[wg-c] Position paper

[Milton Mueller <mueller@syr.edu>]

Comments and people willing to sign on as supporters welcome

NEW GTLDS: COMPETITION, INNOVATION, AND CULTURAL DIVERSITY.
AN ALTERNATIVE REPORT FROM MEMBERS OF WORKING GROUP C

1. Adding new gTLDs to the root is an important part of ICANN’s
mandate
ICANN was created because the institutions that preceded it were
unable to resolve the intense political and economic conflicts
created by demand for new top-level domain names. The US
Department of Commerce White Paper included “oversee policy for
determining the circumstances under which new TLDs are added to
the root system” as one of the 4 key purposes of the new
corporation. Thus, ICANN’s handling of new TLDs is an important
test of its DNS stewardship. We urge a flexible, diverse, and
open approach to this problem.

2. Demand for new gTLDs
There is abundant evidence of significant end user demand for
expanding the name space. While some users will continued to be
satisfied registering under ccTLDs, a very large number prefer
gTLDs and want alternatives to .com, .net, and .org. Evidence of
the intense demand for new, semantically interesting gTLDs is
enumerated below:

a) Second-level domain names under the dot com TLD routinely
change hands for enormously inflated prices. (See Table) These
are not cases of “cybersquatting” but legitimate trades of
ordinary, untrademarked words. High prices reflect the artificial
scarcity of common names in existing gTLDs, and the premium on
.com names in particular.

Domain Name Price
Bingo.com US$ 1.1 million
Wallstreet.com US$ 1.03 million
Rock.com US$ 1.0 million
Eflowers.com US$ 1.0 million
Drugs.com US$ 800,000
University.com US$ 530,000
Computer.com US$ 500,000
Blackjack.com US$ 460,000
BBC.com L 200,000
Business.com US$ 150,000
Internet.com US$ 100,000
Trade.com US$ 40,000

b) There are widespread complaints among users that it is
becoming increasingly difficult to find simple domain names in
the NSI gTLDs. The basis for these complaints was verified in an
April 14, 1999 Wired News survey, which found that of 25,500
standard dictionary words, only 1,760 were free in the com
domain. At the time of that article, only about 7.5 million
domain names had been registered. More than 3 million have been
registered in the ensuing five months.

c) Currently, the weekly growth rate of domain name registrations
is over 270,500. Projecting that growth rate into the future
would put the number of domain name registrations at 67 million
by 2003. Current gTLDs simply will not be able to contain such
growth.

d) The growing demand for domain names cannot be satisfied by
ccTLD registries. The problem is not capacity, but consumer
choice. Given a choice, most users prefer gTLDs. 74% of the
world’s domain names are registered in gTLDs; 61% are in dot com.
The gTLD’s share of total domain name registrations has remained
constant since 1997. This is true despite the fact that the
proportion of Internet users outside North America has grown
significantly in the same period. Internet users’ preference for
gTLDs over ccTLDs has remained evident despite the limited choice
of gTLD names available now. Expanding the number of gTLDs and
making their semantic content relate to different cultures and
languages will make this preference even stronger. We believe
that TLD policy ought to reflect end user preferences, not
top-down, preconceived notions of where users “ought” to
register.

e) On the supply side, there are numerous potential suppliers of
registry services willing and able to administer new gTLDs. These
include, but are not limited to, Core, Image Online Design,
Name.space, VRx, and MHSC. Thousands of registrants have paid to
reserve domain names under TLDs not carried in the legacy IANA
root. Several ccTLD registries, such as .NU, .CC, and .TO, have
transformed themselves into gTLDs, marketing their names globally
as alternatives to .com, .net and .org. It is obvious that these
businesses perceive a serious demand for gTLD services.

3. The WG-C Recommendations
WG-C’s report to the DNSO proposes to add only 6-10 new gTLDs to
the root as a “testbed.” It would then engage in an evaluation
period before adding any more. There is no commitment or
presumption that more would be added, and the duration and
criteria of the evaluation are not specified. An entirely new
process would have to be initiated to add more. Although many of
the signatories of this report supported the 6-10 proposal as the
lowest-common denominator for widespread agreement in the working
group, we believe that the proposal is far too limited and
imposes undue restrictions on competition, innovation, and
cultural diversity.

