contrast, found fair use only 20% of the time. These data suggest that
WIPO panels apply the UDRP fair use defense more favorably to U.S.
respondents than to other respondents.
Although a variety of factors likely are responsible for these phenomena,15
I argue they occurred for two basic reasons. First, the UDRP
contains a vague choice of law provision. This vagueness allowed WIPO
panels to apply whatever law they “deem[ed] applicable.”16 U.S. panels
then took greater advantage of this provision than their foreign counterparts.
In other words, U.S. panels began importing U.S. law—law
that is generally (viewed as) more sensitive than the law of other countries
to free speech interests—into UDRP disputes with higher frequency
than did foreign panels. More than non-U.S. panels, U.S. panels also
applied U.S. law in cases where the respondent was from the United
States. Because U.S. law is (viewed as) more friendly to speech interests
than foreign law, the use of U.S. law favored respondents.
But why did U.S. panels—and not foreign panels—apply their
home country’s law? There are five potential reasons. First, because
U.S. panels decide nearly fifty percent of all fair use cases, they have
more opportunities than their foreign counterparts to import U.S. laws.
Second, more than U.S. arbitrators, other arbitrators believe that the
UDRP should be a uniform policy. To preserve uniformity, other arbitrators
do not apply other countries’ local laws—be it the arbitrators’,
complainants’, or respondents’ law(s)—to decide UDRP disputes.
Third, other arbitrators do not think about what law to apply. Because
they view the UDRP as precedential, they simply apply prior decisions.
The UDRP contains a U.S./non-U.S. dichotomy in terms of rules about
fair use. With two rules to choose from, arbitrators did not invent new
ones. Fourth, much more than U.S. panels, foreign panels decide cases
15 I did not run a regression using my data. Therefore, my data are descriptive. My argument
is based only on these descriptive data. I do not purport to account for other factors
that could be influencing the findings; indeed, there may be other factors that are
influencing the data. I leave that statistical analysis for another study.
16 UDRP Rules, supra note 5, § 15(a).
2012] An Empirical Analysis of Fair Use Decisions 69
involving parties of nationalities different than the (foreign) panel. Finally,
many countries do not have cybersquatting laws, or trademark
laws with as many speech protections as U.S. laws. Without law—or law
that is as speech-friendly—to apply, panels from other countries simply
rely on the language of the UDRP (which may better reflect their own
countries’ laws). That is, they interpret the UDRP as a self-contained
document, rather than import local legal rules or principles.
The importation of U.S. law has practical consequences. As different
panels applied different rules, two competing sets of rules arose in
fair use cases. The first set adheres to U.S. laws. The second set is comprised
of sui generis UDRP rules developed by WIPO panels (“WIPO
Rules”). These rules then replicated as panels cited them in subsequent
fair use cases. The practical effect was twofold. First, because the imported
U.S. law allowed individuals greater leeway to use trademarks in
domain names than did WIPO Rules, the rule-dichotomy privileged cases
in which panels applied U.S. law. Second, this same “legal leeway”
gave U.S. respondents a greater ability to “speak” than other respondents.
In other words, U.S. respondents could use others’ trademarks in
their domain names in more situations than other respondents because
arbitrators deciding cases with U.S. respondents usually applied U.S. law.
Although fair use is an important defense to protect speech interests,
most empirical studies of UDRP cases do not focus on how panels
interpret specific provisions. Instead they examine broader issues, such
as the UDRP’s fairness or its potential to suppress speech generally.
Some studies, for example, focus on respondents’ overall success rate in
UDRP actions. One study found that the UDRP disfavors speech by
those who do not hold trademarks—that is, it disfavors respondents.17
The study notes that the UDRP, by giving complainants the ability to
select an arbitration provider, encouraged forum shopping.18 To sup-
17 Milton Mueller, Rough Justice: A Statistical Assessment of ICANN’s Uniform Dispute Resolution
Policy, 17 Info. Soc’y 151, 156 (2001) (noting that, in 2001, complainants won around
eighty percent of disputes); see also Milton Mueller, Ctr. for Convergence & Emerging
Network Techs., Success by Default: A New Profile of Domain Name Trademark Disputes
Under ICANN’s UDRP 1, 15 (2002), available at http://dcc.syr.edu/PDF/marklereport-final.pdf
(explaining that complainants win ninety-six percent of the time when a
respondent defaults, admitting that sometimes default may indicate cybersquatting, but concluding
that many times defaulting respondents have valid defenses).
18 Mueller, supra note 17, at 161 (“After only 1 year of operation, there is a statistically
significant correlation between market share and the tendency to take away domain names
from respondents.”). But see Ned Branthover, Int’l Trademark Ass’n, UDRP—A Success
Story: A Rebuttal to the Analysis and Conclusions of Professor Milton
Mueller in “Rough Justice” 1, 4–7 (2002), available at http://www.inta.org/Advocacy
70 Boston College Law Review [Vol. 53:65
port this claim, the study pointed to data showing that the providers
with the biggest market shares were also those with the highest complainant
win rates.19
Some were not convinced by this hypothesis. One study, for instance,
argued that provider selection is mostly a matter of provider
efficiency rather than provider bias.20 In other words, the complainants
choose providers that decide disputes the fastest.21 Results of that study
showed that provider efficiency had been overlooked as a factor by previous
studies.22
Further studies revealed other factors that helped explain outcomes
under the UDRP. One author, for example, used data on UDRP
decisions to analyze the influence of panel size on outcomes (i.e., wins
or losses).23 Under the UDRP, the complainant can elect to have the
dispute heard by a one- or three-member panel.24 When the complai/Documents/INTAUDRPSuccesscontraMueller.pdf
(attempting to rebut Mueller’s analy-
sis).
19 Mueller, supra note 17, at 161.
20 Jay P. Kesan & Andres A. Gallo, The Market for Private Dispute Resolution Services—An
Empirical Re-assessment of ICANN-UDRP Performance, 11 Mich. Telecomm. Tech. L. Rev. 285,
295–96 (2005) (“Using a multinomial logit regression model to determine if complainants
select the provider based on bias or the duration of the procedure, we show that duration
is at least as important as bias in selecting providers.”).
21 See id.
22 Still others have examined how cases are distributed among panels and providers.
John Selby, Competitive Provider Selection Under the ICANN UDRP: Are ICANN’s Goals Being
Achieved?, AusWeb (2004), http://ausweb.scu.edu.au/aw04/papers/refereed/selby/paper.
html (noting, without performing any statistical tests, that “the Providers that resolve disputes
faster than average do have larger market shares,” and that, “although the time differences
are not that great between the slowest and fastest Provider’s Panels, it would appear
that there is a slight correlation between the time taken to resolve a dispute and the
market share of each Provider”).
23 Michael Geist, Fair.com?: An Examination of the Allegations of Systematic Unfairness in the
ICANN UDRP, 27 Brook. J. Int’l L. 903, 922 (2002) (explaining that respondents win
more often with three-member panels than with one-member panels even when controlling
for defaults); see also Mueller, supra note 17, at 15 (stating that, for disputes heard by
the National Arbitration Forum, “[respondents that] have the resources to pay an additional
$1500 to select one of three panelists . . . win slightly over half of the cases,” whereas
respondents generally win only 33% of the time when they mount a defense and 1% of the
time when they default). But see The UDRP: Fundamentally Fair, But Far from Perfect, 6 Electronic
Com. & L. Rep. (BNA) No. 34, at 937 (Aug. 29, 2001) (challenging Geist’s conclusions
regarding respondent win rates).
24 UDRP Rules, supra note 5, § 3(b)(iv). In the complaint, the complainant must
[d]esignate whether Complainant elects to have the dispute decided by a single-member
or a three-member Panel and, in the event Complainant elects a
three-member Panel, provide the names and contact details of three candidates
to serve as one of the Panelists (these candidates may be drawn from
any ICANN-approved Provider’s list of panelists) . . . .
2012] An Empirical Analysis of Fair Use Decisions 71
nant elects a one-member panel, the arbitration provider selects the
arbitrator.25 How, exactly, each provider selects the arbitrator is unknown.26
But the complainant need not elect a one-member panel—
and the respondent need not accept one. Either party can elect that a
three-member panel decide their case.27 When any party elects a threemember
panel, both parties must provide the names of three Internet
Corporation for Assigned Names and Numbers (ICANN)-approved arbitrators
to the provider.28 The provider selects two arbitrators, one
from each party’s list.29 It then selects the third arbitrator on its own “in
a manner that reasonably balances the preferences of both Parties.”30
Again, just how the arbitration provider makes its selection is un-
known.31
The author of one study decided to examine panel size after noticing
two facts. First, providers employed many of the same panelists.32
Second, a discrepancy existed in respondent win rates among three
providers (existing at the time).33 The author found that respondents
won more often with three-member panels than with one-member panels.34
Because some providers employed more one-member panels
than three-member panels, this fact explained the discrepancy in respondent
win rates.35 (The author also found that providers with the
Id.
25 UDRP Rules, supra note 5, § 6(b).
If neither the Complainant nor the Respondent has elected a three-member
Panel (Paragraphs 3(b)(iv) and 5(b)(iv)), the Provider shall appoint, within
five (5) calendar days following receipt of the response by the Provider, or the
lapse of the time period for the submission thereof, a single Panelist from its
list of panelists. The fees for a single-member Panel shall be paid entirely by
the Complainant.
Id.
26 Geist, supra note 23, at 910 (“Although the ICANN Rules and provider supplemental
rules indicate how panelists are selected, little is known about how providers determine
precisely which panelists serve on what cases. Panelist allocation has become particularly
important as the providers’ panelist rosters have converged.”).
27 UDRP Rules, supra note 5, §§ 3(b)(iv), 5(v).
28 Id.
29 Id. § 6(e).
30 Id.
31 See supra note 27 and accompanying text.
32 Geist, supra note 24, at 910 (footnote omitted).
33 Id. at 911 (“[I]f many of the same panelists are deciding cases for multiple providers,
how is it that complainants win over 80% of the time with WIPO and NAF, and only
61% of the time with eResolution?”).
34 Id. at 922.
35 Id.
72 Boston College Law Review [Vol. 53:65
highest win percentage fed disputes to a small number of complainantfriendly
arbitrators.)36 The results did not change even when controlling
for respondent defaults.37 In other words, panel size influenced the
likelihood of a respondent winning.
All of these studies have offered reasons for general trends in
UDRP disputes. None, however, focused on whether the fair use provision
protects the speech of domain-name users.38 None, that is, explored
how effectively the UDRP’s speech “safety valve” —fair use—
operates.
This Article does just that. It reports the results of an empirical
study examining how WIPO panels apply the UDRP in fair use cases.
Specifically, it asks what factor(s) influence the probability that a respondent
will successfully defend a UDRP proceeding on fair use
grounds. The data suggest that the answer is nationality: respondents
from the United States won fair use cases more often than respondents
from other countries. The data also show that panels with a majority of
U.S. arbitrators (“U.S. panels”)39 found for respondents in fair use cases
more often than did other panels.
Because the UDRP lacks a principled choice of law strategy and a
clear panel assignment procedure, this Article recommends revising
the UDRP in two ways. First, ICANN should change the current choice
of law provision. The new choice of law provision should instruct panels
to apply the local law of the respondent’s home country. Second,
ICANN should revise and make transparent its panel assignment pro-
36 Id. at 928.
A review of the 3881 single-member panel cases as of February 18, 2002 indicates
that single panel cases may not be allocated in an entirely random manner.
Most disturbing are the case allocation trends at the NAF. Of the NAF’s
1379 single-member panel cases, only six panelists decided an astonishing
56.4% (778 of 1379) of the cases. The sheer number of cases assigned to only
six people alone is surprising. The 778 cases represent 20% of the entire
UDRP single panelist caseload.
Id. (footnotes omitted).
37 Id. at 924.
38 But see Annette Kur, UDRP: A Study by the Max-Planck-Institute for Foreign
and International Patent, Copyright and Competition Law, Munich 1, 30–32 (2002)
(on file with author) (discussing various categories of cases including fair use, but attempting
to explain the results of cases by legal rules rather than by contextual factors such as the
identity of the parties involved).
39 As noted in footnote 11, this term includes only cases where two or more U.S. panelists
appeared on a three-member panel. So, for example, I did not classify a panel as a
“U.S. panel” if it contained arbitrators from the United States, German, and Spain. The
coding protocols are contained in a separate document.
2012] An Empirical Analysis of Fair Use Decisions 73
cedures. Given the proposed choice of law provision, ICANN should
appoint to panels a majority of arbitrators from the same country as the
respondent’s home country. Where the parties select a three-member
panel, one arbitrator should be from the complainant’s home country.
Part I briefly explores the UDRP and how scholars have criticized
it.40 Part II describes the methodology used to collect and analyze fair
use UDRP decisions.41 Part III presents the study’s findings.42 Part IV
discusses the implications of the findings, attempting to explain two
phenomena.43 First, why do U.S respondents win more than other respondents?
Second, why do U.S. panels apply U.S. law more than other
panels? This Part answers these questions using three tools: the UDRP’s
choice of law provision, WIPO-specific rules, and panel nationality. After
this explanation, Part IV explores potential reforms that could be
made to the UDRP. It suggests that ICANN should revise the UDRP’s
choice of law and panel assignment provisions in fair use cases.
I. The UDRP and Its Problems
There can only be one website at www.barcelona.com. Should it belong to the
city in Spain, or to Whit Stillman’s 1994 romantic comedy? The answer does
not respect national boundaries. The [Internet Corporation for Assigned
Names and Numbers, the] root authority[,] decides[] on behalf of every Internet
user everywhere in the world.44
ICANN is the “root authority.”45 Among other things, ICANN manages
the registration and allocation of domain names using the Domain
Name System (DNS).46 Although the DNS was developed to allow the
web to function effectively with increasing users and desire for names,47
it also increased competition: with a limited number of names comes
competition for those resources.48 As the Internet grew, so did disputes
40 See infra notes 45–87 and accompanying text.
41 See infra notes 88–110 and accompanying text.
42 See infra notes 111–135 and accompanying text.
43 See infra notes 136–262 and accompanying text.
44 Jack Goldsmith & Tim Wu, Who Controls the Internet?: Illusions of a Borderless
World 31 (2006) (footnote omitted).
45 See Milton Mueller, Ruling the Root: Internet Governance and the Taming
of Cyberspace 3, 7–10 (2002). Although ICANN is the primary “root authority,” other
competing, independent roots exist. Id. at 54–56 (describing several alternative root servers,
including Open NIC, Pacific Root, ORSC, New.net, Name.space, and CN-NIC).
46 Id. at 41–45, 185–208.
47 Id. at 41.
48 See id. at 57–58 (analogizing the Internet to other “unowned” resources such as the
ocean, where domain names are like fish or minerals—resources that engender competi-
74 Boston College Law Review [Vol. 53:65
over domain names.49 In response, ICANN developed50 the UDRP to
provide a quick and cheap alternative to litigating domain-name disputes
involving abusive domain-name registration;51 that is, disputes
where a party registered domain names in “bad faith.”52 To ensure that
the UDRP was more than just a fancy procedure, ICANN made it mandatory
by contracting with Network Solutions, Inc. (NSI) (now Verisign),53
which, until 1999, registered all of the “.com” generic top-level
domain names (known as gTLDs).54 As part of this contract, NSI incorporated
the UDRP into all domain-name registration agreements.55 This
enables ICANN and NSI to bind to the UDRP all individuals who register
a domain name.56 When ICANN allowed new registrars of gTLDs in
1999,57 it bound them to the UDRP as a condition of accreditation.58
tion and produce ownership rules); see also Jonathan Zittrain, ICANN: Between the Public and
the Private Comments Before Congress, 14 Berkeley Tech. L.J. 1071, 1079–83 (1999) (explaining
how “entrepreneurial forces” altered the domain-name system, changing it from space
allocated by one man (Jon Postel) to governance by ICANN).
49 Mueller, supra note 45, at 109, 114–19 (explaining that the commercialization of
the Internet followed its 1991 opening to commerce, with competition “endow[ing] . . .
the name space with economic value” that, in turn, led to an increasing number of disputes
over various domain names).
50 ICANN actually drew heavily on WIPO’s Report when developing the UDRP. See
Froomkin, supra note 14, at 631–88 (explaining that ICANN adopted each of the major
WIPO recommendations—use of the DNS to use mandatory contractual agreements, limiting
disputes to trademark and non-trademark holders, substantive rules governing
disputes, procedures for dispute resolution process, pre-emptive protections for wellknown
marks—and discussing the degree to which it did so regarding each recommendation);
Justin Hughes, The Internet and the Persistence of Law, 44 B.C. L. Rev. 359, 377–78
(2003) (explaining that ICANN asked WIPO to issue a report on domain-name disputes
and that ICANN used the report as the basis of the UDRP).
51 E.g., Elizabeth G. Thornburg, Fast, Cheap, and Out of Control: Lessons from the ICANN
Dispute Resolution Process, 6 J. Small & Emerging Bus. L. 191, 201 (2002) (stating that the
UDRP was meant “to create a remedy for a narrowly defined group of particularly egregious
cases,” which ICANN said would be “‘deliberate, bad faith, abusive, domain-name registrations
or ‘cybersquatting.’” (quoting World Intellectual Prop. Org., Final Report of the
WIPO Internet Domain Name Process § 135(i) (1999) [hereinafter Final Report], available
at http://www.wipo.int/export/sites/www/amc/en/docs/report-final1.pdf)).
52 For a definition of “bad faith” under the UDRP, see UDRP, supra note 5, § 4(b).
53 Graham J.H. Smith, Internet Law and Regulation 153 (4th ed. 2007).
54 Mueller, supra note 45, at 186–90, 192; Smith, supra note 53, at 153.
55 ICANN-NSI Registry Agreement, Internet Corp. for Assigned Names & Nos.
§ 3(A)(ii)(b), http://www.icann.org/en/nsi/nsi-registry-agreement-04nov99.htm (last modified
Nov. 10, 1999).
56 The policy is contractual, and all individuals agree to it when they register a domain
name. Domain Name Dispute Resolution Policies, Internet Corp. for Assigned Names &
Nos., http://www.icann.org/en/udrp/ (last modified Sept. 5, 2011).
57 Mueller, supra note 45, at 188; Smith, supra note 53, at 155.
58 Mueller, supra note 45, at 192 (stating that ICANN “bound all registrars of domain
names under .com, .net, and .org to [the UDRP] as a condition of accreditation”).
2012] An Empirical Analysis of Fair Use Decisions 75
Thus, all registrars are bound by the UDRP. When individuals register a
domain name, they too become contractually bound to the UDRP.59
To provide a quick and cheap resolution of disputes involving domain
names, the UDRP makes filing a claim simple: the owner of a registered
or common law trademark asserts that a “domain name is identical
or confusingly similar to a trademark or service mark in which the
complainant has rights . . . .”60 To assert a claim over a domain name,
one must “submit[] a complaint in accordance with the [UDRP] and
[the UDRP] Rules to any [one of four] Provider[s] approved by
ICANN,”61 which includes WIPO.62 The provider (be it WIPO or otherwise)
then employs a panel of arbitrators to decide the dispute.63 As
mentioned, providers do not explain how they assign particular arbitrators
to serve on panels.
The UDRP thus regulates speech on the Internet. Since its inception
in 1999, WIPO has resolved over 22,000 domain-name disputes
under the UDRP.64 As some authors note, the root—that is, ICANN—
59 Id. (“Through their contracts with registrars, all registrants of domain names under
the generic TLDs are contractually bound to submit to arbitration under the UDRP.”).
60 UDRP, supra note 5, § 4(a)(i). The UDRP does not preclude litigation as a first-order
remedy. Id. § 5.
All other disputes between you and any party other than us regarding your
domain name registration that are not brought pursuant to the mandatory
administrative proceeding provisions of Paragraph 4 shall be resolved between
you and such other party through any court, arbitration or other proceeding
that may be available.
