My name is Vinton G. Cerf, and outside of my regular employment at WorldCom(1),
I am the volunteer Chairman of the Internet Corporation For Assigned Names
and Numbers (ICANN). I appreciate the opportunity to appear before this
Committee to describe the efforts of ICANN to introduce additional competition
into the Internet name space, while at the same time prudently protecting
against possible disruption of this extremely important global resource
for communications and commerce.
The basic message I would like to leave with
you today is that ICANN is functioning well, especially for such a young
organization with such a difficult job. In fact, it has made substantial
progress toward the specific goals it was created to meet, including the
introduction of competition at both the wholesale and retail levels of
the registration of names in the Domain Name System (DNS). The recent
action to introduce seven new Top Level Domains (TLDs) into the DNS will
double the number of global TLDs and at the same time will not, we believe,
create serious risks of destabilizing the Internet -- something I know
none of us wants to see. The fact that ICANN, in just over a year, has
been able to generate global consensus on this issue -- which has been
fiercely debated for most of the last decade -- is a testament to ICANN's
potential to effectively administer the limited but important aspects
of the DNS that are its only responsibility.(2)
A. What is ICANN?
It is probably useful to first provide a
little background about ICANN, which is a unique entity that may not be
familiar to everyone. ICANN is a non-profit private-sector organization
with a 19-member international volunteer Board of Directors drawn from
a set of specialized technical and policy advisory groups, and through
open, worldwide online elections. ICANN was formed in 1998 through a consensus-development
process in the global Internet community, in response to a suggestion
by the United States Government that the private sector create such a
body. It was formed to undertake certain administrative and technical
management aspects of the Domain Name System (DNS) and the Internet address
space. Domain names serve as the visible face of the name and address
mechanism of the Internet -- in short, the way computers know where to
send or receive information.
ICANN performs functions that, prior to ICANN's
creation by the private sector, were performed by contractors to the US
Government (National Science Foundation and DARPA). ICANN is a young,
and still maturing organization; it turns out that achieving global consensus
is not so easy. But it has made great -- and many would say surprising
-- progress toward the objective shared by the vast majority of responsible
voices in the international Internet community: the creation of a stable,
efficient and effective administrative management body for specific technical
and related policy aspects of the DNS and the Internet address space that
is consensus-based, internationally representative, and non-governmental.
B. What are the Guiding Principles
There is nothing quite like ICANN anywhere
in the world, and of course it will be some time before we are certain
that this unique approach to consensus development can effectively carry
out the limited but quite important tasks assigned to it. I am cautiously
optimistic, but we are still at an early stage of evolution, and there
is much work to do. The organizational work has been complicated by the
fact that we have also been asked to simultaneously begin to accomplish
the specific operational goals set out by the US Government in the White
Paper.(3) The situation is analogous to
building a restaurant and starting to serve customers while the kitchen
is still under construction; it is possible, but may occasionally produce
The White Paper set forth four principles
that it described as critical to the success of an entity such as ICANN:
stability; competition; private, bottom-up coordination; and representation.
1. Stability is perhaps
the easiest to understand. The US Government was seeking to extract itself
from what it had concluded was no longer a proper role for the US Government
-- the funding of private contractors to manage important technical aspects
of the global Internet name and number address system -- but only in a
way that did not threaten the stability of the Internet. As the White
Paper said, and as seems obvious, "the stability of the Internet
should be the first priority of any DNS management system." If the
DNS does not work, then for all practical purposes for most people, the
Internet does not work. That is an unacceptable outcome, and thus everything
that ICANN does is guided by, and tested against, this primary directive.
2. Competition was also
an important goal set forth in the White Paper, which stated that "[w]here
possible, market mechanisms that support competition and consumer choice
should drive the management of the Internet because they will lower costs,
promote innovation, encourage diversity, and enhance user choice and satisfaction."