4. An Alternative Proposal
The following policy better reflects the mandate of ICANN and the
needs of the Internet:
a) ICANN announces its intention to accept applications for a
maximum of 500 new gTLDs over the course of the next three years.

b) The applicants for these new gTLDs would be added at a gradual
pace – e.g., 10 the first six months, 40 the next six months, 150
the second year and 300 the third year.

c) A defined proportion of the new gTLDs should be reserved for
names that reflect distinct cultural/linguistic groups. ICANN’s 5
geographic regions could be used as the basis for these
reservations.

d) Instead of an open-ended “evaluation” period, we propose that
ICANN define clear, objective, quantitative indicators of
problems that would justify an interruption or cessation of the
process of adding new gTLD registries and names to the root. In
other words, new gTLDs should be considered innocent until proven
guilty.

A proposal similar to this one received the support of about 35%
of the voting working group participants in a straw poll
conducted by the WG-C chairs. Below, we enumerate the reasons for
the more open, diverse, and competitive approach to the new TLD
problem.

4.1 Competition Policy
The most important reason to specify at the outset a specific and
large number of gTLDs is to ensure a competitive market and a
level playing field in the provision of domain name registration
services under new gTLDs. Only 6-10 new gTLDs, with no commitment
to add more later, risks creating an oligopoly of registries and
a monopoly under each name. Competition would be attenuated, and
the incumbents might have a vested interest in slowing or
stopping the addition of new competitors. The initial grantee(s)
will have a significant and unfair competitive advantage. They
will be able to capitalize on the pent-up demand for new names in
the short term.

4.2 Market Certainty
A highly restricted initial rollout, with no commitment to move
forward, makes the initial grant of the gTLD to new registries a
far less rational process for consumers and producers alike. The
public will not know whether new alternatives will be available
in the near future, and hence their choice of a supplier may be
based on the assumption that the new gTLDs are the only option. A
small number of new gTLDs encourages pre-emptive and speculative
registrations that are based on the possibility of continued
artificial scarcity. Prospective competitors will have no idea
when they will be allowed into the market or whether they will be
allowed at all. Limiting the number makes the initial assignment
decision more controversial and arbitrary as well. ICANN will not
be able to ensure that its initial award is not a permanent and
final grant of a very special privilege, similar to the NSI
monopoly of the past.

A pre-announcement of a larger number, on the other hand, makes
it clear that the initial winners of new gTLD awards can
anticipate plenty of additional competition. Investment and entry
decisions of new competitors will be far more rational.
Consumption decisions will be based on the need for and value of
the domain names themselves, not on attempts to exploit
artificial scarcity.

5. Trademark Concerns and new GTLDs.
Some concerns were expressed in WG-C that the number of new gTLDs
should be restricted and the pace of introduction slowed in order
to guard against trademark infringement. We believe that that
argument is deeply flawed.

It is legitimate for the trademark and intellectual property
interests to advocate case-specific dispute resolution procedures
to protect themselves from abusive domain name registrations. It
is not legitimate, however, for trademark holders to demand what
amounts to a blanket prohibition on entry into a market for a
legal service (domain name registration) solely to make their
policing and enforcement task cheaper and easier. Such a policy
unfairly imposes costs and restrictions on millions of innocent
consumers and suppliers. Operating a TLD is not, per se, abusive
or infringing; infringing and speculative name registrations
constitute a tiny fraction of the total number of registrations
in any TLD. Similarly, we know that the existence of VCRs,
photocopying machines and similar recording devices will result
in copyright violations. No one in this day and age proposes to
ban them or severely restrict the number that can be manufactured
and sold for that reason. Technology should be allowed to
develop, and legal protections should be adjusted, if necessary,
to reflect new realities. We should not, therefore, restrict the
number of gTLDs based on concerns about their impact on trademark
protection.

Even if one does not agree with the reasoning above, the adoption
of a UDRP by ICANN completely severs any linkage between the
number of gTLDs and concerns about trademark protection. Under
the UDRP, contact information will be accessible and accurate,
and challenges to registrations will be easy and inexpensive.
Cybersquatters have lost every court case in which they have been
challenged at any rate, and all objective indications are that
the problem is declining. The UDRP should serve as an additional
deterrent.