Id.; UDRP Rules, supra note 5, § 18(a)–(b) (stating rules for arbitrators when a party to the
dispute commences litigation before or during a UDRP proceeding).
61 UDRP Rules, supra note 5, § 3(a).
62 There are a total of four providers. List of Providers, supra note 6.
63 For information on how panels are selected, see supra notes 24–31 and accompanying
text. WIPO, for example, provides a list of its arbitrators. WIPO Domain Name Panelists,
supra note 11.
64 As of the time of this study in 2010, over 19,000 UDRP complaints had been filed
with WIPO. Total Number of Cases per Year, World Intellectual Prop. Org. http://www.
wipo.int/amc/en/domains/statistics/cases.jsp (last visited Dec. 20, 2011). During this time
WIPO decided 18,624 of these complaints. See Case Outcome by Year(s) (Breakdown), World
Intellectual Prop. Org., http://www.wipo.int/amc/en/domains/statistics/outcome.jsp
(last visited Dec. 20, 2011). When the last comprehensive study of the URDP was conducted
in 2002, WIPO decided the majority of UDRP disputes (fifty-nine percent). UDRP
Decisions, UDRPInfo.com, http://udrpinfo.com/dcsn.php (click on the “View” hyperlink
next to “UDRP market share figures by provider”) (data reflect all decisions as of Feb. 18,
2002). Extrapolating from this study, this means that as of February 2002, providers other
than WIPO decided 12,942 disputes, for a total of 31,566 disputes.
76 Boston College Law Review [Vol. 53:65
decides how to settle these disputes.65 But, the question is how should
the root decide? More precisely, what law applies?
The answer is unclear. When ICANN adopted the UDRP, it included
a provision that commands a panel to “decide a complaint on
the basis of . . . any rules and principles of law that it deems applicable”
(“choice of law provision” or “provision”).66 The UDRP does not, however,
provide any guidance on when, if ever, to apply country-specific
law. If panelists applied the UDRP in a nonprecedential manner,67 this
would not necessarily be a problem.
But WIPO’s “consensus view” is that “panels consider it desirable
that their decisions are consistent with prior panel decisions dealing with
similar fact situations.”68 Yet, despite panels’ use of “precedents” from
prior disputes, no precedent exists on how to apply the UDRP’s choice of
law provision. WIPO, which ICANN charged with making recommendations
for a UDRP,69 suggested that the UDRP use existing “multilateral”
standards of intellectual property and apply them to domain names—
though it was not clear exactly what standards applied.70 Even among
65 Goldsmith & Wu, supra note 44, at 31.
66 UDRP Rules, supra note 5, § 15(a). It also included a provision that let panels “conduct
the administrative proceeding in such manner as it considers appropriate in accordance
with the Policy,” even allowing panels to formulate their own rules on admissibility
of evidence. Id. § 10(a), (d).
67 WIPO Overview of WIPO Panel Views on Selected UDRP Questions, Second Edition (“WIPO
Overview 2.0”) § 4.1, World Intellectual Prop. Org., http://www.wipo.int/amc/en/
domains/search/overview/ (last visited Dec. 20, 2011) [hereinafter WIPO Rules] (“Consensus
view: The UDRP does not operate on a strict doctrine of precedent. However, panels consider
it desirable that their decisions are consistent with prior panel decisions dealing with similar
fact situations. This ensures that the UDRP system operates in a fair, effective and predictable
manner for all parties . . . .”).
68 Id. Indeed, one scholar has shown empirically that the UDRP is a de facto
precedent-based system. In 2002, for example, he published results showing that over fifty
percent of UDRP decisions cite other UDRP decisions. Mueller, supra note 17, at 19.
More significantly, “[t]he majority not only cite other cases but rely on precedent extensively
to reason out and support their decisions.” Id.
69 ICANN Staff Report: Uniform Dispute Resolution Policy for gTLD Registrars, Internet
Corp. for Assigned Names & Nos. (Aug. 24, 1999), http://www.icann.org/en/meetings/
santiago/udrp-staff-report.htm (stating that the U.S. Government, in its 1998 White Paper,
created ICANN and “called upon ‘the World Intellectual Property Organization (WIPO)
to . . . develop recommendations for a uniform approach to resolving trademark/domain
name disputes involving cyberpiracy . . . .’” (quoting Management of Internet Names and
Addresses, 63 Fed. Reg. 31,741, 31,747 (June 10, 1998))).
70 Final Report, supra note 51, § 34.
It is further recognized that the goal of this WIPO Process is not to create new
rights of intellectual property, nor to accord greater protection to intellectual
property in cyberspace than that which exists elsewhere. Rather, the goal is to
give proper and adequate expression to the existing, multilaterally agreed
2012] An Empirical Analysis of Fair Use Decisions 77
academics, no clear consensus exists as to what law applies in any given
dispute.71 Wherever a panel looks, it will find no clear guidance.
Because the choice of law provision does not specify how panels
should apply it, panels have broad discretion in making that determination.72
That is troublesome because a panel’s decision to apply a particular
law has important consequences. When a respondent wants to
mount a fair use defense, for example, the choice of law issue can be
determinative.73 Critics claim that this results in poorly reasoned or inconsistent
decisions.
Inconsistency occurs in two forms. First, panels can apply different
UDRP precedents to similar fact-patterns.74 Alternatively, panels can
standards of intellectual property protection in the context of the new, multijurisdictional
and vitally important medium of the Internet and the DNS that
is responsible for directing traffic on the Internet.
Id.
71 One author has argued that WIPO actually meant to incorporate traditional choice
of law principles. See Froomkin, supra note 14, at 640. Others have disagreed. See Mueller,
supra note 17, at 152 (describing one of the UDRP’s three main objectives as “creat[ing]
global uniformity: that is, [it was designed] to eliminate variety and competition among
the jurisdictions and rule sets applied to domain name-trademark conflicts”).
72 Froomkin, supra note 14, at 643.
[T]he subtlety of the choice of law issue has either been lost on some arbitrators,
or they have chosen to avert their eyes from it. Perhaps the arbitrators
have concluded that “UDRP law” should, after all, be some free-standing body
of rules deracinated from any legislature and made up largely by trademark
lawyers.
Id.; Thornburg, supra note 51, at 210 (noting that the choice of law provision has resulted
in “eclectic and unprincipled ‘choice of law’ decisions,” has caused “uncertainty about
applicable law,” and has “exacerbate[d] the differences among the decisions made by the
unappealable arbitrators”); see Laurence R. Helfer & Graeme B. Dinwoodie, Designing NonNational
Systems: The Case of the Uniform Domain Name Dispute Resolution Policy, 43 Wm. &
Mary L. Rev. 141, 225 (2001).
Without constraints on these open-ended clauses, panels are left with little to
guide the exercise of their discretionary lawmaking powers. Not surprisingly,
this omission has produced a schism between panels that strictly construe the
UDRP and those that interpret the Policy more expansively to curb a broader
range of conduct by domain-name registrants.
Helfer & Dinwoodie, supra, at 225.
73 Froomkin, supra note 14, at 642 (“The choice of law issue is especially important because
choice of law in effect determines what constitutes ‘rights and legitimate interests’
. . . .”).
74 See Mueller, supra note 17, at 24 (“[T]he outcomes and precedents in [cases involving
critics or commentary] are mixed, reflecting the lack of clear standards and dissension
among panelists.”); Jacqueline D. Lipton, Commerce Versus Commentary: Gripe Sites, Parody,
and the First Amendment in Cyberspace, 84 Wash. U. L. Rev. 1327, 1356 (2006) (“[T]here is
no clear test as to when a commentary site is a fair or legitimate use. A comparison of two
78 Boston College Law Review [Vol. 53:65
apply different national law to disputes with similar facts or issues.75
This latter scenario occurred in Wal-Mart Stores, Inc. v. Walmarket Canada,
a 2000 WIPO case.76 In that case, the respondent operated in Thailand
and was domiciled in Canada.77 The complainant, Wal-Mart, was
headquartered in the United States.78 The panel noted that either Canadian
or Thai law could apply.79 Then, however, it discussed and applied
law from the United States and the United Kingdom—even though
the parties’ dispute did not appear to implicate U.K. law.80
Why did this happen? One reason may be the panel’s nationality.
That explanation is at least plausible in the Wal-Mart decision. There,
the lone panelist was from the United Kingdom and applied U.K. law.
In other cases, too, arbitrator nationality may help explain the outcome.
One author has suggested that UDRP disputes—as a result of
UDRP drafting—focus on trademark law (when there are other methods
of resolving domain-name disputes).81 For support, the author
cites one WIPO case where the respondent and complainant both
hailed from the United States.82 Although the author argues for nonearly
UDRP decisions involving consumer commentary and gripe sites demonstrates the
confusion that can arise under the UDRP in this respect.”); Lisa M. Sharrock, The Future of
Domain Name Dispute Resolution: Crafting Practical International Legal Solutions from Within the
UDRP Framework, 51 Duke L.J. 817, 833–36 (2001) (discussing the application of two different
rules in similar cases involving domain-name speculation and in cases involving
criticism (i.e., <trademarksucks.com>) and arguing that the UDRP causes inconsistency
“because it lacks specificity, a flaw magnified by the fact that precedent plays little role in
UDRP decisionmaking”).
75 See, e.g., David Wotherspoon & Alex Cameron, Reducing Inconsistency in UDRP Cases, 2
Canadian J.L. & Tech. 71, 74–75 (2003) (explaining that, among other problems, panels
can apply legal principles from jurisdictions that have no relation to the parties).
76 No. D2000-0150 (WIPO May 2, 2000), http://www.wipo.int/amc/en/domains/
decisions/html/2000/d2000-0580.html.
77 Id.
78 Id.
79 Id. (“Since the Respondent appears to be domiciled in Canada, any legal action
would have to be taken against him in that country. If the Respondent is to operate in
Thailand, then action could be taken there.”).
80 Id. (stating and applying the “legal position in England on domain names of confusingly
similar wording to registered trademarks”).
81 Lipton, supra note 74, at 1356 (“These decisions also show that, regardless of how
the UDRP is drafted, UDRP arbitrators are likely to be influenced, at least to some extent,
by domestic trademark laws.”).
82 Id. at 1356 n.163 (“‘The discussion and decision herein will . . . be governed by the
terms of the [UDRP], although reference by analogy may be made to principles of U.S. law,
as two of the Complainants are U.S. corporations, Respondent is a U.S. resident, and both
parties have cited U.S. law in their submissions.’” (quoting Bridgestone Firestone, Inc. v.
Myers, No. D2000-0190 (WIPO July 6, 2000), http://www.wipo.int/amc/en/domains/
decisions/html/2000/d2000-0190.html)).
2012] An Empirical Analysis of Fair Use Decisions 79
trademark based tools to resolve these disputes, the observation nonetheless
highlights the influence of national law in UDRP disputes.83 Indeed,
we might expect that the law of the panel’s home country will
influence how the panel applies the UDRP. As noted above, this panelbased
explanation seems plausible in the Wal-Mart case.84
Panel nationality, however, highlights a broader concern: how are
panels deciding cases? In particular, scholars are concerned with panels’
ability and tendency to suppress speech by controlling domainname
use. Some, for example, worry that decisional inconsistency
could “chill speech.”85 Because people fear the cancellation or transfer
of a selected domain name, the argument goes, they will be reluctant to
use them at all. Others are concerned that providers like WIPO are biased
in favor of the complainant.86
None of these studies, though, examined how specific provisions
of the UDRP impact individuals’ ability to “speak” using domain names
that incorporate trademarks—an issue of growing importance. As the
number of trademarks and domain names multiply, people increasingly
will use those trademarks in domain names. Yet one’s ability to use domain
names will depend on how ICANN manages that space. Currently,
gTLD space is both expanding and contracting. As ICANN opens
new gTLDs, it restricts those able to apply for them.87 By limiting individuals’
and small businesses’ ability to use new domain-name space,
83 Id. at 1361–64 (arguing for a “zoning” approach that would give trademark holders
the “most obvious .com versions of their trademarks in the domain space”).
84 No. D2000-0150 (WIPO May 2, 2000).
85 See Keith Blackman, The Uniform Domain Name Dispute Resolution Policy: A Cheaper Way
to Hijack Domain Names and Suppress Critics, 15 Harv. J.L. & Tech. 211, 251–52 (2001)
(quoting U.S. Supreme Court Justice Hugo Black and arguing that inconsistent or unclear
law about what speech is permitted can chill protected speech).
86 See supra notes 19–39 and accompanying text.
87 See generally Internet Corp. for Assigned Names & Nos., gTLD Applicant Guidebook
(2011), available at http://www.icann.org/en/topics/new-gtlds/rfp-clean-19sep11en.pdf
(laying out the details for applying for new gTLDs). ICANN is introducing new
gTLDs. Registering for these gTLDs, however, is, in ICANN’s words, “a much more complex
process” than registering for an existing gTLD. Frequently Asked Questions, Internet
Corp. for Assigned Names & Nos., http://newgtlds.icann.org/applicants/faqs/faqs-en
(last visited Dec. 20, 2011). Indeed, applying for a new gTLD requires, among other
things, online user registration, application submission, fee submission, fee conciliation,
and appropriate documentation. gTLD Applicant Guidebook, supra, § 1.1.1 (explaining
the life cycle of new gTLDs), § 1.2.2 (listing required documents). To be eligible to register,
the entity seeking registration must be an “[e]stablished corporation[], organization[],
or institution[] in good standing . . . . Applications from individuals[,] . . . sole proprietorships[,]
[and] . . . yet-to-be formed legal entities . . . will not be considered.” Id. § 1.2.1.
What’s more: the evaluation fee for a new gTLD is expensive, with an estimated price of
$185,000 (including a $5,000 deposit). Id. § 1.5.1.
80 Boston College Law Review [Vol. 53:65
the new gTLD policy shrinks rather than grows the existing space for
these potential registrants.
This de facto shrinking and the prospect of new, expensive gTLDs
amplify the importance of the fair use defense. With the growth of
companies and internet access, people will continue to voice their opinion
using websites—and frequently domain names. The extent to
which people will be able to use trademarks in their domain names depends,
at least in part, on the fair use defense. To date, however, no
study has systematically examined this defense, and whether or how the
fair use defense works; i.e., whether it applies uniformly to all parties,
regardless of nationality. This study undertook this task.
II. Methodology Used to Collect and Analyze
Fair Use UDRP Decisions
This study examined fair use domain-name disputes arbitrated by
WIPO under the UDRP. As mentioned above, section 4(c)(iii) of the
UDRP allows respondents to “mak[e] a legitimate noncommercial or
fair use of the domain-name [that incorporates a trademark], without
intent for commercial gain to misleadingly divert consumers or to tarnish
the trademark or service mark at issue.”88 Nevertheless, to better
understand where the fair use provision fits within the UDRP, it is helpful
to describe the three main elements required to prove a prima facie
case under the UDRP.89
First, the complainant must show that the “domain name is identical
or confusingly similar to a trademark or service mark in which the
complainant has rights . . . .”90 Second, the complainant must show that
the respondent has no “rights or legitimate interests” in the domain
name.91 Third, the complainant must show that the respondent “registered
and is . . . us[ing] [the domain name] in bad faith.”92 The complainant
must make a prima facie case (on these elements) to suc-
88 UDRP, supra note 5, § 4(c)(iii).
89 In addition to substantive provisions, the UDRP contains a variety of procedural
rules. See generally UDRP Rules, supra note 5. One such procedural mechanism is that the
respondent must file a response within twenty days after receiving provider notification of
the complaint. See id. § 5(a); id. § 4(c) (explaining that the proceeding commences when
the provider fulfills its notification responsibilities under Rule 2(a)); id. § 2(a) (detailing
the provider’s responsibility to forward the complaint to the respondent). Failure to file a
response counts as a “default,” which simply means that the panels decide the case without
a response. Id. § 14(a). A default is not an automatic loss.
90 UDRP, supra note 5, § 4(a)(i).
91 Id. § 4(a)(ii).
92 Id. § 4(a)(iii).
2012] An Empirical Analysis of Fair Use Decisions 81
ceed.93 Even if the complainant makes a prima facie case, though, the
burden shifts, and the respondent may defeat the complaint by demonstrating
“rights or legitimate interests” in the domain name.94
The UDRP specifies three ways in which a respondent can have
rights or legitimate interests in a domain name: the respondent (1) is
using the domain name in connection with a bona fide offering of
goods or services;95 (2) is commonly known by the domain name;96 or
(3) is “making a legitimate noncommercial or fair use of the domain
name, without intent for commercial gain to misleadingly divert consumers
or to tarnish the” plaintiff’s trademark.97 I examined only decisions
substantively discussing this last issue: whether the respondent’s
use of the alleged mark in the domain name qualified as legitimate
noncommercial or fair. I refer to these decisions as “fair use cases.” Figure
1 (below) shows the process by which I selected, sampled, and categorized
cases. The following two Sections explain this process in detail.
A. Sample Selection: Method of Selection
WIPO makes all of its decisions publicly available and textsearchable
on its website (“WIPO database”).98 I collected cases decided
between January 1, 1999 and September 24, 2010. I searched the WIPO
database using only one search term, “4(c)(iii),”99 indicating that the
search results (i.e., the decisions) “must” contain this phrase.100 The
93 WIPO Rules, supra note 67, § 2.1 (“[A] complainant is required to make out a prima
facie case that the respondent lacks rights or legitimate interests.”).
94 Id.
Once such prima facie case is made, the burden of production shifts to the respondent
to come forward with appropriate allegations or evidence demonstrating
rights or legitimate interests in the domain name. If the respondent fails to
come forward with such appropriate allegations or evidence, a complainant is
generally deemed to have satisfied paragraph 4(a)(ii) of the UDRP.
Id.; see also UDRP, supra note 5, § 4(c)(i)–(iii) (describing ways in which a domain name
owner can show rights or legitimate interests in a domain name).
95 UDRP, supra note 5, § 4(c)(i).
96 Id. § 4(c)(ii).
97 Id. § 4(c)(iii).
98 Search WIPO Cases and WIPO Panel Decisions, World Intellectual Prop. Org., http://
www.wipo.int/amc/en/domains/search/ (last visited Dec. 20, 2011). Because extracting the
decisions from WIPO’s database, WIPO changed its search interface. This new interface does
not extract results in the same manner as the interface I used to conduct my search.
99 Section 4(c)(iii) of the UDRP governs “fair use” of domain names. UDRP, supra note
5, § 4(c)(iii).
100 I collected cases using a search on September 24, 2010. I have not updated cases
since.
82 Boston College Law Review [Vol. 53:65
search revealed 2688 non-unique results from January 1, 1999 through
September 24, 2010.101 These decisions appeared in languages other
than English.102
After compiling the sets of search results, I eliminated the duplicate
cases. This left 1415 unique cases. I first randomly sampled 499
decisions. From this sample, I found only a small number of respondent
wins. Because the number of wins was too small to draw inferences
(n < 20), I drew another random sample of 500 cases. From the random
sample totaling 999 decisions, I found 148 cases that substantively
discussed the fair use defense. In 40 of those cases, the respondent
101 “Non-unique” results contain duplicate decisions.
102 To review cases not in English, I used GoogleTranslate.
Figure 1: Case Selection and Sampling Method
1415 Decisions:
Where 4(c)(iii) Appeared
18,624 Decisions
WIPO Online Database
2,688 Decisions:
Where 4(c)(iii) Appeared
999 Decisions:
Total Random Sample
40 Decisions:
Respondent Wins
on Fair Use
108 Decisions:
Respondent Loses and Panel
Discusses Fair Use
Sorted Decisions Based on Result & Fair
Use Determination
499 Decisions:
Random Sample
500 Decisions:
Random Sample
+
Drew Two
Random Samples
Searched Database Using Term “4(c)(iii)”
Eliminated Duplicate Decisions
2012] An Empirical Analysis of Fair Use Decisions 83
won. The complainant won in the remaining 108 decisions. I did not
examine the remaining 416 cases.
B. Types of Cases Selected for Sample
I manually reviewed all 999 randomly sampled cases to determine
whether they were relevant. Relevant decisions included two kinds of
cases. First, I identified cases in which the respondent succeeded in defending
the complaint on fair use grounds. I identified these cases easily
because the panels would discuss why the domain name constituted a
noncommercial fair use.103 Put simply, a respondent who succeeds on a
fair use defense defeats the complaint; i.e., the respondent “wins.”104 I
identified forty of these decisions.
Second, I identified cases in which, although the respondent lost,
the panel discussed fair use in the decision. Nearly every case I examined,
however, mentioned section 4(c)(iii) in some way. Often, for
example, decisions would summarily find no fair use defense, or cite
the relevant provision but never discuss it. Therefore, I had to narrow
the cases to those in which the panel substantively addressed the fair
use defense.
Generally, I excluded decisions finding no fair use because the
website was commercial. I kept cases in which the panel decided the
case on commerciality grounds but noted that the website may have
been for a legitimate purpose, such as criticism.105 Because these cases
specifically addressed whether a use was fair and legitimate, they were
relevant. I excluded cases in which the panel decided summarily that
the respondent used the domain name commercially. Including these
cases would have skewed the results because this criterion encompassed
nearly every case where a respondent lost.
When the respondent lost and its domain name did not display a
live website, I included only those cases where the respondent asserted
that it was planning on making a fair use. I included these cases for
consistency, as some respondent fair use victories occurred when the
103 For respondent wins, the decision must have resolved the fair use issue in the respondent’s
favor. I did not count decisions that punted on the fair use question but dismissed the
complaint on other grounds. E.g., Kendall v. Mayer, No. D2000-0868 (WIPO Oct. 26, 2000),
http://www.wipo.int/amc/en/domains/decisions/html/2000/d2000-0868.html.
104 If the respondent succeeded in proving a fair use defense as well as another defense
under the UDRP, I coded the case because it still qualifies as one in which the respondent
won on fair use grounds.
105 E.g., Pfizer Inc. v. e2eTech LLC, No. D2007-0099 (WIPO Apr. 6, 2007), http://
www.udrpsearch.com/wipo/d2007-0099.
84 Boston College Law Review [Vol. 53:65
respondent had not yet developed a website. Like the aforementioned
commerciality cases, I included these cases because they substantively
addressed the issue of fair use. For the opposite reason, I excluded cases
in which the panel summarily rejected the respondent’s fair use defense;
namely, because they did not substantively discuss fair use.
Occasionally, a decision would defer the fair use discussion to the
section on bad faith—one of three elements the complainant must show
to win a UDRP complaint.106 I included these decisions and classified
them as “other” unless the bad faith discussion turned specifically on
the nature of the website (i.e., criticism, fan, information, and personal).
If the bad faith discussion turned on the nature of the site, I classified
it as one of the existing categories. I included these decisions because
the primary issue—regardless of where in the decision the panel
resolved it—was whether the respondent made a fair use of the domain
name. Because the fair use inquiry sometimes rolled into the bad faith
question, it made sense to include cases where this overlap occurred.
I also eliminated cases in which, although the panel discussed fair
use to some degree, fair use was not really disputed. So, for example, I
excluded cases in which, even though the panel discussed the fair use
defense, the website that the domain name displayed was simply a pornographic
one.107 I excluded these cases because they did not substantively
implicate the fair use defense.
I also excluded decisions involving websites that offered only advertisements
or domain names that displayed a “parked” website containing
links to other vendors or websites. I excluded these cases because
they generally could not qualify as either noncommercial or
fair,108 and, as a result, the respondent could not and would not assert a
fair use defense. In general, panels tended to resolve these cases summarily
in favor of the complainant.
After obtaining all 148 cases substantively discussing the fair use
defense, I coded them for a variety of variables. One variable was sub-
106 UDRP, supra note 5, § 4(a)(iii) (requiring the complainant to show that, in addition
to the domain name being confusingly similar to the trademark and the respondent not
having rights or legitimate interests in the domain name, it has been registered and is being
used in bad faith).
107 E.g., Microsoft Corp. v. Horner, No. D2002-0029 (WIPO Feb. 27, 2002), http://
www.wipo.int/amc/en/domains/decisions/html/2002/d2002-0029.html.
108 But see CP Masters B.V. v. RareNames, WebReg, No. D2009-1673 (WIPO Feb. 11, 2010),
http://www.wipo.int/amc/en/domains/decisions/html/2009/d2009-1673.html (finding
the registration of a domain name “speculatively, to either generate click-through revenue
and be part of the portfolio of domain names in which the Respondent deals” fair because
no bad faith existed).
2012] An Empirical Analysis of Fair Use Decisions 85
ject-matter: each case fell into a category that designated the type of site
displayed upon entering the domain name. I coded them into five categories:
criticism, fan site, informational, personal, and other. “Criticism”
means that the website displayed at a domain name made critical
statements about some subject. The most common subject was the
trademark holder, who was seeking to cancel or transfer the domain
name. “Fan site” means that a supporter or proponent of a topic, person,
or entity created the website displayed at the domain name. “Informational”
means that the website displayed at the domain name
provided information about some kind of topic, person, or entity. Frequently
this was the trademark owner who attempted to cancel or transfer
the domain name. “Personal” means that an individual created a
website displayed at the domain name for personal use, such as e-mail
or displaying family photographs. “Other” refers to fair use decisions
that did not fit any of these categories.
I also coded for other variables, including (1) panel, respondent,
and complainant nationality; (2) whether the panel cited U.S. law, foreign
law, or neither; and (3) whether the respondent won in whole or
in part.
I coded the nationality of each arbiter using WIPO’s website,
which lists its arbitrators by country.109 I coded the nationality of respondents
and complainants using the decision itself, which indicated
from where the parties hailed.
I also coded whether and what law each panel cited. Whenever a
decision either mentioned or listed the law of another country, it was
counted as “citing” that country’s law. Occasionally, a decision would
cite a non-national law, such as a law of the European Union. These
decisions also were counted as “citing” foreign law.
Of course, my analysis would have been incomplete without coding
for the ultimate result of the case. Here, I coded whether a respondent
“won” on the fair use defense. Given that multiple domain names
may have been at issue, a respondent could win the dispute and lose the
fair use defense in the same case. Thus, I recorded whether the respondent
won as to all domain names (a “whole win”) or as to less than
all the domain names at issue (a “partial win”). I also recorded whether
the respondent “lost” on the fair use defense (and typically, but not always,
on all defenses).110
109 See WIP Domain Name Panelists, supra note 11.
110 For a detailed description of how the variables were coded, please refer to the protocols,
which are available at http://www.bc.edu/content/dam/files/schools/law/bclaw
review/pdf/53_1/simon_protocol.pdf.
86 Boston College Law Review [Vol. 53:65
III. Findings
The purpose of this analysis was to determine whether the nationality
of respondents or panels influenced the outcome of fair use proceedings.
First, I analyzed whether respondents who were from the
United States won more often than those from other countries.111
Second, I analyzed whether U.S. panels were more likely to find fair use
than non-U.S. panels.112 Third, I analyzed the nationalities of panels
and respondents in fair use cases.113 I report the findings in the next
three Sections.
A. U.S. Respondents Win Twice as Many Fair Use Cases as Other Respondents
Figure 2 shows that U.S. respondents won twice as often as other
respondents. Specifically, U.S. respondents won 35% of the time while
other respondents won only 17% of the time. The differences were statistically
significant (p = 0.016).114
Figure 2. Respondents’ Outcomes in UDRP Fair Use Cases, by Respondent Nationality
Respondent
Wins
Respondent
Losses
Total
U.S.
Respondent
29
(35%)
54
(65%)
83
(100%)
Other
Respondent
11
(17%)
54
(83%)
65
(100%)
Total 40
26%
108
74%
148
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were statistically significant (p = 0.016).
These data also show that respondents in fair use cases win 26% of the
time. That is slightly higher than the overall percentage of respondent
111 See infra notes 114–120 and accompanying text.
112 See infra notes 121–127 and accompanying text.
113 See infra notes 128–135 and accompanying text.
114 I used a Fisher’s Exact test to compute statistical significance. A Fisher’s Exact test is
used to make comparisons across a two-by-two table (i.e., to compare proportions). H.
Russell Bernard, Social Research Methods: Qualitative and Quantitative Approaches
569–70 (2000); Michael O. Finkelstein & Bruce Levin, Statistics for Lawyers
154–56 (2d ed. 2001). Typically, a Fisher’s Exact test is used when the cells in a table
have an expected value of below five (i.e., for a small sample). Bernard, supra, at 569. The
test tells us the likelihood that our table shows an actual difference (i.e., that the differences
are not a result of chance). See id. at 569–70; Finkelstein & Levin, supra, at 156 n.2.
Put another way, it tells us in Figure 2, for example, that we could expect these results by
chance 1.5% of the time.
2012] An Empirical Analysis of Fair Use Decisions 87
wins (20%) found in the last comprehensive empirical study.115 The
differences in the proportions of respondent wins in fair use cases
compared to WIPO decisions generally was not statistically significant (z
= -1.76, p = 0.08).116
In addition to respondent nationality, I tested panel nationality
against respondent nationality. Figure 3 shows that respondents win
more often when a U.S. panel decides the dispute than when a panel
from another country decides the dispute. Respondents win 36% of the
time when a U.S. panel decides their case and 20% when a panel from
another country does so. The difference was statistically significant (p =
0.04).
Figure 3. Respondents’ Outcomes in UDRP Fair Use Cases, by Panel Nationality
Respondent
Wins
Respondent
Losses
Total
U.S.
Panel
24
(36%)
43
(64%)
67
(100%)
Other
Panel
16
(20%)
65
(80%)
81
(100%)
Total 40
27%
108
73%
148
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were statistically significant (p = 0.040).
After grouping the cases into criticism and noncriticism categories,
U.S. respondents won more than other respondents in both categories.
(“Noncriticism” cases mean all cases that did not involve the “criticism”
category described in the previous Part.)117 The differences in winning
percentage, however, were statistically significant only for noncriticism
cases. That is, U.S. respondents won more often than other respondents
in both criticism and noncriticism cases. But in only noncriticism
cases were those differences statistically significant. Figure 4 shows that,
in criticism cases where in which respondents won, U.S. respondents
comprised 70% of cases and won 37% of the time.118
115 Geist, supra note 23, at 910 (stating “that complainants win 80.6% of the time with
WIPO”); see Mueller, supra note 17, at 156–57 (showing a roughly 80% win-rate for complainants
in all cases in which a WIPO panel rendered a decision).
116 I used a proportions test to generate the z-statistic. I ran the test based on the number
of WIPO cases reported on UDRPinfo.com (2565). UDRP Decisions, supra note 64 (click
the “View” hyperlink for “UDRP market share figures by provider”).
117 See supra notes 108–110 and accompanying text.
118 I found that, in criticism cases, respondents won approximately 30% of the time.
Another author found that respondents won 36% of the time in cases classified as “critic or
commentary.” Mueller, supra note 17, at 11.
88 Boston College Law Review [Vol. 53:65
Figure 4. Respondents’ Outcomes in UDRP Criticism Fair Use Cases,
by Respondent Nationality
Respondent
Wins
Respondent
Losses
Total
U.S.
Respondent
14
(37%)
24
(63%)
38
(100%)
Other
Respondent
6
(22%)
21
(78%)
27
(100%)
Total
20
31%
45
69%
65
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were not statistically significant (p = 0.278) (one-tailed, p =
0.162).
Other respondents, by contrast, comprised only 30% of the cases and
won 22% of the time. The differences fell short of statistical significance
(n = 63, p = 0.278) (one-tailed, p = 0.162).119
In noncriticism cases in which the respondent won, U.S. respondents
comprised 75% of the cases and won 33% of the time. Other respondents
comprised 25% of the cases and won 13% of the time. As
Figure 5 shows, the differences in win percentage for U.S. and other
respondents for noncriticism cases were statistically significant (n = 83,
p = 0.041) (one-tailed, p = 0.028).120
Figure 5. Respondents’ Outcomes in UDRP Noncriticism Fair Use Cases,
by Respondent Nationality
Respondent
Wins
Respondent
Losses
Total
U.S.
Respondent
15
(33%)
30
(67%)
45
(100%)
Other
Respondent
5
(13%)
33
(77%)
38
(100%)
Total 20
24%
63
76%
83
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were statistically significant (p = 0.041) (one-tailed, p = 0.028).
Although the difference between the winning percentage of U.S.
and non-U.S. respondents was not statistically significant, I still needed
to examine panel nationality. The difference in winning percentage
between U.S. and non-U.S. panels was significant for criticism but not
noncriticism cases. Figure 6 shows that U.S. panels find for respondents
119 I used a Fisher’s Exact to test the differences.
120 I used a Fisher’s Exact to test the differences.
2012] An Empirical Analysis of Fair Use Decisions 89
in 50% of the cases compared to 18% for other panels. This difference
was statistically significant (n = 65, p = 0.012).
Figure 6. Respondents’ Outcomes in UDRP Criticism Fair Use Cases, by Panel
Nationality
Respondent
Wins
Respondent
Losses
Total
U.S.
Panel
13
(50%)
13
(50%)
26
(100%)
Other
Panel
7
(18%)
32
(82%)
39
(100%)
Total 20
31%
45
69%
65
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were statistically significant (p = 0.012) (one-tailed, p = 0.007).
The story for noncriticism cases is different. Figure 7 shows that
respondent wins and losses differ based on the nationality of the deciding
panel. In cases decided by U.S. panels, respondents won 27% of the
time. When other panels decided the case, respondents won 21% of
the time. The difference between these percentages, however, was not
statistically significant (n = 83, p = 0.798).
Figure 7. Respondents’ Outcomes in UDRP Noncriticism Fair Use Cases, by Panel
Nationality
Respondent
Wins
Respondent
Losses
Total
U.S.
Panel
11
(27%)
30
(73%)
41
(100%)
Other
Panel
9
(21%)
33
(79%)
42
(100%)
Total 20
24%
63
76%
83
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were not statistically significant (p = 0.615) (one-tailed, p = 0.375).
To summarize, the data revealed two findings. First, U.S. respondents
win more often in fair use cases than respondents from other
countries. Second, in fair use cases, U.S. panels rule for respondents
more often than panels from other countries. Respondent nationality
and panel nationality have opposite relationships with criticism and
noncriticism cases. In noncriticism cases, the respondent’s nationality is
related to whether they win or lose their fair use defense. The same is
not true in criticism cases—respondent nationality is not related to
whether the respondent wins a fair use dispute. Panel nationality shows
just the opposite. In noncriticism cases, the nationality of the panel is
90 Boston College Law Review [Vol. 53:65
not related to the outcome. In criticism cases, however, U.S. panels decide
more disputes in favor of the respondent than other panels.
B. U.S. Law Dominates UDRP Decisions
To investigate further, I determined whether and when panels
cited U.S. law in their decisions. To qualify as “citing” U.S. law, a panel
merely had to cite a U.S. case in any portion of the decision. In many
decisions, panels cited U.S. law in discussions of subjects other than fair
use. I used the “citation anywhere” criterion because many fair use decisions
did not cite U.S. law in their fair use discussion. Nevertheless, a
citation to U.S. law in another part of the decision shows that the panel
considered U.S. law in some respect. Because it was difficult to devise a
metric that would measure the influence of U.S. law, this variable is an
indication of whether U.S. law influenced the decision making process.
Even those numbers alone, however, would not tell the whole story.
Recall that the primary findings show that U.S. panels give, and U.S.
respondents get, more favorable treatment than other panelists or respondents.
Thus, it is important to see both the nationality of the panels
citing U.S. law and the nationality of the respondents in cases where panels
cite U.S. law. Figures 8 and 9 show the data for these two groups.
Overall, panels cited U.S. law in fair use cases 30% of the time. Panels
cited foreign law in fair use cases only 6% of the time. In the remaining
64% of fair use cases, panels did not cite any nation’s law.
Figure 8. Panels Citing U.S. Law in UDRP Fair Use Cases, by Respondent Nationality
U.S. Law
Cited
Foreign Law
Cited
No Law Cited Total
U.S.
Respondent
34
(41%)
1
(1%)
48
(58%)
83
(100%)
Other
Respondent
10*
(15%)
8
(12%)
47
(73%)
65
(100%)
Total 44
30%
9
6%
95
64%
148
100%
Source: WIPO Online Database of UDRP Decisions
A Yates’ Chi-Square showed the differences were statistically significant, X2
(2, N = 148) = 13.837, p<
0.000.
*Three additional cases cited U.S. law and foreign law, so I excluded them from the U.S. Law
Category.121
121 Coast Hotels Ltd. v. Lewis, No. D2009-1295 (WIPO Nov. 23, 2009), http://www.
udrpsearch.com/wipo/d2009-1295; Volkswagen Aktiengesellschaft v. MapleOne Solutions,
No. D2003-0326 (WIPO June 16, 2003), http://www.wipo.int/amc/en/domains/decisions/
html/2003/d2003-0326.html; Espirito Santo Fin. Grp. S.A. v. Colman, No. D2001-1214
(WIPO Jan. 22, 2002), http://www.wipo.int/amc/en/domains/decisions/html/2001/d2001-
1214.html. Excluding these cases entirely does not alter the chi-square analysis.
2012] An Empirical Analysis of Fair Use Decisions 91
Figure 8 also shows the data specific to respondent nationality. Panels
cited U.S. law in 41% of cases with U.S. respondents. By contrast, they
cited U.S. law in only 15% of cases with other respondents.122
I also examined patterns of foreign law citations. In cases with U.S.
respondents, panels almost never cited or applied foreign law. In cases
where the respondent was not from the U.S., however, panels cited foreign
law 12% of the time. These were statistically significant, X2
(2, N =
148) = 13.837, p = 0.000.123
Finally, I investigated the nationalities of panels applying U.S. law.
If a panel-specific legal bias exists, it would manifest itself in citation
patterns. Figure 9 shows that U.S. panels cited U.S. law 47% of the time
and cited foreign law 0% of the time.124
Figure 9. Panels Citing U.S. Law in UDRP Fair Use Cases, by Panel Nationality
U.S. Law
Cited
Foreign Law
Cited
No Law Cited Total
U.S.
Panel
32
(47%)
0
(0%)
35
(52%)
67
(100%)
Other
Panel
12*
(15%)
9
(11%)
60
(74%)
81
(100%)
Total 44
30%
9
6%
95
64%
148
100%
Source: WIPO Online Database of UDRP Decisions
A Yates’ Chi Square showed the differences were statistically significant, X2
(2, N = 148) = 17.572, p <
0.000.
*Three additional cases cited U.S. law and foreign law ,so I excluded them from the U.S. Law Category.
They were included in Foreign Law Citations.125
Other panels cited U.S. law 15% of the time, and cited foreign law 11%
of the time. The differences were statistically significant X2
(2, N = 148)
122 Additionally, panels applied U.S. law in 20% of the cases with U.S. respondents. In cases
with other respondents, however, panels applied U.S. law 6% of the time. The differences
were statistically significant (n =148, p = 0.017). Simply applying U.S. law did not always
mean the respondent won. E.g., Wal-mart Stores, Inc. v. Walsucks & Walmarket Puerto Rico,
No. D2000-0477 (WIPO July 20, 2000), http://www.wipo.int/amc/en/domains/ deci-
sions/html/2000/d2000-0477.html.
123 I used a Yates’ Chi-Square to test these differences. Yates’ Chi-Square is a version of
Chi-Square used with small samples. Bruce M. King, Patrick J. Rosopa & Edward W.