Finally, we would point out that narrow restrictions on the
number of gTLDs actually contribute to many “cybersquatting”
problems. Name speculation is fueled by the premium value
attached to domain names in gTLDs. That premium value is largely
a product of artificial scarcity in the TLD space. Also, if there
are only four or five new gTLDs, the most logical course of
action for major trademark holders will simply be to
pre-emptively register names across all gTLDs. That defeats the
purpose of adding gTLDs.

6. Diversity and Competition in Registry Models
Perhaps the most controversial issue in Working Group C concerned
the proper economic model for registries and the process for
selecting names. All of the following questions were debated at
length:
a) Should registries be shared, exclusive, or should both be
allowed?
b) Should registries be non-profit, for-profit, or should both be
allowed?
c) Should ICANN define the names to be added to the root, or
should applicants come to ICANN with proposals for names? Should
the names available represent a fixed, standardized taxonomy?

All of these questions can be boiled down into one fundamental
issue. Does ICANN control the market structure for domain name
registrations and then license specific firms to fit into its
pre-ordained structure? Or, do end users and suppliers,
interacting in a marketplace, determine the market structure of
registries, registrars, and names, and ICANN in turn coordinates
their activities?

We believe that the latter alternative is the best one. ICANN
should not impose any specific model upon registries, nor should
it centrally impose any specific pattern of names. It should
allow the choices of end users in the marketplace to decide which
models and names succeed and which fail. The content of the
top-level name space should be driven by applications submitted
by prospective registries. Registries should contract with
registrars on a free market basis, with no pre-ordained pattern.
Competition in the marketplace and user preferences will
determine which approaches succeed. Regulatory and legal remedies
to consumer protection problems that develop should be left to
professional regulators in national governments. ICANN should
concentrate exclusively on technical and administrative
coordination of registry operators to ensure stability,
interoperability, and accountability. It should establish basic
qualifications for top-level domain name registries, and these
should be confined exclusively to technical stability and
financial responsibility.

The following are the reasons for this approach:

6.1 Product differentiation and innovation may require
integration of the registrar-registry function.
A registry that wants to create a distinct identity and unique
features for a TLD may need to control who registers within it.
It may also want to control the front-end software interface for
registration or other technical and business parameters. For
example, a TLD devoted to North American aboriginals, as was
proposed to WG-C, may want to ensure that specific tribal names
are only assigned to legitimate members of that tribe. Or a
privacy-enhanced gTLD, which was also proposed in comments to
WG-C, may want to dictate certain technical parameters and
protect the integrity of its data. Either requirement might best
be implemented by integrating the registry-registrar function.
This should be an option available to applicants.

6.2 Imposing uniform models on new entrants will probably make it
more difficult for them to compete with NSI.
Dot com already has enormous market dominance and a huge economic
premium is attached to names in that TLD. Com and the other NSI
gTLDs are already shared, and their wholesale price is regulated.
If new entrants into the marketplace are forced to adhere to the
exact same business model, their ability to generate the profits,
mindshare, and investment required to challenge NSI may be
hampered.

6.3 Compulsory sharing requires detailed technical and economic
regulation.
As the US Department of Commerce has learned, imposing “equal
access” upon a registry requires: a) fixing the wholesale price;
b) ensuring that the same SRS software is used; c) trying to
determine the economic costs of the registry; and other complex
monitoring and regulatory/contractual issues. ICANN lacks the
resources and the expertise to engage in such activity on a
global basis. Furthermore, ICANN is not the only line of defense
against abuse of market power. Antitrust actions, price
regulation, consumer fraud proceedings, and other remedies are
available in national and international jurisdictions.

We believe that discussion of the issue of models has been
distorted by the case of Network Solutions and its near-term
market dominance. NSI’s monopoly on gTLDs may have justified
requiring sharing in .com, .net, and .org. But NSI’s short-term
dominance should not dictate how the domain name market works in
the long term

6.4 The benefits of sharing can be realized without making it
compulsory.
Shared-registry TLDs are one option in the marketplace. The NSI
gTLDs are already shared. Several ccTLDs are also operating on
that model. If consumers express a clear preference for the price
and service delivered by this model, then it will be emulated.
If, however, consumers willingly choose services offered by
businesses following a different model, why should ICANN
interfere?