Minium, Statistical Reasoning in the Behavioral Sciences 375 (6th ed. 2011). It is
designed to reduce the overestimation of statistical significance by reducing the value of
X2
, which makes statistical significance less likely. Id. The Chi-Square uses a sample frequency
or proportion to test a hypothesis about a corresponding population from which
the sample is drawn. Frederick J. Gravetter & Larry B. Wallnau, Statistics for the
Behavioral Sciences 607 (8th ed. 2009).
124 See infra Figure 9.
125 Coast Hotels, No. D2009-1295; MapleOne Solutions, No. D2003-0326; Espirito Santo, No.
D2001-1214. Excluding these cases entirely does not alter the chi-square analysis.
92 Boston College Law Review [Vol. 53:65
= 17.572, p = 0.000.126 U.S. panels found fair use in most of the cases
(75%) in which they applied U.S. law.127
C. U.S. Panels Decide Mostly Cases Involving U.S. Respondents
To investigate nationality further, I examined the nationality of
panels deciding cases with U.S. respondents. The Introduction explains
the UDRP’s procedures for selecting both one-member and threemember
panels.128 In spite of the procedures, little is known about how
WIPO (and other providers) select specific arbitrators.129 Figure 10
shows that most cases that U.S. panels decide involve U.S. respondents.
Other panels, by contrast, decided mainly cases with respondents from
other countries.
Figure 10. Nationality of Panels Deciding Fair Use Cases, by Respondent Nationality
U.S. Panel Other Panel Total
U.S.
Respondent
56
(67%)
27
(33%)
83
(100%)
Other
Respondent
12
(18%)
53
(82%)
65
(100%)
Total 68
46%
80
54%
148
100%
Source: WIPO Online Database of UDRP Decisions
A Fisher’s Exact showed the differences were statistically significant (p = 0.00)
The differences were statistically significant (p = 0.00). Additionally,
U.S. panels are deciding almost 46% of all fair use cases, compared
with 54% for other panels.
But how does this compare to the number of arbitrators by nationality?
There are a total of 453 WIPO arbitrators.130 Of those, 109 are
from the United States.131 The remaining 344 arbitrators are from other
countries.132 Thus, of all WIPO arbitrators, 24% are from the U.S.
and the remaining 76% are from other countries. Figure 11 shows that,
after the United States, the next four countries with the highest percen-
126 Again, I used a Yates’ Chi-Square to test these differences.
127 Of the fifteen cases decided by U.S. panels that applied U.S. law, eleven found fair
use.
128 See supra notes 23–30 and accompanying text.
129 Geist, supra note 23, at 910 (“Although the ICANN Rules and provider supplemental
rules indicate how panelists are selected, little is known about how providers determine
precisely which panelists serve on what cases. Panelist allocation has become particularly
important as the providers’ panelist rosters have converged.”) (footnote omitted).
130 WIPO Domain Name Panelists, supra note 11.
131 Id.
132 Id.
2012] An Empirical Analysis of Fair Use Decisions 93
tage of arbitrators are the United Kingdom (10%), Australia (6%),
Switzerland (5%), and France (5%).133
Figure 11. Proportion of WIPO UDRP Arbitrators, by Arbitrator Nationality
Arbitrator Nationality
Number of
Arbitrators
Proportion of Arbitrators(%)
United States 109 24%
United Kingdom 46 10%
Australia 29 6%
Switzerland 23 5%
France 22 5%
All Others 225 50%
Total 453 100%
Source: WIPO online list of arbitrators. The list is not entirely accurate as several arbitrators that decided
disputes were not listed in this database.
The other fifty countries not listed in Figure 11 comprised the remaining
50% of arbitrators. So, although U.S. arbitrators comprise 24% of
all arbitrators, U.S. panels decided 45% of all fair use cases. In other
words, U.S. panels are assigned to more than their proportional share
of fair use cases.
But, there is one other question: who are the respondents when
the U.S. panels decide fair use cases? If they are from the United States,
then the difference between the total percentage of U.S. arbitrators
and the percentage of U.S. panels deciding fair use can be explained by
a panel assignment policy: WIPO assigns U.S. panels to decide cases
involving U.S. respondents.134 Specifically, of the sixty-seven cases decided
by U.S. panels, fifty-two involved U.S. respondents and U.S. complainants.
Of those cases, four cases involved only U.S. respondents.
Five other cases included only a U.S. complainant.
Thus, the data show that 78% of cases with U.S. panels involved
U.S. respondents and complainants. When I included all cases that involved
at least one U.S. party, that number jumped to 88%. Only in
12% of all cases decided by U.S. panels was neither party from the
United States. Put simply, most U.S. panels decided cases involving at
least one party from the United States.
Panels from countries other than the United States had lower
rates. Figure 12 shows that other panels decided cases involving parties
133 The following are the country-specific number of arbitrators: United Kingdom
(forty-five), Australia (twenty-nine), Switzerland (twenty-three), and France (twenty-two).
Id.
134 Because three-member panels decided only twenty out of 148 cases (13.5%), WIPO
assigned arbitrators in 86.5% of all cases.
94 Boston College Law Review [Vol. 53:65
with no relation to their home countries much more frequently than
U.S. panels.
Figure 12. Cases in Which at Least One Party Is from the Same Country as the Panel,
by Total Number of Cases Decided
Panel Country Number of Cases
At Least 1 Party
From Panel Country (%)
United States 67 59 (88%)
Australia 20 4 (20%)
United Kingdom 16 8 (50%)
Canada 9 2 (22%)
New Zealand 7 1 (14%)
To summarize, the data show three things. First, panels deciding
cases with U.S. respondents cited U.S. law more frequently than panels
with other respondents. Second, U.S. panels cited U.S. law more frequently
than other panels cited U.S. law. In fact, the respondent won
every case in which a U.S. panel cited U.S. law. Finally, U.S. panels decided
mostly cases with at least one U.S. party.135 Other panels, by contrast,
decided a majority of cases in which neither party shared the
same nationality as the panel.
When combined with the previous findings, the data show that, in
fair use cases decided by WIPO, U.S. panels provide, and U.S. respondents
receive, treatment more favorable than panels or respondents
from other countries. I refer to these phenomena as “U.S. Favoritism.”
IV. Implications: Explaining U.S. Favoritism and
Reforming the UDRP
These findings raise two issues. First, what explains U.S. Favoritism?
Second, given this explanation, what reforms, if any, should ICANN
make to the UDRP?
A. Explaining U.S. Favoritism
Two factors help to explain U.S. Favoritism. The first is the UDRP’s
choice of law provision. The second is panel nationality. But before
reaching the issue of panel nationality, this Section addresses WIPO’s
135 Out of all 148 disputes, 83 (56%) involved U.S. respondents. See supra Figure 2. In
all disputes before WIPO, U.S. respondents comprise nearly 37% of all respondents. Respondent
Country Filing (Ranking), World Intellectual Prop. Org., http://www.wipo.int/
amc/en/domains/statistics/countries.jsp?party=R (last visited Dec. 20, 2011).
2012] An Empirical Analysis of Fair Use Decisions 95
official opinion on rules applied by panels.136 Although not a factor
that caused U.S. Favoritism, it does elucidate how the phenomenon
arose. Thus, the analysis proceeds in three steps. First, this Section discusses
how the choice of law provision’s vagueness enabled panels to
cite and apply—or avoid citing and applying—local (national) law.137
Second, it explains how WIPO Rules that favored U.S. parties developed,
and thus contributed to U.S. Favoritism.138 The Section ends with
a discussion of how panel nationality, combined with the choice of law
provision and WIPO rules, played a role in U.S. Favoritism.139
1. Choice of Law Provision Influences What Law Applies
The UDRP’s choice of law provision can help explain why U.S. Favoritism
exists. As noted in Part I, the choice of law provision is open
ended: it gives panels discretion to apply any “principles of law that
[they] deem[] applicable.”140 One could construe this as a command to
apply the law of a party’s home country. Indeed, panels in fair use cases
do just that in 41% of the cases involving U.S. respondents. But, not all
panelists follow that route. Sometimes arbitrators in fair use cases apply
the law of their own country.141 U.S. panels do this in 47% of all cases
they hear; other panels do this in less than 12% of cases. More often,
however, arbitrators apply no law; instead, they invent or adopt UDRP
rules to govern particular factual situations. These different applications
show that panels in fair use cases have not developed a consistent
method of choosing nations’ laws. They have, though, cited U.S. law
more often than other nations’ laws in fair use cases. The question then
becomes, why? Part of the answer lies in the rules WIPO panels have
developed and applied in fair use cases.
2. WIPO Rules of Decision
In February 2005, WIPO published on its website a variety of rules
its panels have developed, entitled “WIPO Overview of WIPO Panel
Views on Selected UDRP Questions” (“Rules” or “WIPO Rules”).142
136 See infra notes 140–198 and accompanying text.
137 See infra notes 140–141 and accompanying text.
138 See infra notes 142–158 and accompanying text.
139 See infra notes 159–198 and accompanying text.
140 UDRP Rules, supra note 5, § 15(a); supra notes 66–73 and accompanying text.
141 No arbitrators apply their home country’s law when it differs from the law of the
parties’ respective countries.
142 WIPO Rules, supra note 67.
96 Boston College Law Review [Vol. 53:65
These Rules developed inductively—much like common law.143 For the
most part, they represent an informal consensus or agreement on how
to apply a rule in a given factual situation.144 Although the Rules are
not binding, panels frequently cite them: out of the eighty-three fair
use cases included in my sample since 2005, twenty-seven decisions (or
32%) explicitly cited WIPO’s Rules (discussed below). Given that these
Rules represent de facto precedents,145 a WIPO Rule favoring U.S. respondents
(or disfavoring other respondents) in fair use cases could
help explain U.S. Favoritism.
On certain issues, WIPO Rules have “split” into two competing
viewpoints.146 One set of these Rules addresses domain names used for
criticism websites (“Criticism Rules”).147 The second set of Rules relates
to fan sites (“Fan Site Rules”), websites about famous people created by
fans or supporters.148
The first Criticism Rule explains that “[t]he right to criticize does
not extend to registering a domain name that is identical or confusingly
similar to the owner’s registered trademark or conveys an association
While some of [the listed] issues arise only infrequently, all of them are, or
are perceived to be, relevant to the operation of the UDRP. On most of these
issues, consensus or clear majority views have developed. Certain other questions
attract a diversity of views. The WIPO Center’s identification of questions
and careful and conservative evaluation of opinions is based on some
20,000 UDRP cases it has administered through March 31, 2011.
Id.
143 Id. (“On most of these issues, consensus or clear majority views have developed. Certain
other questions attract a diversity of views.”) (emphasis added).
144 Id.
145 Mueller, supra note 17, at 19 (explaining that, in 2002, over half of the UDRP decisions
cite other UDRP decisions, and that “[t]he majority [of these decisions] not only
cite other cases but rely on precedent extensively to reason out and support their deci-
sions”).
146 E.g., Asset Loan Co. Pty Ltd v. Rogers, No. D2006-0300 (WIPO May 2, 2006),
http://www.wipo.int/amc/en/domains/decisions/html/2006/d2006-0300.html. The court
noted that, although
a number of distinguished Panelists take the view that a bona fide noncommercial
criticism site connected to a domain name featuring the trade
mark of a subject of the criticism is fair use of the domain name for the purposes
of the Policy[,]’ . . . Other panelists . . . have taken a contrary view.
Id. (quoting Bonneterie Cevenole S.A.R.L. v. Sanyouhuangong, No. D2001-1309 (WIPO
Jan. 21, 2002), http://www.wipo.int/amc/en/domains/decisions/html/2001/d2001-1309.
html).
147 WIPO Rules, supra note 67, § 2.4.
148 Id. § 2.5.
2012] An Empirical Analysis of Fair Use Decisions 97
with the mark.”149 The second Criticism Rule notes that, “[i]rrespective
of whether the domain name . . . connotes criticism,” the respondent
can make a noncommercial use of the mark in the domain name of a
criticism site.150 WIPO goes on to assert that “[i]n cases involving only
US parties or the selection of a U.S. mutual jurisdiction, panelists tend
to adopt the reasoning in [the second Rule] (though not universally).”151
It does not, however, explore this phenomenon further.
The other potentially relevant set of Rules involves fan sites. These
Fan Site Rules are similar to those WIPO has recognized in criticism
cases.152 The first Fan Site Rule states that domain names incorporating
trademarks can be distinguished either by other words or by the content
of the site.153 In other words, fan sites can avoid an adverse decision
if they are “clearly distinctive from any official site.”154 The second
Fan Site Rule states that any domain name that incorporates the
trademark is confusingly similar to the trademark.155 The number of
cases involving fan sites is small (n = 16). There are only three cases in
which respondents won (19%).156 Two of those cases—Société Anonyme
de La Colline Champel v. Angelini, in 2005, and White Castle Way, Inc. v.
149 Id. § 2.4 (“The right to criticize does not necessarily extend to registering and using
a domain name that is identical or confusingly similar to the complainant’s trademark.”).
150 Id. (“[T]he respondent has a legitimate interest in using the trademark as part of
the domain name of a criticism site if such use is fair and noncommercial.”).
151 Id.
152 See id. § 2.5.
153 WIPO Rules, supra note 67, § 2.5 (“View 1: The registrant of an active and noncommercial
fan site may have rights and legitimate interests in the domain name that includes
the complainant’s trademark. The site should be actually in use, clearly distinctive
from any official site, and noncommercial in nature.”).
154 Id.
155 Id.
View 2: A respondent does not have rights or legitimate interests in expressing
its view, even if positive, on an individual or entity by using an identical or
confusingly similar domain name, if the respondent is intentionally misrepresenting
itself as being (or as in some way associated with) that individual or
entity, or seeks to derive commercial advantage from its registration and use.
Also, where the domain name is identical to the trademark, panels have
noted that such respondent action prevents the trademark holder from exercising
its rights to the trademark and managing its presence on the Internet.
Id.
156 Estate of Francis Newton Souza v. ZWYX.org Ltd., No. D2007-0221 (WIPO June 18,
2007), http://www.udrpsearch.com/wipo/d2007-0221; Société Anonyme de La Colline
Champel v. Angelini, No. D2005-1059 (WIPO Dec. 5, 2005), http://www.wipo.int/amc/
en/domains/decisions/html/2005/d2005-1059.html; 2001 White Castle Way, Inc. v. Jacobs,
No. D2004-0001 (WIPO Mar. 26, 2004), http://www.wipo.int/amc/en/domains/de-
cisions/html/2004/d2004-0001.html.
98 Boston College Law Review [Vol. 53:65
Jacobs, in 2004—involved U.S. respondents.157 All three of these cases
applied the first Fan Site Rule.
These Rules explain at least part of U.S. Favoritism. Recall that, in
criticism cases, U.S. respondents won more often than other respondents.158
The data show that the Criticism Rules and Fan Site Rules
have been applied to and favor U.S. respondents, though the effect of
the latter Rules is not large. WIPO also has stated as much by recognizing
that the Criticism Rules exist and apply in cases involving some U.S.
party or panel. Although 43% of all fair use cases are criticism cases, the
Criticism Rules may help to explain U.S. Favoritism. The difference
between U.S. and non-U.S. respondent winning percentages in criticism
cases is not significant (p > 0.05). More importantly, though, the
differences between respondent winning percentages when U.S. and
non-U.S. arbitrators decide their case is significant (one-tailed, p =
0.007). Thus, while WIPO Rules can help explain U.S. Favoritism, the
Rules cannot explain everything. Indeed, how panels applied the Rules
is related to both panel and respondent nationality.
3. Panel Nationality
Panels seem to have created the speech-friendly Criticism Rule by
repeatedly applying U.S. law, or what they perceive to be U.S. law.159
Why, though, did panels import U.S. laws into the UDRP? The primary
explanation seems to be panel nationality. Out of the twenty-one fair
use cases in which U.S. law has been applied, only four times (19%) was
the panel from a country other than the United States.160 Additionally,
157 See Angelini, No. D2005-1059; Jacobs, No. D2004-0001.
158 U.S. respondents win thirty-eight percent of the time in criticism cases. Other respondents,
however, win only twenty-one percent of the time. See supra Figure 4.
159 Indeed, all five decisions cited by WIPO as examples of this Rule involve U.S. respondents
and U.S. panels applying U.S. law. WIPO Rules, supra note 67, § 2.4 (citing: (1)
Sutherland Inst. v. Continuative LLC, No. D2009-0693 (WIPO July 10, 2009), http://www.
wipo.int/amc/en/domains/decisions/html/2009/d2009-0693.html (complaint denied);
(2) Sermo, Inc. v. CatalystMD, LLC, No. D2008-0647 (WIPO July 2, 2008), http://www.
wipo.int/amc/en/domains/decisions/html/2008/d2008-0647.html (complaint denied); (3)
Howard Jarvis Taxpayers Ass’n v. McCauley, No. D2004-0014 (WIPO Apr. 22, 2004), http://
www.wipo.int/amc/en/domains/decisions/html/2004/d2004-0014.html (complaint denied);
(4) TMP Worldwide Inc. v. Potter, No. D2000-0536 (WIPO Aug. 5, 2000), http://www.
wipo.int/amc/en/domains/decisions/html/2000/d2000-0536.html (complaint denied); (5)
Bridgestone Firestone, Inc., v. Myers, No. D2000-0190 (WIPO July 6, 2000), http://www.wipo.
int/amc/en/domains/decisions/html/2000/d2000-0190.html (complaint denied)).
160 RapidShare AG v. Yanpeng, No. D2010-0893 (WIPO July 28, 2010), http://www.
wipo.int/amc/en/domains/search/text.jsp?case=d2010-0893; Ryanair Ltd. v. Coulston, No.
D2006-1194 (WIPO Dec. 12, 2006), http://www.wipo.int/amc/en/domains/decisions/
html/2006/d2006-1194.html; Sabanci v. iu, No. D2003-0498 (WIPO July 23, 2003), http://
2012] An Empirical Analysis of Fair Use Decisions 99
as noted, U.S. panels found in favor of fair use more than their foreign
counterparts in criticism cases.
Why might U.S. panels have done this? One explanation is sheer
opportunity. The United States has the highest percentage of arbitrators,
comprising 24% of all WIPO arbitrators. As noted in Part II, the
next four countries with the highest proportion of arbitrators are the
United Kingdom (10%), Australia (6%), Switzerland (5%), and France
(5%).161 Additionally, U.S. panels are deciding nearly 50% of all fair use
cases.
The parties’ nationality might also have influenced U.S. panels’
decisions to import U.S. law. Most cases decided by U.S. panels (75%)
involved both U.S. respondents and complainants. When you include
all cases that involve at least one U.S. party that number jumps to 88%.
Thus, in addition to the sheer number of cases U.S. panels have decided,
the parties involved may influence the decision to apply U.S. law.
U.S. panels primarily are responsible for importing U.S. legal principles
into the UDRP. And, typically, this occurs when resolving a dispute
involving U.S. parties. So the nationalities of the parties and the
panels seem to matter. Once imported, the law is used to decide a dispute.
This, in turn, results in a rule that reflects U.S. law, which WIPO
panels perceive as more protective of free speech than most other
countries.162
Although it is unclear whether that perception of U.S. law is accurate,
it seems likely to be. U.S. trademark law, for example, allows for
criticism, comment, and even parody of mark, though typically not
when the use of the mark is confusing.163 Of the thirty-three other
countries whose arbitrators decided cases in my study, it seemed that
the laws of most other countries did not attempt to further similar policies.164
There were, however, some exceptions. In Germany, for examwww.udrpsearch.com/wipo/d2003-0498;
Voigtländer GmbH v. John Voigtlander, No. D2003-
0095 (WIPO Apr. 18, 2003), http://www.wipo.int/amc/en/domains/decisions/html/2003/
d2003-0900.html.
161 See supra Figure 11.
162 This may be because U.S. trademark law allows for uses of marks in order to criticize
or comment on them, either under the fair use defense or a parody defense. 15 U.S.C.
§ 1125(c)(3)(A) (2006) (establishing a fair use defense for dilution when the use parodies
or criticizes the mark); id § 1125(c)(3)(B) (establishing a fair use defense for dilution for
news reporting and news commentary); David A. Simon, The Confusion Trap: Rethinking
a Defense for Parody in Trademark Law (forthcoming presentation).
163 See Simon, supra note 162.
164 To determine whether other countries had trademark laws that accommodated free
speech interests I consulted two sources: Ethan Horwitz, World Trademark Law and
Practice (2d ed. 2011) and Edward J. Fennessy, Trademarks Throughout the World
100 Boston College Law Review [Vol. 53:65
ple, one court has held that the German Constitution provides a defense
for satirical uses of trademarks.165 Other countries, like France,
exempt from liability noncommercial “informational” and “political”
uses of trademarks.166 Other countries, like Brazil, exempt from liability
“noncommercial,” “truthful” trademark uses, but also “prevent reference
to the mark in speeches or publications” when the reference is
commercial and causes “detriment to the trademark.”167
On the whole, most countries did not expressly exempt uses for
comment, criticism, parody, or satire. Even the European Union’s (EU)
Directive to harmonize European Community trademark law does not
specify any free speech protections for parody, commentary, or criticism.168
Indeed, at least one court in Europe (the Austrian Supreme
Court) has ruled that the EU Directive in conjunction with an EU Regulation
prohibits parodic uses of trademarks on commercial products.169
That said, the Directive does provide some safeguards, though they
are not robust. Article 3, for example, states that certain words cannot
be registered as trademarks, such as those without distinctive character.170
Likewise, Article 6 limits the rights of mark owners by allowing
for uses of a trademark to describe or compare products.171 Despite
these limited protections, some argue that the EU Directive has an ad(5th
ed. 2011). Using these two sources, I searched the law of each country for whether
the law permitted “expressive uses.” Loosely defined, this meant that I searched for whether
the law immunized from liability criticism, comment, parody, or satire of trademarks.
Because these sources provided mere synopses of many countries’ laws, it was difficult to
determine whether the law punished or exempted respondents from liability for expressive
uses. If the source did not affirmatively indicate that a country’s law immunized expressive
uses, I presumed that the law punished such uses.
165 2-7 Horwitz, supra note 164, § 7.01[4] (“It is also a defense to infringement if the
mark is used in a satirical way based on the constitutional freedom of art.”).
166 FRA-7 id. § 7.01[4] (“It is a defense to infringement that the mark was used for information
purposes and not for commercial or advertising purposes. The use of the mark
in a political statement is not infringement.”).
167 BRA-7 id. § 7.01[4] (noting that “[t]he owner of a mark may not . . . [p]revent reference
to the mark in speeches or publications, as long as it is done with no commercial
connotation and no detriment to the trademark”).
168 See Directive 2008/95/EC, Approximate the Laws of the Member States Relating to
Trade Marks, 2008 O.J. (L 299) 25 (EU).
169 Ferdinand Graf, Supreme Court Rules on Permissibility of Trademark Parodies, Int’l Law
Office (Apr. 26, 2010), http://www.glpi.com.br/templates/conteudo_geral_en.aspx?page=
2497&idiom=1 (noting that the Austrian Supreme Court held, on September 22, 2009, that
the STYRIAGRA mark used on pumpkin seeds and touted as aphrodisiac diluted the VIAGRA
mark and explaining how the Austrian Supreme Court distinguished the German court
decision allowing the use of the mark on parody grounds).
170 Council Directive 2008/95/EC, art. 3(1)(b), 2008 O.J. (L 299) 25, 27.
171 Id. art. 6, at 29. It is interesting to note that the Directive uses the word “unfair” five
times, but never once uses the word “fair.” See id.
2012] An Empirical Analysis of Fair Use Decisions 101
ditional safeguard: it limits liability only to those uses “in the course of
trade.”172
So, some panels developed competing rules based on their perception
of U.S. and non-U.S. law. As other panels draw on that rule, it
replicates and becomes more prevalent. Indeed, WIPO’s Criticism
Rules illustrate this fact.173 Although WIPO acknowledges that some
U.S. component plays a role in the Criticism Rules, it does not claim
that panels are applying U.S. law. Instead, the Criticism Rules are based
on panel decisions. Some of these decisions, of course, cited U.S. law.
In contrast, foreign law has not birthed any specific rule in fair use
cases. This is likely because panels have imported other countries’ laws
a total of six times, with only Spain’s laws referenced more than once.174
As a result, no other country’s law has influenced the application of the
UDRP to the extent as U.S. law. Or, perhaps all other countries’ laws
are just less speech protective; thus, they are the default UDRP Rule.
The outcome is a division in UDRP application: panels apply either
U.S. law or a sui generis UDRP rule.
Still, the question lingers: why have panels not imported foreign law?
Unfortunately, most panels do not even discuss the choice of law provision.
In only six cases did a panel discuss whether to apply a particular
country’s law.175 Sometimes, other cases would discuss why a potential
172 Id. art. 5, at 29 (providing rights “to prevent all third parties not having [the trademark
owner’s] consent from using [the mark] in the course of trade”).
173 WIPO Rules, supra note 67, § 2.4.
174 Consejería de Sanidad de la Comunidad de Madrid v. Morán Llanes, No. D2008-1400
(WIPO Nov. 20, 2008), http://www.wipo.int/amc/en/domains/decisions/html/2008/
d2008-1400.html; Canonais, S.L. v. Díaz Ros, No. D2000-0592 (WIPO Oct. 2, 2000), http://
www.wipo.int/amc/en/domains/decisions/html/2000/d2000-0592.html; Metro Bilbao, S.A.
v. Fernández, No. D2000-0467 (WIPO Aug. 31, 2000), http://www.wipo.int/amc/en/
domains/decisions/html/2000/d2000-0467.html.
175 Coast Hotels, Ltd. v. Lewis & UNITE HERE, No. D2009-1295 (WIPO Nov. 23, 2009),
http://www.udrpsearch.com/wipo/d2009-1295; Grupo Costamex, SA de C.V. v. Smith &
Oneandone Private Registration / 1&1 Internet Inc., No. D2009-0062 (WIPO Mar. 25,
2009), http://www.wipo.int/amc/en/domains/decisions/html/2009/d2009-0062.html.
Accordingly, the Panel is not disposed to apply either the law of the United
States of America (the country of domicile of the Respondents and the registrar)
or the law of Mexico (the Complainant’s country of domicile), in deciding
whether the Respondents’ website as it stood on June 12, 2008, constituted
a use of the Domain Name falling within paragraph 4(c)(iii) of the
Policy. Nor is the Panel attracted to the idea, proposed in the Sermo, Inc. case,
that conflicts of laws provisions should be applied to assist in the selection of
some particular national law by which to judge the concepts of ‘fair’ and ‘legitimate’
in paragraph 4(c)(iii) of the Policy. The interpretation of a standardform
contract (i.e., that between a domain name registrant and the domain
name registrar), the relevant terms of which are mandated by the Policy and
102 Boston College Law Review [Vol. 53:65
split arose in the Criticism Rules (e.g., the United States has a strong
tradition of free speech that other countries do not share).176 It did not,
however, discuss the choice of law provision or the merits of applying a
particular country’s law.
One explanation for the lack of foreign law, then, is thoughtless
panels. Foreign panels do not think that choice of law is an issue warranting
attention. But that explanation is not very helpful. It is still important
to know why panels seem indifferent to choice of law issues.
One arbitrator suggests that the reason is that panels from countries
other than the United States favor a “unified” or “self-referential”
UDRP.177 That is, panels favor a UDRP that develops and applies its own,
independent rules over one that relies on local laws.178 Why? Because
the UDRP and trademark laws serve different purposes: “[T]he Policy
and local laws as to cybersquatting and trademark infringement are different.
They do not do, and are not intended to do, the same thing.”179
have been designed to operate in a cross-border environment, should not differ
in accordance with the location or nationalities of the parties, the registrar,
or the panel, and nor should it depend on the location of the relevant
website.
Grupo, D2009-0062; see also 1066 Housing Ass’n Ltd. v. Morgan, No. D2007-1461 (WIPO Jan.
18, 2008), http://www.wipo.int/amc/en/domains/decisions/html/2007/d2007-1461.html;
Heineken España, S.A. v. Domingo González Ruiz, No. D2001-1202 (WIPO Dec. 12, 2001),
http://www.wipo.int/amc/en/domains/decisions/html/2001/d2001-1202.html; Arthur Guinness
Son & Co. (Dublin) Ltd. v. Dejan Macesic, No. D2000-1698 (WIPO Jan. 25, 2001), http://
www.wipo.int/amc/en/domains/decisions/html/2000/d2000-1698. html; Bridgestone Firestone,
Inc. v. Meyers, No. D2000-0190 (WIPO July 6, 2000), http://www.wipo.int/amc/en/
domains/decisions/html/2000/d2000-0190.html. Although the Grupo panel rejected the
choice of law issue, it went on to apply U.S. law’s initial interest confusion doctrine to find
bad faith. No. D2009-0062.
176 E.g., CBS Broad. Inc. v. Z. aghloul, No. D2004-0988 (WIPO Feb. 5, 2005), http://
www.wipo.int/amc/en/domains/decisions/html/2004/d2004-0988.html.
177 1066 Housing Ass’n, No. D2007-1461.
178 Id. (agreeing with a previous decision that found that, “‘[a]s a matter of principle, this
Panel would not have thought that it was appropriate to import unique national legal principles
into the interpretation of paragraph 4(c) of the Policy. This is so even if the effect of
doing so is desirable in aligning decisions under the Policy with those emerging from the
relevant courts and thus avoiding instances of forum shopping” (quoting Covance, Inc. &
Covance Labs. Ltd. v. The Covance Campaign, No. D2004-0206 (WIPO Apr. 30, 2004),
http://www.wipo.int/amc/en/domains/decisions/html/2004/d2004-0206.html)).
179 Id. (footnote omitted) (citation omitted). This statement is questionable, at least as
to existing cybersquatting laws. The U.S. cybersquatting law, for example, is nearly identical
to the UDRP. Compare 15 U.S.C. § 1125(d) (2006) (promoting cyberprivacy prevention
by creating civil liability for a bad faith intent to profit from the use of domain names that
are “identical or confusingly similar” to a distinctive or famous mark or is a protected
trademark), with UDRP, supra note 5, § 4(a) (defining an applicable dispute as one in
which the domain name is “identical or confusingly similar” to a trademark or service
2012] An Empirical Analysis of Fair Use Decisions 103
Beyond having different purposes, there is a worry about uniformity
and certainty: using local laws “risks the UDRP fragmenting into a series
of different systems, where the outcome to each case would depend
upon where exactly the parties happened to reside. That way chaos
lies.”180 Additionally, “bringing local law into the assessment of the Policy
. . . undermines the very goal of commercial certainty . . . .”181 Since
64% of panels do not apply any law, this explanation might have pur-
chase.
Explaining panels’ decisions not to import law, though, has a
shortcoming: virtually all of the fair use cases (96%) say nothing about
the choice of law provision. And, we wouldn’t expect them to if most
panels view the UDRP as self-referential. We would expect, as is the
case, most panels to apply WIPO Rules or precedents from other cases.
Put simply, panels do not think about this issue; they just use WIPO decisions
as de facto precedents. To them, the choice of law provision
gives license to select among WIPO Rules or precedents, not national
laws.
Additionally, in many cases the parties are from different countries
than the panel. In cases involving Australian panels, for example, the
panel and at least one party shared the same nationality only 20% of
the time. Low numbers also exist for other panels deciding large numbers
of cases, including the United Kingdom (50%), Canada (22%),
and New Zealand (14%). In the cases where the panel and none of the
parties share the same nationality, the panel likely does not know the
law of the parties’ countries. Lacking knowledge of the law, the panel
cannot apply it. This could explain why other panels do not apply foreign
law as much as U.S. panels.
Still, there is another, related explanation for panels’ reluctance to
import foreign law: most countries simply do not legislate cybersquatting.
If no law exists to import, this would explain why non-U.S. panels
have not applied it. In the sample of 148 cases, arbitrators, complainants,
and respondents represented thirty-four different countries.182
mark and it has been registered in “bad faith”). They are designed to do very similar, if not
the exact same, things. The question is how each achieves the same goal.
180 1066 Housing Ass’n, No. D2007-1461.
181 Id.
182 In alphabetical order, those countries were Australia, Bahamas, Belgium, Brazil,
Canada, China, Denmark, Finland, France, Germany, India, Indonesia, Ireland, Israel,
Italy, Jamaica, Japan, Luxemburg, Mexico, Netherlands, New Zealand, Nigeria, North Korea,
Norway, Panama, Poland, Russia, Slovenia, Spain, Switzerland, Turkey, United Kingdom,
United States, and West Indies (Saint Kitts & Neviss).
104 Boston College Law Review [Vol. 53:65
Of these thirty-four countries, only seven (20%)183 had legislation relating
to domain names:184 the United States,185 Belgium,186 Finland,187
Japan,188 Russia,189 Spain,190 and Italy.191 Only some of these countries’
183 To find whether a country had national law relating to domain names, I used WIPO’s
searchable database of national laws and treaties on intellectual property. WIPO Lex, World
Intellectual Prop. Org., http://www.wipo.int/wipolex/en/ (last visited Dec. 20, 2011). I
searched each country individually, specifying that the “subject matter” field should contain
only “domain names.” I also consulted a text on Internet law. See generally Smith, supra note
53. Finally, I attempted to locate any online resource that indicated whether a country had
legislation addressing domain names specifically. Sometimes this meant consulting the websites
of various countries (that list the country’s laws). For every country, I attempted to locate
the official government website containing the text of all laws. When successful, I cited
the source in the footnotes to this Article.
184 “Relating to” does not mean that the country had specific cybersquatting laws akin
to the Anticybersquatting Consumer Protection Act. 15 U.S.C. § 1125(d) (2006).
185 Id.
186 Loi relative à l'enregistrement abusif des noms de domaine [Act on the Abusive
Registration of Domain Names] of June 26, 2003, Moniteur Belge [M.B.] [Official Gazette
of Belgium], Sept. 9, 2003, 45225, available at http://www.wipo.int/wipolex/en/text.jsp?
file_id=187949; see Smith, supra note 53, at 206–09 (stating that the Act on the Abusive
Registration of Domain Names created a cause of action for cybersquatting that applies
only to “abusive registration of domain names by third parties who have no rights in the
distinctive signs specified in the act,” but not to “freedom of speech disputes”).
187 Finnish law includes the Domain Name Act, which applies only to the Finnish
country-code top-level domain (ccTLD), “.fi.” Verkkotunnuslaki [Domain Name Act],
translation available at http://www.finlex.fi/en/laki/kaannokset/2003/en20030228.pdf; see
also Smith, supra note 54, at 229–31 (describing the act).
188 Fusei kyousou boushihou [Unfair Competition Prevention Law], Law No. 47 of
1993, art 2(xii) (Japan), translation available at http://www.wipo.int/wipolex/en/text.jsp?
file_id=128372 (defining unfair competition to include “acquiring or holding” domain
names that are confusingly similar to a markholder or that otherwise cause injury to the
markholder); id. art. 5(3)(iv) (providing a damages remedy for trademark holders whose
rights have been infringed by a domain name); see also Brent T. Yonehara, landoftherisingsun.co.jp:
A Review of Japan’s Protection of Domain Names Against Cybersquatting, 43 IDEA 207,
221–24 (2003) (describing Japan’s cybersquatting law).
189 In 2002, Russia amended its trademark law to include protection for trademarks
used in domain names. Mariyetta Meyers, Russia and the Internet: Russia’s Need to Confront
and Conquer Trademark Infringement in Domain Names and Elsewhere on the Web, 9 Gonzaga J.
Int’l L. 200, 213–16 (2006).
190 Law 17/2001 of December 7, 2001, on Trademarks arts. 34(2), (3)(e) (B.O.E. 2001,
23093) (Spain), translation available at http://www.wipo.int/wipolex/en/text.jsp?file_id=
126736 (providing a registered trademark owner with the right to prohibit use of its mark
in a domain name in instances that meet Spanish infringement standards under art.
34(2)(a)–(c)).
191 See Decreto Legislativo 10 febbraio 2005, n. 273, in G.U. 4 marzo 2005, n. 52, art. 22
(It.); see also Smith, supra note 54, at 257 (explaining that Article 22 of Italy’s Industrial
Property Code “establishes that it is forbidden for an undertaking to use the sign or domain
name of the trademark of another if the activities of the parties are either similar or
dissimilar . . . so as to lead consumers to believe that there is a possible link between the
two undertaking[s]”).
2012] An Empirical Analysis of Fair Use Decisions 105
legislation was directed at only cybersquatting. Two other countries—
France and Norway—have laws addressing domain names. Both laws,
however, became effective on a date outside that covered by my sampled
data.192
At first glance, that fact aligns with this explanation. With the U.S.
being one of only four countries with cybersquatting laws, there just are
not many laws to import.193
That explanation, however, is not totally convincing. Recall that the
most frequently cited foreign law was Spain’s (cited three times). In the
three other cases citing foreign law, the panel cited a European treaty,194
192 France’s law, which applies only to “.fr,” “.re,” “.mq.,” “.yt,” and a few other top-level
domain name (TLDs) (but not gTLDs) went into effect on August 4, 2011. Loi 2011-302
du 22 mars 2011 [Law 2011-302 of March 22, 2011], Journal Officiel de la République
Française [J.O.] [Official Gazette of France], Mar. 23, 2011, p. 5186, art. 19(v), available
at http://www.legifrance.gouv.fr/affichTexteArticle.do;jsessionid=5008CEF4C7F4E
F1FC34ADA73470C14BB.tpdjo04v_1?idArticle=LEGIARTI000023752116&cidTexte=LEG
ITEXT000023752099&dateTexte=20111116; Décret 2011-926 du 1 août 2011 relatif à la
gestion des domaines de premier niveau de l'internet correspondant aux codes pays du
territoire national [Decree 2011-926 of August 1, 2011 on the Management of Top-level
Domains of the Internet Country Code Corresponding to the National Territory], Journal
Officiel de la République Française [J.O.] [Official Gazette of France], Aug.
3, 2011, p. 13294, arts. R. 20-44-31 to R. 20-44-44, available at http://www.legifrance.
gouv.fr/affichTexte.do?cidTexte=JORFTEXT000024425287&dateTexte=&categorieLien=
id. The legislative part of the Code was adopted as part of a March 22, 2011 law; government
norms were adopted on August 4, 2011. Thus, the operative date of all provisions is
August 4, 2011. (Thanks to Cedric Manara for this information.) Specifically, these laws
provide that a domain name can be withheld or deleted when it is “[l]ikely to infringe
intellectual property rights or personality, unless the applicant provides proof of a legitimate
interest and is acting in good faith.” L. 45-2-2. A legitimate interest can include “a
non-commercial use of the domain name or a name related with no intention of misleading
the consumers or harming the reputation of a name which is recognized or established
a right.” R. 20-44-43.
Norway’s law became effective on July 1, 2010. Lov om varemerker [The Trademarks
Act], translation available at http://www.patentstyret.no/Global/Filarkiv/regelverk/Trade
marks_Act_valid_01072010.pdf. The relevant language states: “The court may also decide
on other action to prevent new trademark infringements, including that the infringing
party shall take steps to delete or assign to the injured party any domain names used in
contravention of the trademark right.” Id. § 59. Norway’s previous trademark law went into
effect on March 3, 1961. Id. § 83. I could not determine whether and when the 1961 law
was amended. Because my study included WIPO decisions through September 2010, the
French and Norwegian laws are not relevant.
193 See also Przemyslaw Paul Polanski, Customary Law of the Internet: In the
Search for a Supranational Cyberspace Law 322 (2007) (“With the exception of few
countries such as the United States, no [specific, statutory] rights have been created for
defending interests in a domain name.”).
194 Moss & Coleman Solicitors v. Kordowski, No. D2006-1066 (WIPO Oct. 23, 2006),
http://www.wipo.int/amc/en/domains/decisions/html/2006/d2006-1066.html (citing Article
10 of the European Convention on Human Rights in the context of free speech).
106 Boston College Law Review [Vol. 53:65
EU law,195 and United Kingdom trademark law.196 None of the laws cited
came from countries with cybersquatting statutes. In those cases, the
panels cited trademark or unfair competition law, which tends to be—or
at least is perceived as—less friendly to speech than U.S. trademark law.
That is not surprising considering that, even in countries that do not
have specific cybersquatting statutes, courts have applied traditional
trademark and unfair competition laws to resolve domain-name disputes.197
Thus, although most countries have law that could be applied
to domain-name disputes, not many arbitrators apply it.
In light of this fact, panel nationality seems to best explain the lack
of foreign law in the UDRP. Many foreign panels do not decide disputes
with any parties from their home country. For this reason, they
likely do not know the law of either party’s home country.198 Without
such knowledge, they can apply either their home country’s law or the
UDRP. Nearly all elect the latter course and apply only what they know,
or what they can make up as needed.
B. Reforming the UDRP
Some scholars state that “the most basic question about the bordered
Internet . . . is whether speech should be regulated globally or
195 Espirito Santo Fin. Grp. S.A. v. Colman, No. D2001-1214 (WIPO Jan. 22, 2002), http://
www.wipo.int/amc/en/domains/decisions/html/2001/d2001-1214.html (citing European
Court of Justice case C-63/97, BMW v. Deenik, 1999 E.C.R. I-905, 1 C.M.L.R. 1099 (1999)).
196 Lanware Ltd. v. Iveson, No. D2004-1061 (WIPO Feb. 9, 2005), http://www.wipo.
int/amc/en/domains/decisions/html/2004/d2004-1061.html (citing U.K. law on passing-
off).
197 Smith, supra note 53, at 200–04, 210–54, 258–97 (explaining how traditional
trademark and unfair competition law have been applied in Australia, Canada, China,
France, Germany, Hong Kong, Israel, Japan, the Netherlands, New Zealand, Singapore,
Spain, Sweden, and Switzerland); Antonella Carminaati & Alexandre da Cuha Lyrio,
Trademarks v. Domain Names—The Brazilian Experience, CBSG.com.br, http://www.cbsg.com.
br/pdf_publicacoes/trademarksxdomain_names.pdf (last visited Dec. 20, 2011) (describing
various cases applying traditional legal principles, including trademarks and domainname
disputes); Swiss Legal Practice (1999–2005), trademark.ch, http://www.trademark.
ch/?sub_id=136&leng=1 (click the hyperlinked word “link” to obtain the abstracts of these
decisions in German) (last updated Oct. 16, 2007) (listing cases involving domain names
decided by Swiss courts).
Additionally, many of these countries have UDRP-like procedures to settle cybersquatting
disputes involving domain names using their respective country’s ccTLD. In Australia,
for example, the .au Dispute Resolution Policy (auDRP) provides more protection for
trademark holders. Complaints can be brought for “business names or other names, not
just trade marks and service marks.” Smith, supra note 53, at 202.
198 Also, the complaints filed generally do not rely on local law. Therefore, the arbitrators
are not provided with the relevant provisions to interpret.
2012] An Empirical Analysis of Fair Use Decisions 107
locally.”199 My study has shown that, in fair use UDRP cases, WIPO panels
have answered this “basic question” without addressing its normative
component: in cases where the respondent is from the United
States, panelists are more likely to apply U.S. law, or what they perceive
to be U.S. law. In other words, U.S. panelists are more likely to apply
local law when the dispute involves U.S. respondents. Yet, in cases
where the respondent is from any other country, the odds of a panel
applying that country’s law are low.
Despite having inadvertently answered this question, U.S. Favoritism
does not provide a structured way of thinking about which laws to
apply—local, U.S., or international—in UDRP fair use disputes.
1. Problems with the U.S. Law/UDRP Division
This U.S. law/UDRP division in fair use cases is problematic for
several reasons. First, it privileges U.S. law without justification.200 Why,
for example, should panels import U.S. law but not French law into the
UDRP? One arbitrator has remarked:
There is something rather curious . . . [about the existence of
a] US approach and a non-US approach . . . [to criticism websites].
Laws protecting . . . freedom of speech are not unique
to the United States. They exist in many countries and frequently
form part of those countries basic constitutional rules
. . . . If local US law is to influence the issue in the United
199 Goldsmith & Wu, supra note 44, at 150. In addition, countless others have devoted
their scholarship to this question. E.g., Matthew Burnstein, A Global Network in a Compartmentalised
Legal Environment, in Internet: Which Court Decides? Which Law Applies?
23, 24 (Boele-Woelki & Kessedjian eds., 1999); David R. Johnson & David Post, Law and
Borders—The Rise of Law in Cyberspace, 48 Stan. L. Rev. 1367, 1370–76 (1996); Jonathan
Zittrain, Be Careful What You Ask For: Reconciling a Global Internet and Local Law, in Who
Rules the Net?: Internet Governance and Jurisdiction 13, 14 (Thierer & Crews eds.,
2003).
200 One professor wonders whether we might justify this rule on the expectations of
the parties: “If U.S. law is mainly privileged in disputes involving U.S. participants then,
isn’t the justification arguably that the imposition of U.S. legal principles best meets the
expectations of the parties? Is there anything wrong with this result in practice?” E-mail
from Jacqueline D. Lipton, Professor of Law, Case Western Reserve Univ. Sch. of Law, to
David A. Simon, Summer Academic Fellow, Harvard Law Sch. (Apr. 11, 2011, 12:31 GMT)
(on file with author). This explanation is problematic for two reasons. First, it is circular. It
assumes the validity of an expectation that is based on a practice of privileging U.S. law.
But that says nothing about whether a privilege is warranted in the first instance. Second,
and relatedly, the expectation that U.S. law governs when U.S. parties resolve UDRP disputes
suggests that the opposite should hold true: that non-U.S. law would govern when
other parties resolved UDRP disputes. But that simply has not happened.
108 Boston College Law Review [Vol. 53:65
States, then logically, local laws not only can, but should, do so
elsewhere in the world. And yet for the most part, panels in
other jurisdictions have not considered it necessary to take into
account their local laws in coming to their views on this
particular issue.201
This arbitrator questions whether there is a reason for importing only
U.S. law.202 But even if panels had developed a rule or reason justifying
this practice, the UDRP would not help us resolve whether we should
accept the rule. This is because neither the UDRP nor WIPO take a
stance on applicable law. The UDRP is completely silent on the subject.
WIPO Rules, on the other hand, equivocate even on when the U.S.based
rule should apply.203
This lack of clarity illustrates another reason why the U.S.
law/UDRP division is problematic: arbitrators lack a principled way of
deciding when, if ever, to apply law, WIPO rules, or simply use their
own judgment. The data bear this out. Panels in fair use cases apply
U.S. law 30% of the time, foreign law 6% of the time, and no law (or sui
generis rules) 64% of the time. Additionally, since WIPO published its
Rules in 2005, panels have cited them in 32% of fair use cases.
This decisional freedom produces biased or inconsistent decisions.
Inconsistent decisions threaten individuals’ and business’ ability to rely
on the UDRP as a predictable form of dispute resolution. Such uncertainty
can have beneficial or deleterious consequences. On the one
hand, applying narrow speech protections to domain names critical of
a trademark holder can chill or skew speech. That result is orthogonal
to the purpose of the fair use provision: to provide a safety valve for
speech. On the other hand, applying U.S. law to UDRP disputes involving
parties from speech-restrictive countries can increase the amount of
201 1066 Housing Ass’n, No. D2007-1461. Matthew Harris from the United Kingdom decided
this case as the sole panelist. Id. He also said:
In short, the act of bringing local law into the assessment of the Policy when
taken to its logical conclusion as a matter of practice undermines the very
goal of commercial certainty identified in the Howard Jarvis case and with
which this Panel whole heartedly agrees. There may be cases where it is inevitable,
but in general it should be resisted. . . . [S]o far as the specific issue
of rights and legitimate interest are concerned[,] the Panel sees no reason to
import national rules and good reason not to do so.
Id.
202 Id.
203 WIPO Rules, supra note 67, § 2.4 (stating that “[i]n cases involving only U.S. parties
or the selection of a U.S. mutual jurisdiction, panelists tend to adopt the reasoning in [the
U.S.-based rule] (though not universally)”).
2012] An Empirical Analysis of Fair Use Decisions 109
speech online. The result, the data show, typically is related to respondent
and arbitrator nationality.
Finally, this U.S. law/UDRP division threatens to undermine the
UDRP’s legitimacy—if not in the eyes of certain trademark holders,
then at least in the eyes of non-trademark holders (i.e., Internet users).204
If the UDRP prides itself as representing international standards,
it fails to do so by creating distinct rights for U.S. respondents
and everyone else. On the one hand, this approach may benefit more
speakers online, but only if they are from the United States. Other
speech, regardless of what each country’s law has to say, will be provided
less protection. On the other hand, this approach may encourage
corporations and other large trademark holders to bring more UDRP
actions when someone criticizes them online and is not from the United
States. So, in the end, the UDRP would promote two contradictory
positions based on a rather simplistic conception of both U.S. and nonU.S.
law: provide breathing space for speech originating in the United
States while simultaneously encouraging its suppression when it originates
elsewhere. All this happens simply because of an oversimplified
view of speech protections for trademark users in various countries
around the world.
What exactly this will mean for the future of the UDRP is unclear.
If UDRP decisions privilege United States law, other entities (e.g., governments
and corporations) may become disillusioned with its perceived
unfairness or applicability to disputes involving parties from
their countries. If the entities that depend on the UDRP view it as unfair
or inapt, they may opt out205 of the system or seek to change it. This
policy may also alienate Internet users who expect certain speech protections,
but do not receive them because of the U.S./non-U.S. dichotomy.
Perhaps none of this will happen immediately; but as disparities in
treatment between parties grow, and as more countries and users hop
online, the chances only increase.
204 Helfer & Dinwoodie, supra note 72, at 230 (“Panel decisions that diverge widely
from the text of the UDRP, from each other, or from national laws, risk undermining the
UDRP’s legitimacy.”).
205 The UDRP, unlike other entrenched legal institutions, is optional in two basic
senses. First, any party wishing to cancel or transfer domain names can do so in a court of
law; using the UDRP is optional. Second, the UDRP provides a contract-based remedy. As
such, it depends on the continued existence of the contract between ICANN and various
registrars. ICANN—although heavily influenced by the United States—is responsive to
concerns of various parties, particularly national governments. As countries become dissatisfied
with the UDRP, they will opt-out of its procedures. Either they will sue in court, or, if
court-based options become too expensive, they will work to change the UDRP.
110 Boston College Law Review [Vol. 53:65
2. Four Potential Reforms to the UDRP
Although the UDRP is not operating perfectly, we could leave it
alone—which is what it appears ICANN will do. In this scenario, we
would expect U.S. law to continue to dominate UDRP fair use cases,
foreign law to be a nonfactor, and WIPO Rules to solidify the distinction
between these Rules and U.S. law. That is one possible path, but, as
noted above, there are costs for following it. For one, the current application
of the UDRP seems to undercut the policy’s international purpose.
After all, there is no principled reason to privilege U.S. law over
other countries’ laws. That is, if U.S. law is privileged in cases where the
parties are from the United States, other countries’ laws should be given
similar deference when deciding disputes involving non-U.S. res-
pondents.
If we want a principled solution, four options exist. The first three
options involve choice of law—they address U.S. Favoritism by altering
what law applies under UDRP. The last option is substantive; it suggests
one methodological approach panels could use to resolve certain kinds
of domain-name disputes that involve free speech. This last substantive
option could be used in conjunction with any of the three choice of law
options.
First, ICANN could prohibit panels from referencing or applying
any law. There are two problems with this approach. First, without reference
to any law, the UDRP may become increasingly unpredictable. If
panels simply apply ad hoc interpretations of UDRP provisions, prospective
domain-name holders cannot determine their ability to use a domain
name ex ante. Second, divorcing the UDRP from national law
would be impracticable. The UDRP depends upon a protectable trademark
interest, which is a legal right created by national law.206
Despite this risk of ex ante uncertainty, several facts ameliorate this
concern. The UDRP already operates on a de facto precedent system.
Most panels apply past decisions, not law. Moreover, at least in UDRP
disputes before WIPO, the provider has recognized informal rules—
such as the Criticism Rule—that both respondents and complainants
might rely on. There is no reason to suspect these practices would disappear
if we eliminated references to law. Indeed, it might actually encourage
their use.
Even so, the mere fact that panels cite other decisions does not
ensure that a desirable pattern will emerge—or, even if it does, that it
206 Thanks to Jacqueline Lipton for this point.
2012] An Empirical Analysis of Fair Use Decisions 111
will prevent panels from “changing” precedents for the worse. Indeed,
although the UDRP may be de facto precedential, panels cite applicable
WIPO Rules in only 32% of all fair use cases. Additionally, a no-law
system assumes that prohibiting the use of national law will successfully
eliminate it from the UDRP. Disgorging law from the UDRP entirely,
however, is impracticable. Arbitrators’ worldviews—their conceptions of
fairness and justice—drive their conclusions. Shaping these worldviews
are the culture and laws of the countries in which they reside. Thus, we
cannot totally separate particular countries’ legal norms from entering
UDRP decisions. We should expect “uniform” UDRP to reflect arbitrators’
different worldviews and, therefore, the law of their home countries.
Minimizing this effect would require a more substantive revision
of UDRP provisions.
A no-law provision is not the only solution. ICANN also could take
the opposite approach and implement a formal choice of law provision.
ICANN would have to articulate a procedure for selecting a particular
country’s law. Requiring panels to import their countries’ laws would
balance the interests of all participating countries. Countries have different
conceptions of free speech and different trademark laws.207 A
provision that accounts for these differences treats each country’s law
equally. Additionally, it provides some measure of predictability and
legitimacy to the UDRP.
Like other approaches, this one is not impermeable to criticism.
One problem relates to whose law applies. The law of the panel’s home
country? The respondent’s? The complainant’s? Some combination?
Picking one of these options just raises more questions. Why should the
law of the respondent’s home country apply, for example, when a domain
name can cause “harm” to the trademark owner anywhere in the
world? The other problem relates to how panels are selected. Should
panels of particular nationality hear certain disputes? The answer to
this question likely will depend on whose law governs. In cases where a
panel must apply Finnish law, for example, it would make sense for a
Finnish arbitrator to hear the dispute.
207 Graeme B. Dinwoodie, (National) Trademark Laws and the (Non-National) Domain
Name System, 21 U. Pa. J. Int’l Econ. L. 495, 511 (2000).
[There are a] wide range of free speech notions that exist around the world
once we move beyond [certain] core principles. And the UDRP system does
not presently require us to venture into the murky waters of enforcement because
the remedies (transfer of domain name) are limited to those that can
be effectuated by the registrars without the aid of national courts.
Id.
112 Boston College Law Review [Vol. 53:65
Third, ICANN could require all panels to apply only U.S. law to all
UDRP disputes. The advantages and drawbacks of this approach are
straightforward. For respondents, this approach would increase their
ability to use trademarks in their domain names. It also would increase
clarity in UDRP fair use cases and, because the U.S. has a cybersquatting
law (the Anticybersquatting Consumer Protection Act, or the “ACPA”),208
presumably in other types of cases as well.
But the drawbacks are substantial. Complainants would oppose the
change because it produces less favorable outcomes for them. Currently,
complainants enjoy an 83% win rate in fair use cases in which the
respondent is not from the United States.209 When the respondent is
from the United States, however, the complainant’s win rate drops
18%, to 65%.210 We see the same trend when we examine arbitrator—as
opposed to respondent—nationality. Complainants win 80% of the
time when a non-U.S. panel resolves the dispute.211 But that percentage
falls by 15%, to 65%, when a U.S. panel resolves the dispute.212 Because
it seems that U.S. law partially explains this phenomenon, using only
U.S. law would likely favor respondents.
Furthermore, the proposal would alienate other countries. If only
U.S. law applied, the UDRP would become a “mini” U.S. tribunal. But
ICANN did not adopt the UDRP so panels could apply U.S. law to all
domain-name disputes. The purpose of the UDRP is to provide an international
dispute-resolution process. Using only U.S. law—or worse,
perceptions of U.S. law—to decide disputes undermines the international
application of the UDRP. Indeed, arbitrators from other countries
would be unnecessary; only U.S. panels would be required to apply
U.S. law. For these reasons, the U.S. law approach would likely meet
strong resistance.
The last reform is one that focuses on remedies, rather than choice
of law. Currently the UDRP gives panels who find for the complainant
the power to cancel or transfer the respondent’s domain name.213 Under
this proposal, however, the UDRP would limit the complainant’s
208 15 U.S.C. § 1125(d) (2006).
209 See supra Figure 2.
210 See supra Figure 2
211 See supra Figure 3.
212 See supra Figures 2, 3.
213 UDRP, supra note 5, § 4(i) (“The remedies available to a complainant pursuant to
any proceeding before an Administrative Panel shall be limited to requiring the cancellation
of your domain name or the transfer of your domain name registration to the com-
plainant.”).
2012] An Empirical Analysis of Fair Use Decisions 113
remedy to only cancellation when the website implicates free speech
interests.”
This new proposal has two benefits. First, it is flexible. We can integrate
this approach with any of the aforementioned approaches. Second,
it protects speech space rather than individual defendants per se. By preventing
complainants from controlling unused “free speech domains,”
this option preserves that space for people to express themselves. Take, for
example, the domain name <http://livingspacessucks.com>214 that resolves
to a website that states in large print, “This website is for sale!”
Underneath this, further text reads, “The future blog for consumers to
voice their frustrations with Living Spaces!” The website seems to be designed
solely for resale, albeit to gripers. The noncommercial fair use
defense does not seem to apply to this respondent. It could, however, apply
to a subsequent griper. Yet, the complainant who succeeds in a cybersquatting
claim under the UDRP can request that the panel transfer
the domain name to it. That transfer, however, essentially closes off space
for free speech: the opportunity to use the domain name for a critical,
creative, or informational message is no longer available. A remedy that
allows for cancellation but not transfer preserves the space in which free
speech can be made. Thus, the cancellation-only policy would operate
much like the current domain-name system—the principal difference
would be that sites invoking speech interests would not automatically be
transferred to successful complainants.
Because the current UDRP is designed for a first-come, first-served
domain-name registration system, this proposal seems to face difficulties.215
In particular, a cancellation-only remedy leaves the domain
name open to further abusive registrations. We have, in other words,
the “whack-a-mole” problem: shut down one cybersquatter and another
one registers the identical domain again. But characteristics of potential
cybersquatters may actually mitigate, rather than increase, this risk.
We might, for example, expect prospective cybersquatters to know
something about domain names, domain-name registration, and do-
214 The website “www.livingspacessucks.com” was taken down after a UDRP complaint
was filed. Easy Win for Living Spaces UDRP?, domainmonster.com (July 27, 2011), http://
www.domainmonster.com/editorials/easy-udrp-win/; see also Case Details for WIPO Case D2011-
1262, World Intellectual Prop. Org., http://www.oapi.wipo.net/amc/en/domains/
search/case.jsp?case_id=21376 (last visited Dec. 20, 2011).
215 Repeat cybersquatters will be less of a problem under the new gTLD registration
system, which allows registration of gTLDs by only “[e]stablished corporations, organizations,
or institutions” and requires a review and objection process. See gTLD Applicant
Guidebook, supra note 87, § 1.2.1.
114 Boston College Law Review [Vol. 53:65
main-name cancellation or transfer under the UDRP.216 Thus, cancellations
of domain names like <[trademark]sucks.com> should deter
future abusive registrations, not encourage them.
How much a cancellation-only remedy deters potential cybersquatters
depends on the aggressiveness and reputation of the trademark
owner. The more valuable the mark’s reputation, the more likely the
mark owner will pursue a UDRP action. Likewise, wealthy trademark
owners are more likely to pursue UDRP action than moderatelyendowed
ones. Thus, a cancellation-only remedy may increase trademark
owners’ need to police their marks. Nevertheless, they likely will
continue to do so for the same reasons they have in the past.217
That is not to say the costs to mark owners will be insignificant in
all cases. The whack-a-mole problem, however, may be a cost public policy
demands the trademark holder to bear. Trademark owners, after all,
are filing UDRP complaints in any case—and they seem to be in a better
position than the registrar to monitor the use of their marks. If the
costs to trademark owners are too high, there are other available options.
Some kind of burden-shifting approach, for example, might be
workable in cases of repeat, abusive registrants.
Whatever the precise solution, the purpose of this Article is not to
propose it; it is to point out two facts. First, a cancellation-only remedy
for certain cases better protects speech-interests than a transfer.
Second, while this approach faces difficulties, solutions exist to deal
with them.
All of the aforementioned approaches have benefits and shortcomings.218
Doing nothing is easy, but it risks delegitimizing the UDRP
and creating a further divide between U.S. and other countries. Applying
U.S. law has the same—but probably stronger—effects. Excluding
law from UDRP decisions altogether may engender new rules, but
216 It is not unreasonable to assume that potential cybersquatters have been subject to
UDRP action in the past—or, if not, that they at least know something about their potential
liability under the policy.
217 It may turn out, though, that trademark owners decide that policing these speech
spaces under a cancel-only policy is too costly. If that happens, we might worry about
speech space being clogged by cybersquatters. Assuaging that fear, however, should be the
fact that most individuals who use domain names incorporating trademarks in a noncommercial,
fair way are not able to purchase the domain names for a large sum. Thus, the
market price for these domain names should be only marginally higher than the cost of
registration.
218 One might wonder why parties worried about U.S. Favoritism simply do not select
non-U.S. arbitrators. One reason is that most fair use cases are decided by only one arbitrator.
In any case, that does not solve the problem of the U.S./non-U.S. dichotomy. It merely
allows parties to choose between two existing options.
2012] An Empirical Analysis of Fair Use Decisions 115
these rules are likely to reflect the law and customs of the panel’s home
country. A meaningful choice of law provision would balance nations’
interests, but doing so simply and effectively seems difficult. Given
these problems, which approach is best in fair use cases?
3. ICANN Should Revise the UDRP to Apply Local Law to UDRP Cases
At least in fair use cases, ICANN should revise and clarify both the
panel assignment process and the UDRP’s choice of law provision. The
new choice of law provision should specify that a panel apply the law of
the respondent’s home country. The new panel assignment procedure
should require that arbitrators share the same nationality as the respondent.
In practice, this would not greatly affect U.S. respondents.
WIPO’s de facto policy assigns U.S. panels to U.S. respondents. But for
other respondents, WIPO’s assignment pattern is not clear. It is in this
realm that the policy would have the greatest effect.
Although this proposal requires adopting two separate provisions,
the proposed choice of law rule and panel assignment procedure are
tied together. Indeed, to understand the proposal, one must understand
the relationship between these two mechanisms. This Section
starts by discussing panel selection and then discusses applicable law.
Before entering this discussion, it must be made clear that this Article is
not arguing about how to govern the Internet with law or some other mechanism.
That topic has been debated feverishly.219 It is, however, arguing
that the UDRP should incorporate local law in fair use disputes.220
219 See generally James Boyle, Foucault in Cyberspace: Surveillance, Sovereignty, and Hardwired
Censors, 66 U. Cin. L. Rev. 177 (1997) (arguing that technological solutions offer a
way for the state to exert legal power over the internet); Dan L. Burk, Federalism in Cyberspace,
28 Conn. L. Rev. 1095 (1996) (arguing that the due process clause of the Fourteenth
Amendment and the Dormant Commerce Clause are constitutional limitations on individual
states' ability to regulate internet activity); Jack L. Goldsmith, Against Cyberanarchy,
65 U. Chi. L. Rev. 1199 (1998) (arguing that internet regulation is at least “feasible and
legitimate from the perspective of jurisdiction and choice of law”); Johnson & Post, supra
note 199 (suggesting that regulating cyberspace will require modifying our understanding
of law and its connection to physical boundaries); Joel R. Reidenberg, Governing Networks
and Rule-Making in Cyberspace, 45 Emory L.J. 911 (1996) (suggesting that, in the context of
the global information infrastructure, “a new ‘network governance paradigm’ must
emerge to recognize the complexity of regulatory power centers . . . and shift the role of
the state toward the creation of an incentive structure for network self-regulation”).
220 Other scholars have suggested different reforms to the UDRP. One scholar, for example,
has suggested that the UDRP can take one of three approaches: apply U.S. legal
principles, apply the laws of all countries, or apply UDRP sui generis law. See generally Laurence
R. Helfer, Whither the UDRP: Autonomous, Americanized, or Cosmopolitan?, 12 Cardozo
J. Int’l & Comp. L. 493, 494 (2004). Others have argued for a wholesale reworking of the
116 Boston College Law Review [Vol. 53:65
More specifically, this Article proposes a regime that is appropriate given
my study of fair use cases at WIPO. Therefore, the proposal applies
only to fair use cases. Although this proposal could apply to the UDRP
generally, scholars should study the other provisions more broadly before
making changes wholesale.221
Panel selection and assignment under the UDRP has only a few
rules. As noted previously,222 the complainant and respondent can offer
lists of arbitrators in cases where a three-member panel decides the dispute.
The provider assigns two arbitrators from the lists submitted by
the parties—one from each party. It then assigns one arbitrator on its
own. In a one-member panel, the provider assigns the arbitrator. How
providers assign specific arbitrators is unknown. The UDRP does not
state how this does or should occur, and providers have not been forthcoming
with information.
The data, however, show two patterns of assignment in fair use disputes:
one for U.S. panels and one for other panels. U.S. panels almost
always decide cases with at least one U.S. party. Indeed, in 88% of all
fair use disputes decided by U.S. panels, at least one party was from the
United States.223 That means that in 12% of all fair use disputes decided
by U.S. panels, the panel had absolutely no connection to the jurisdiction
of either party.224
For fair use cases decided by other panels, this problem is particularly
acute.225 Of all the cases decided by Australian panels, for example,
80% did not involve any party from Australia.226 The numbers are
similar for other countries. For all fair use cases involving panels from
UDRP to be more speech protective. E.g., Dawn C. Nunziato, Freedom of Expression, Democratic
Norms, and Internet Governance, 52 Emory L.J. 187, 190–91 (2003).
221 Because this proposal applies only to fair use cases, it faces one major, but not insurmountable,
difficulty. Namely, it requires changing the panel assignment provision for
all disputes, even though the findings—and subsequent choice of law recommendation—
apply only to fair use disputes. As I argue below, changing the panel assignment policy
need not be viewed negatively. But, even if it is, there may be other ways around making
the wholesale change. Some screening mechanism prior to panel assignment, for example,
could be used to determine whether the dispute substantively involved fair use.
222 See supra notes 25–31 and accompanying text.
223 See supra Figure 12.
224 See supra Figure 12.
225 E.g., Native Instruments GmbH v. Williams, No. D2007-0973 (WIPO Aug. 21, 2007),
http://www.udrpsearch.com/wipo/d2007-0973 (complainant was from Germany, respondent
was from the United States, and the sole panelist was from the United Kingdom);
Pfizer Inc. v. Zhan, No. D2006-1134 (WIPO Oct. 30, 2006), http://www.wipo.int/amc/
en/domains/decisions/html/2006/d2006-1134.html (complainant was from the United
States, respondent was from Canada, and the sole panelist was from China).
226 See supra Figure 12.
2012] An Empirical Analysis of Fair Use Decisions 117
the United Kingdom, Canada, and New Zealand, the percent of cases
in which neither party was from those countries was 50%, 78%, and
86%, respectively.227 Thus, many non-U.S. panels decide cases involving
parties from countries other than their own. Unlike the assignment
pattern for U.S. arbitrators, no clear assignment pattern exists for nonU.S.
arbitrators.228
Normally, random assignment of arbitrators is a virtue. Not here.
The reason is because panel assignment and choice of law are closely
connected. What law panels apply depends on who comprises those panels.
Recall that without a principled basis for applying laws, UDRP panels
will continue to apply national laws or WIPO Rules—or make ad
hoc decisions—at varying rates. Given the data, we would expect that, in
fair use cases, U.S. panels would continue to apply U.S. law more than
other panels apply foreign law. But, for other panels, what principles or
law they will apply is uncertain. Part of this uncertainty is tied to panel
assignment. If a panel has no national connection to either party, its decision
is more likely to be ad hoc, unprincipled, and unpredictable.
To eliminate this risk, ICANN should clarify and explain how providers
should assign panels. Specifically, panel assignment should be
tied to what law applies in fair use cases. As I stated at the beginning of
this Section, panels should apply the law of a respondent’s home country
in fair use cases.
To facilitate these choice of law rules and procedures, panels must
have some knowledge of the law they apply. Thus, I propose that the
provider assign to cases arbitrators that share the same nationality as
the respondent. In the case of three-member panels, this means two
arbitrators should be from the same country as the respondent; the
third should be from the complainant’s country.229 Providing two panelists
from the respondent’s country ensures that the law is applied
accurately. Retaining at least one party from the complainant’s country
allows for a flexible system. Because the complainant’s fair use law may
227 See supra Figure 12.
228 It may be that there just have not been enough cases for a pattern to develop. Although
plausible, that explanation does not seem likely. The more likely explanation is
that panel assignment does not matter much, at least as concerns fair use, in cases where
neither party is from the United States.
229 There are different ways to implement this proposal. One method would allow each
party to choose a home country panelist when either party elects a three-member panel.
The provider would then choose the third panelist, who would hail from the respondent’s
home country. In cases where neither party elects to choose arbitrators, the provider could
assign panelists—two from the respondent’s home country and one from the complainant’s
home country.
118 Boston College Law Review [Vol. 53:65
govern in some cases, it is important for the panel to be knowledgeable
of the complainant’s home-country law.
What happens, though, when the complainant’s law governs in a
one-member panel? When the respondent default, there is no choice
of law problem. Under the UDRP Rules, the provider appoints a onemember
panel after it receives the respondent’s response.230 Thus, if the
respondent defaults, the provider can assign a panelist from the complainant’s
home country.
But a problem arises in cases where the respondent answers and
the complainant contests the respondent’s purported location. The
arbitrator in such a case would hail from the respondent’s home country.
So that arbitrator likely would not have working knowledge of the
complainant’s home-country law. Under the proposed burden-shifting
approach (discussed below), however, the complainant’s law could govern.
Thus, the panelist deciding the case could be ignorant of the
complainant’s law. Although this problem could be addressed in many
different ways, one solution comes to mind: eliminate one-member panels
altogether. In addition to facilitating the proposed choice of law
rule, using only three-member panels probably would promote fairer
resolutions all around.231
Even if we accept these panel assignment rules, we would still need
to know what constitutes “a respondent’s home country.” This term
should mean “country listed as the registrant’s in the WHOIS database.”
The WHOIS database is a publicly available, searchable database
that contains the contact information of domain-name registrants.232
When an individual or corporation registers a domain name, the information
they provide automatically enters the WHOIS database.233 If
I registered <www.stevejobslikesapples.com>, for example, I would
have to enter my name and contact information. Because I reside in
the United States, I would enter that as my home country. If Apple filed
a complaint with WIPO, a U.S. panel would hear the dispute. In mak-
230 UDRP Rules, supra note 5, § 6(b).
231 See Geist, supra note 23, at 930–33.
232 Mueller, supra note 45, at 235 (“The WHOIS database allows one to type in a domain
name and pull up the name and address of the individual or company that registered
the domain.”); What Is WHOIS?, Network Solutions, http://www.networksolutions.com/
whois/index.jsp (last visited Dec. 20, 2011) (“Once your listing appears in this online directory,
it is publicly available to anyone who chooses to check domain names using the
WHOIS search tool.”).
233 What Is WHOIS?, supra note 232 (“When you register a domain name, the Internet
Corporation for Assigned Names and Numbers (ICANN) requires your domain name
registrar to submit your personal contact information to the WHOIS database.”).
2012] An Empirical Analysis of Fair Use Decisions 119
ing a decision, the panel would apply U.S. law to interpret or “fill the
gaps” in the (fair use portion of the) UDRP.
Because a domain-name registrant enters information into the
WHOIS database, one might question the reliability of the information.
Hence, there is some risk that the registrant will provide incorrect, misleading,
or false information. Although current ICANN agreements
address this risk, their effectiveness is uncertain.234 The new choice of
law provision should address this risk in two ways. First, it should require
any complainant who asserts that the respondent does not reside
in the home country to provide evidence of that fact. Evidence must
consist of more than mere assertions. If the complainant can produce
such evidence, the burden shifts to the respondent, who will have to
provide evidence rebutting the presumption. If, on the other hand, the
234 ICANN’s Accreditation Agreement requires registrars to obtain—and domain-name
registrants to provide—accurate contact information. Registrar Accreditation Agreement ,
§ 3.7.7.1, Internet Corp. for Assigned Names & Nos. (May 21, 2009), www.icann.org/en
/regis-trars/ra-agreement-21may09-en.htm.
The Registered Name Holder shall provide to Registrar accurate and reliable
contact details and promptly correct and update them during the term of the
Registered Name registration, including: the full name, postal address, e-mail
address, voice telephone number, and fax number if available of the Registered
Name Holder; name of authorized person for contact purposes in the
case of an Registered Name Holder that is an organization, association, or
corporation; and the data elements listed in Subsections 3.3.1.2, 3.3.1.7 and
3.3.1.8.
Id. It also requires registrars to verify, and re-verify, this information for accuracy. If anyone
notifies the registrar of an inaccuracy, the registrar has a duty to investigate. Id. § 3.7.8.
Registrar shall, upon notification by any person of an inaccuracy in the contact
information associated with a Registered Name sponsored by Registrar,
take reasonable steps to investigate that claimed inaccuracy. In the event Registrar
learns of inaccurate contact information associated with a Registered
Name it sponsors, it shall take reasonable steps to correct that inaccuracy.
Id. The registrar may cancel the domain names of registrants who supply inaccurate contact
information to the Registrar. Id. § 3.7.7.2.
A Registered Name Holder’s willful provision of inaccurate or unreliable information,
its willful failure promptly to update information provided to Registrar,
or its failure to respond for over fifteen calendar days to inquiries by
Registrar concerning the accuracy of contact details associated with the Registered
Name Holder’s registration shall constitute a material breach of the
Registered Name Holder-registrar contract and be a basis for cancellation of
the Registered Name registration.
Id. Domain-name registrants can be held liable for uses of their domain names by third
parties, licensed or unlicensed. Id. § 3.7.7.3 (requiring a licensor of a domain name to
keep current contact records and to accept liability for any use by the licensee).
120 Boston College Law Review [Vol. 53:65
respondent cannot meet the burden, the complainant’s law applies.
Second, in the cases of respondent default, the panel will presume that
the respondent does not reside in the country listed in the WHOIS database,
and the complainant’s law will govern. Using either the law of
the complainant’s or the respondent’s home country allows providers
to appoint panels with knowledge of any potentially applicable law.
Despite this burden-shifting approach, some may worry that corporate
registrants may try to “game” the system by registering corporations
in countries with speech-friendly domain-name laws. While this
risk exists, at least three factors ameliorate it. First, the fair use defense
covers noncommercial fair uses. This excludes uses designed merely to
profit off the goodwill in a trademark. For these reasons, corporations
are not likely to (successfully) game the choice of law provision.
Second, companies seeking to protect their trademarks will be enforcing
their marks; they will be the complainants. Because the law of the respondent’s
home governs, where the complainant resides is important only
in cases of respondent default. Third, it costs time and money to set-up
and operate a corporation in a country. Costs will diminish the risk that
companies will try to register companies in domain-friendly countries
merely to trade off the goodwill of others’ trademarks. Under the new
UDRP provision, a company will not be able to circumvent the residence
requirement by merely registering a corporation; it must operate
in that country as well.235
At this point, one might be concerned that the burden on the
complainant is too high. That is, forcing the complainant to gather information
about the respondent’s whereabouts reduces the efficiency
of the UDRP. But, that ignores the complainant’s current duties under
the UDRP. A complaint under the current UDRP must provide much
of this information, including
the name of the Respondent (domain-name holder) and all
information (including any postal and e-mail addresses and
telephone and telefax numbers) known to Complainant regarding
how to contact Respondent or any representative of
Respondent, including contact information based on precomplaint
dealings, in sufficient detail to allow the Provider to
send the complaint . . . .236
235 What constitutes “operating” could be defined either by local law or by the UDRP.
236 UDRP Rules, supra note 5, § 3(b)(v).
2012] An Empirical Analysis of Fair Use Decisions 121
One may read the language “known to Complainant” to absolve the
complainant of any duty to investigate facts. But such a reading would
not make sense. Most of the time, complainants will know little, if anything,
about the domain-name owner. At the very least, the complainant
must take reasonable steps to gain such information, such as by using
the WHOIS database. Since the information in the WHOIS database
would govern the choice of law issue, this new requirement would not
be a burden.
4. Using Local Law Addresses Concerns of Speech, Uniformity, Cost,
and Predictability
There are further concerns with this system. Four are particularly
relevant: speech, uniformity, cost, and predictability. Although using
local law is promising, it raises free speech concerns. Some countries’
trademark and cybersquatting laws are not speech-protective. Some
commentators worry that oppressive regimes will use the trademark
law—and thus the UDRP—as cover to censor unpopular viewpoints or
speech. Although this risk exists, it is not as serious as it appears. Most
oppressive governments wishing to suppress speech do so using direct
government action, not trademark or cybersquatting laws. The Chinese
government, for example, censors speech by blocking websites directly.237
Thus, UDRP disputes are not typically the subject of speech disputes
in authoritarian countries about which we normally would be
concerned.
Nevertheless, a risk of private censorship exists in countries with
speech-restrictive laws. In these countries, trademark owners will be
able to suppress speech that infringes their trademarks, even if another
country would decide the dispute differently. This risk, though, is one
that traditional trademark law readily accepts. Each country’s trademark
laws generally do not apply outside its borders. It is reasonable to
expect individuals of various countries to bear that same risk when using
a trademark in a domain name. Thus, using the law of the respondent’s
home country to resolve UDRP disputes is in line with traditional
territoriality principles.
Using local laws, though, decreases the overall uniformity of the
UDRP. Indeed, we saw in Section IV.A that at least one panel fears that
237 See Jonathan Zittrain & Benjamin Edelman, Empirical Analysis of Internet Filtering in
China, Harvard Law Sch., http://cyber.law.harvard.edu/filtering/china/ (last updated
Mar. 20, 2003).
122 Boston College Law Review [Vol. 53:65
applying local laws would disturb the uniformity of the UDRP.238 That
fear, however, seems overstated. Currently, various splits already occur
within the WIPO Rules.239 Furthermore, many decisions (68%),240 do
not cite WIPO Rules at all. Although importing national law may fracture
the UDRP, maintaining the current U.S. law/UDRP dichotomy
trades one large fracture line for a more nuanced and controlled approach.
The U.S. law/UDRP dichotomy, for example, sacrifices the territoriality
that defines all other trademark law. Instead, it opts to favor
U.S. parties, leaving others under less-friendly interpretations of the fair
use defense, even in cases where their home country’s law (like Germany’s)
may allow such a use.241 Additionally, it ignores the fact that the
parties can file lawsuits in various countries even after a UDRP resolution.
Put another way, parties can pursue their legal options under the
laws of other countries.242 If the UDRP purports to be a system founded
on common international principles, it makes little sense to disregard
those legal principles that would govern in its absence. Moreover, there
always exists the possibility of an increasingly unpredictable UDRP. Although
the U.S. law/UDRP dichotomy exists now, further breaks could
develop on unforeseen grounds. Although this risk currently appears
low, it would, if realized, increase ex ante costs for both (potential)
complainants and respondents.
Finally, the objection also assumes that the costs of using local laws
would be much higher than a self-referential system. That assumption
does not hold. Under a policy that uses local law to settle disputes, the
respondent would have at least as much notice as under the UDRP. Instead
of combing through WIPO decisions, which may not be in the
native language of the party, the parties can consult the binding legal
precedents of the applicable jurisdiction.
For a similar reason, a UDRP that imported local law could reduce
social costs by specifying what law applies and who applies it. As one
238 See supra notes 136–198 and accompanying text.
239 See supra notes 146–157 and accompanying text.
240 The 68% includes only decisions after WIPO published its Rules in 2005.
241 See supra notes 159–198 and accompanying text.
242 Taking account of local laws also recognizes the realities of the political economy. If
the UDRP does not reflect the laws of specific countries, the usefulness of the UDRP diminishes,
at least in particular countries. In the United States, for example, the ACPA reflects
principles embodied in the UDRP; thus, the UDRP should be a strong predictor of outcomes
under the ACPA. But, if the UDRP does not reflect U.S.-ACPA law, the UDRP risks
becoming disconnected from the ACPA, at least when applied to U.S. parties. How this
would affect the use of U.S. courts and the state of U.S. law is uncertain. At the very least, it
would change the current incentive structure for filing UDRP complaints. Similar concerns
exist for countries that have specific cybersquatting laws.
2012] An Empirical Analysis of Fair Use Decisions 123
scholar has noted: “the differences between the trademark and domain-name
systems lead to these (and other) social costs ex post; if
these are costs that can be reduced by one system having regard to the
other ex ante, then some coordination is clearly worthwhile.”243 A system
defined by reference to specific laws is also better equipped to address
noneconomic trademark law concerns, such as free speech.244
This is important because “different countries deal with the collision
[between trademark law and free speech] in different ways . . . .”245
What if, however, a country has no specific domain-name law? Of
the thirty-four countries in my study, for example, only seven countries
in my sample have laws that discuss domain names: the United States,
Belgium, Finland, Japan, Russia, Spain, Italy, and, very recently, France
and Norway.246 Only a few of these laws deal explicitly with cybersquatting.
How could panels apply the laws of their home country when
none existed? That question erroneously assumes that having no cybersquatting
law is the same as having no law at all. Although many countries
do not have cybersquatting laws, they do have laws regarding
speech and trademarks. These laws can be applied in the context of
UDRP disputes involving fair use. Prior to the enactment of the ACPA,
for example, U.S. courts ably applied free speech and trademark principles
to domain-name disputes.247 Indeed, they still do—the ACPA is
an alternative to traditional trademark infringement (and dilution)
actions.248 There is no reason to suspect other countries’ courts or arbitrators
would be less capable. Some countries already have done this.249
Alternatively, panels could incorporate existing country-specific,
domain-name resolution policies. Many countries have UDRP-like policies
that pertain to domain names containing the country’s “code” —
like “.us” for the United States—known as a “country-code top-level
243 Dinwoodie, supra note 207, at 505–06.
244 Id. at 511 (stating that “trademark protection, especially for words, collides with
free speech concerns”).
245 Id.
246 See supra notes 184–191 and accompanying text.
247 Courts in the United States—prior to Congress enacting a domain-name-specific
law (the ACPA)—used traditional trademark principles to resolve domain-name disputes.
See, e.g., Brookfield Commc’ns, Inc. v. W. Coast Entm’t Corp., 174 F.3d 1036, 1046 (9th Cir.
1999) (applying traditional trademark law principles to a dispute involving domain
names); Panavision Int’l, L.P. v. Toeppen, 141 F.3d 1316, 1324–27 (9th Cir. 1998) (applying
the Federal Trademark Dilution Act to a cybersquatting dispute).
248 E.g., Newport News Holdings Corp. v. Virtual City Vision, Inc., 650 F.3d 423, 431
(4th Cir. 2011) (dealing with a plaintiff who alleged a violation of the ACPA in addition to
trademark law).
249 See supra note 196.
124 Boston College Law Review [Vol. 53:65
domain” (ccTLD). Perhaps unsurprisingly, WIPO helped shape each of
these agreements.250 These agreements, like the UDRP, are contractual.
When one wants to register a domain name with ccTLD, one must do
so through a registrar in that country.251 Many registries have UDRPlike
agreements that the registrant must agree to when registering a
domain name.252
To obtain a ccTLD for Canada (.ca), for example, one must register
with the Canadian Internet Registration Authority (CIRA).253 Like
ICANN, CIRA binds registrants through a mandatory “Registrant
Agreement.”254 This agreement requires registrants to agree to resolve
disputes255 using the CIRA Domain Name Dispute Resolution Policy
(CDRP), which is similar to the UDRP.256
Although different countries’ agreements have different provisions,
they closely resemble the UDRP. The CDRP, for example, provides
a list of what constitutes “legitimate interests.”257 One of these includes
using the domain name “in good faith in association with a noncommercial
activity including, without limitation, criticism, review or
250 Domain Name Dispute Resolution Service for Country Code Top Level Domains (ccTLDs),
World Intellectual Prop. Org., http://www.wipo.int/amc/en/domains/cctld/ (last
visited Dec. 20, 2011) (“WIPO has provided advice to many ccTLDs with a view to establishing
registration conditions and dispute resolution procedures that conform with international
standards of intellectual property protection while taking into account the particular
circumstances and needs of the individual ccTLD.”).
251 Smith, supra note 53, at 153.
A wide variety of organizations, some state, some private, have responsibility
delegated by IANA [(The Internet Assigned Numbers Authority)] for managing
country code domains . . . . Each ccTLD is able to organize and sub-divide
its own domain space as it wishes within the general policy laid down by IANA
. . . . Each ccTLD tends to have its own local requirements for registration
. . . .
Id. (footnotes omitted).
252 See Arbitration and Mediation Center, ccTLD Database, World Intellectual Prop.
Org., http://www.wipo.int/amc/en/domains/cctld_db/codes/ar.html (last updated Oct.
24, 2011) (displaying a searchable database of “ccTLD registration agreements, WHOIS
services, and alternative dispute resolution procedures”).
253 FAQ, Canadian Internet Registration Auth., http://www.cira.ca/why-ca/faq/
(click through “What is a registry”) (last visited Dec. 20, 2011) (“CIRA is the registry for the
.CA domain.”).
254 Registrant Agreement, Version 2.0, Oct. 12, 2010, GoDaddy.com, https://www.godaddy.
com/domains/popups/caagreement.aspx (last visited Dec. 20, 2011).
255 Id.
256See CIRA Policies, Rules, and Procedure: CIRA Domain Name Dispute Resolution Policy, Version
1.2, Canadian Internet Registration Auth., http://www.cira.ca/assets/Documents/CDRP
policy.pdf (last visited Dec. 20, 2011).
257 Id. § 3.6.
2012] An Empirical Analysis of Fair Use Decisions 125
news reporting . . . .”258 The fair use provision of the “.au” Dispute Resolution
Procedure (auDRP), which resolves disputes involving Australia’s
ccTLD, is identical to the UDRP.259
Although using ccTLD dispute resolution procedures has promise,
there are two problems. First, most of these policies do not differ from
the UDRP. Thus, they do not provide much substantive guidance. This
likely reflects WIPO’s role in helping registrars craft dispute resolution
policies.260 Second, even if the UDRP did use specific ccTLD policies to
resolve domain-name disputes, it probably would not reflect the legal
differences in each country. The reason is because WIPO already decides
many ccTLD disputes.261 The data, however, show that arbitrators
are not importing their national law. For the UDRP to truly account for
the nuanced distinctions between countries, it must do more than look
to ccTLD dispute resolution policies. Thus, using ccTLD dispute resolution
procedures may not resolve the problem we are trying to address.
Specifically, it will not provide guidance to panels about what legal
principles they should use to resolve a domain-name dispute.
For this reason, it is best to decide fair use domain-name disputes
by reference to the trademark and unfair competition law(s) of each
country. This kind of decision making also has benefits. Like a federalstate
system, it frees entities to craft new laws according to various customs.
These customs approximate the values that permeate particular
societies. Over time, individuals and societies compare the emerging
legal rules with values. Based on this comparison, some rules are discarded,
revised, or retained. New rules also emerge. The proposed approach
facilitates this process. It promotes more informed and
nuanced legal decision making. And, it does so by drawing on the resources
and minds of the entire world.
258 Id. § 3.6(d).
259 .au Dispute Resolution Policy (auDRP) (2010-05) § 4(c)(iii), .au Domain Admin. Ltd.
(Aug. 13, 2010), http://www.auda.org.au/policies/auda-2010-05/ (stating as a right or
legitimate interest that “you are making a legitimate non-commercial or fair use of the
domain name, without intent for commercial gain to misleadingly divert consumers or to
tarnish the name, trademark or service mark at issue”).
260 Domain Name Dispute Resolution Service for Country Code Top Level Domains (ccTLDs),
World Intellectual Prop. Org., http://www.wipo.int/amc/en/domains/cctld/ (last visited
Mar. 6, 2011) (“WIPO has provided advice to many ccTLDs with a view to establishing
registration conditions and dispute resolution procedures that conform with international
standards of intellectual property protection while taking into account the particular circumstances
and needs of the individual ccTLD.”).
261 For a list of auDRP disputes and those resolved by WIPO, for example, see auDRP Proceedings
Archive, .au Domain Admin. Ltd., http://www.auda.org.au/audrp/proceedingsarchive/
(last updated Nov. 16, 2011).
126 Boston College Law Review [Vol. 53:65
Conclusion
Revising and clarifying the choice of law provision will improve the
functioning and enhance the legitimacy of the UDRP. What law applies
in a UDRP dispute—fair use or otherwise—can determine who wins. As
more people crawl the Web, an increasing number of individuals use
domain names for a variety of purposes, including as comment, criticism,
personal, informational, and fan websites. The rise in Internet use
is not likely to reduce the number of domain-name disputes.262 As the
number of disputes increases, the UDRP will become more important.
Without intervention, the effectiveness and fairness of the UDRP will
likely suffer. To avoid this failure, ICANN should revise the current
choice of law provision.
The results of my study show that U.S. respondents have an advantage
in fair use domain-name disputes; this Article calls this phenomenon
U.S. Favoritism. Specifically, U.S. respondents win in fair use cases
35% of the time, compared to 17% of the time for other respondents.
Additionally, U.S. panels rendered decisions favorable to respondents
in 36% of cases, compared to 20% for other panels. Although my study
was limited to WIPO fair use cases, further study might reveal similar
U.S. Favoritism in UDRP decisions generally.
U.S. Favoritism occurred for two reasons. First, the UDRP’s choice
of law provision allowed panels to apply any country’s law in their complete
discretion. Second, U.S. panels took advantage of this provision at
higher rates than their counterparts: they imported U.S. law into UDRP
fair use cases more frequently than other panels—or, at the very least,
other panels perceived U.S. law as generally more speech-friendly than
non-U.S. law. The “imported” law provided more protection to respondents
than the UDRP as applied without reference to any country’s law.
Once imported, the law replicated and now populates WIPO decisions
as a rule.
Despite the hazards of its current choice of law provision, ICANN
does not seem concerned with the issue. Recently, for example, ICANN
debated how to add new domain-name space to the Internet.263 Part of
262 With the exception of 2009, the number of UDRP decisions decided by WIPO generally
has risen each year since 2003. Total Number of Cases per Year, World Intellectual
Prop. Org., http://www.wipo.int/amc/en/domains/statistics/cases.jsp (last visited Dec.
20, 2011) (listing the numbers from 1999–2011 in ascending order: 1, 1857, 1557, 1207,
1100, 1176, 1456, 1824, 2156, 2329, 2107, 2696, and 2390).
263 Specifically, ICANN is adding new generic gTLDs, which the reader may recognize
as the familiar “.com,” “.edu,” etc. Among the new gTLDs will be Internationalized Domain
Names (“IDNs”), which are “represented by local language characters, or letter equi-
2012] An Empirical Analysis of Fair Use Decisions 127
this process included formulating a dispute resolution procedure for
domain names registered in this new space.264 Like the UDRP, disputes
over these new online territories may favor certain parties if the choice
of law issue is not settled. Unfortunately, ICANN’s new gTLD dispute
procedure ignores this issue. Instead, it allows legal objections to newly
registered domain names based on “generally accepted and internationally
recognized principles of law.”265 Although different from that
in the current UDRP—which allows panels to apply “any . . . principles
of law . . . [they] deem[] applicable”266—it is not likely to provide
choice-of-law guidance for panels. The final draft of the new gTLD
procedure allows “legal rights objections” for when the “applied-for
gTLD string infringes the existing legal rights of the objector.”267 The
new procedure does not take a stance on the choice of law issue and
leaves it up to the panels; they decide what constitutes an “infringement”
of legal rights. Panels also decide, when a limited public interest
objection is filed, what constitute “general principles of international
law for morality and public order.”268 Still, the new procedure does not
leave panels free to do whatever they choose; it lays out factors for pavalents.”
Frequently Asked Questions, supra note 87. One example is “.テスト” (this kanji
means “test” in Japanese), which would represent a Japanese IDN. Root Zone Database, Internet
Assigned Nos. Auth., http://www.iana.org/domains/root/db/ (last visited Dec.
20, 2011).
264 ICANN has posted a new version of this dispute resolution procedure. See Internet
Corp. for Assigned Names & Nos., gTLD Applicant Guidebook: Module 3 § 3.1.2.2
(2011), available at http://www.icann.org/en/topics/new-gtlds/objection-procedures-clean-
19sep11-en.pdf (explaining that rights-holders can object by claiming that the new gTLD
infringes a registered or unregistered trademark); id. § 3.4.2 (listing a variety of factors for a
panel to consider). See generally Internet Corp. for Assigned Names & Nos., Attachment
to Module 3: New gTLD Dispute Resolution Procedure (2011) [hereinafter Attachment
to Module 3], available ar http://www.icann.org/en/topics/new-gtlds/new-gtld-drp-
clean-19sep11-en.pdf.
265 Attachment to Module 3, supra note 264, art. 2(e)(ii) (allowing for an “Existing
Legal Rights Objection” when “the string comprising the potential new gTLD infringes the
existing legal rights of others that are recognized or enforceable under generally accepted
and internationally recognized principles of law”); id. art. 2(e)(iii) (allowing for a “Limited
Public Interest Objection” when “the string comprising the potential new gTLD is contrary
to generally accepted legal norms relating to morality and public order that are recognized
under international principles of law”). The proposed procedure allows specific
potential objections that are not based on any kind of legally recognized right. See id. art.
2(e)(i) (specifying objections based on confusing similarity to an existing gTLD (not
trademark)); id. art. 2(e)(iv) (specifying objection when a significant portion of a community
“to which the string may be explicitly or implicitly targeted” so requests).
266 UDRP Rules, supra note 5, § 15(a).
267 gTLD Applicant Guidebook, supra note 87, § 3.2.1.
268 Id. § 3.5.3.
128 Boston College Law Review [Vol. 53:65
nels to consider when determining what these standards mean.269 In
other words, ICANN prefers dispute resolution procedures that can,
but need not, account for local law.
The findings of this study, however, suggest equivocating or avoiding
the choice of law issue will lead to favoritism of some nationality, or
perhaps for existing domain-name holders. This has the potential to
undermine the legitimacy and efficacy of the UDRP as an international
dispute resolution mechanism.
ICANN, of course, could prevent that from happening—and the
time is ripe for it to do so. Recently, ICANN reviewed the UDRP. In February
2011, the Generic Names Supporting Organization (GNSO)
Council requested an Issue Report on “[h]ow the UDRP has addressed
the problem of cybersquatting to date, and any insufficiencies/inequalities
associated with the process.”270 Following a request for
public comments, two community members identified safe harbors for
free speech as an area that needs attention.271 The issue is all the more
pressing given that the GNSO council previously resolved to make no
action on at least one pressing free speech issue (Gripe Sites).272 ICANN
issued a Final Report on the Current State of the Uniform Dispute Resolution
(“Final Report”),273 which was submitted to the GNSO Council
in October 2011.274 The GNSO Council will then consider whether “to
commence a policy development process (PDP) on the UDRP” —that is,
the GNSO Council will decide whether to revise the UDRP.275 The Final
Report recommends delaying a PDP until the new gTLD dispute-
269 Id. § 3.5.
270 Margie Milam, Internet Corp. for Assigned Names & Nos., Final GNSO Issue
Report on the Current State of the Uniform Domain Name Dispute Resolution Policy
1, 33 (2011), available at gnso.icann.org/issues/udrp/udrp-final-issue-report-03oct11-
en.pdf.
271 Id. at 37, Annex 2.
272 Id. at 35, Annex 1.
Rough Consensus: Make no recommendation. The majority of RAPWG
members expressed that gripe site and offensive domain names that use
trademarks should be addressed in the context of cybersquatting and the
UDRP for purposes of establishing consistent registration abuse policies in
this area, and that creating special procedures for special classes of domains,
such as offensive domain names, may present problems.
Id.
273 Id.
274 The GNSO Council is an ICANN organization that identifies and recommends
changes to gTLD policies. GNSO, ICANNWiki, http://icannwiki.com/index.php/GNSO
(last modified Sept. 10, 2011).
275 See Milam, supra note 270, at 3.
2012] An Empirical Analysis of Fair Use Decisions 129
resolution system—the Uniform-Rapid Suspension System (URS)—has
operated for “at least eighteen months.”276 Because ICANN modeled the
URS on the UDRP, it suggests that the data from the URS would inform
any decision on the UDRP.277 The GNSO Council has agreed, and it
will revisit the issue after receiving another ICANN staff report within
the next eighteen months.278
Although ICANN has decided not to reform the UDRP immediately,
it has left open the possibility that it will do so in the near future.
When it next examines the issue, ICANN should consider the results
of this study and the variety of studies and suggestions offered by
others concerned with the UDRP. The results of this study show that
panel assignment in fair use cases is different for U.S. and non-U.S.
parties. And this panel assignment—along with a vague choice of law
provision—helps explain what I call U.S. Favoritism: U.S. panels provide,
and U.S. respondents receive, more favorable treatment than
their counterparts in fair use disputes. U.S. Favoritism raises concerns
of speech, fairness, predictability, and legitimacy. One way to address
these concerns is to specify what law, if any, applies in fair use disputes.
Doing so would likely require altering the panel assignment procedures.
And that, in turn, should force two other beneficial changes.
First, ICANN should make an otherwise secret panel-assignment procedure
public. Second, ICANN should ensure panelists have knowledge
of the law that applies in fair use disputes. If ICANN makes these
changes in the context of fair use—and considered them for the UDRP
at large—it would improve the UDRP.
276 Id. at 4 (recommending “that a PDP on the UDRP not be initiated at this time”).
277 Id. (stating that this “would allow the policy process to be informed by data regarding
the effectiveness of the URS, which was modeled on UDRP, to address the problem of
cybersquatting”).
278 E-mail from Glen de Saint Géry, GNSO Secretary, to GNSO Council (Dec. 16, 2011,
02:58), available at http://gnso.icann.org/mailing-lists/archives/council/msg12522.html.
RESOLVED further, the GNSO Council requests a new Issue Report on the
current state of all rights protection mechanisms implemented for both existing
and new gTLDs, including but not limited to, the UDRP and URS, should
be delivered to the GNSO Council by no later than eighteen (18) months following
the delegation of the first new gTLD.
Id.
INSERTED BLANK PAGE