Competition in the DNS structure as it stands today is theoretically possible
at both the registry (or wholesale) level, and the registrar (or retail)
level. Increasing competition at the retail level involves only adding
additional sellers of names to be recorded in existing registries; as
a result, it generates relatively minor stability concerns. For this reason,
adding new competition at the retail level was the first substantive goal
that ICANN quickly accomplished after its formation. On the other hand,
adding new registry (or wholesale) competition -- which is the subject
of this hearing -- requires the introduction of additional Top Level Domains
into the namespace, and thus does raise potential stability issues of
various kinds. As a result, and given its prime directive to protect stability,
ICANN has moved forward in this area in a prudent and cautious way, consistent
with recommendations from many constituencies interested in the Internet,
which I will describe in more detail later in this testimony.
3. A third principle was
private sector, bottom-up consensus development, and
the entirety of ICANN's processes are controlled by this principle. ICANN
is a private-sector body, and its participants draw from the full range
of private- sector organizations, from business entities to non-profit
organizations to foundations to private individuals. Its policies are
the result of the complex, sometimes cumbersome interaction of all these
actors, in an open, transparent and sometimes slow progression from individuals
and particular entities through the ICANN working groups and Supporting
Organizations to ICANN's Board, which by its own bylaws has the role of
recognizing consensus already developed below, not imposing it from above.
Like democracy, it is far from a perfect system, but it is an attempt,
and the best way we have yet been able to devise, to generate global consensus
without the coercive power of governments.
4. Finally, the fourth core
principle on which ICANN rests is representation. A body
such as ICANN can only plausibly claim to operate as a consensus development
organization for the Internet community if it is truly representative
of that community. The White Paper called for ICANN to "reflect the
functional and geographic diversity of the Internet and its users,"
and to "ensure international participation in decision making."
To satisfy these objectives, all of ICANN's structures are required to
be geographically diverse, and the structures have been designed to, in
the aggregate, to provide opportunities for input from all manner of Internet
stakeholders. This is an extremely complicated task, and we are not yet
finished with the construction phase; indeed, we have just initiated a
Study Committee chaired by the former Prime Minister of Sweden, Carl Bildt,
to oversee a new effort to find a consensus solution for obtaining input
from and providing accountability to the general user community, which
might not otherwise be involved in or even knowledgeable about ICANN and
its activities. Other organizational tasks necessary to ensure that ICANN
is fully representative of the entirety of the Internet community are
also ongoing. This is hard work, and there is more to do to get it done
C. What Has ICANN Accomplished So
Obviously, ICANN is still a work in progress.
Nevertheless, it has, in my view, already made remarkable progress in
its young life. ICANN was created in November of 1998, and did not really
become fully operational until a year later (November of 1999) with the
signing of a series of agreements with Network Solutions Inc., then the
sole operator of the largest and most significant registries -- .com,
.net, and .org. So ICANN really has only about 14 months of operating
history. Still, even in that short span of time, some significant things
1. The Introduction of Retail Competition.
As one of its very first actions, ICANN created an accreditation
system for competitive registrars and, pursuant to its NSI agreements,
gave those new competitors access to the NSI-operated registries. When
ICANN was formed, there was only a single registrar (NSI) and everyone
had to pay the single price for the single domain name product that sole
registrar offered: $70 for a two-year registration. There are now over
180 accredited registrars, with more than half of those actively operating,
and you can now register a domain name in the .com, .net, and .org registries
for a wide range of prices and terms - some will charge zero for the name
if you buy other services, while others will sell you a ten-year registration
for significantly less than the $350 it would have cost pre-ICANN (even
if it had been available, which it was not). While there are no precise
statistics, in part because the market is so diverse, a good estimate
of the average retail price today of a one-year domain name registration
in the NSI registries is probably $10-15 -- or less than half the retail
price just 18 months ago.
At the time of ICANN's creation, NSI had
100% of the registration market for the .com, .net and .org TLDs. Today,
we estimate that NSI is registering less than 40% of new registrations
in those TLDs -- a market share drop of more than half in that same 18-month
period. There are still issues that must be dealt with in this area; some
registrars have not lived up to their contractual commitments, and ICANN
needs to ensure that they do. And indeed, there may be too many registrars;
94% of all registrations come from the 10 largest registrars, with the
other 80 or 90 active registrars sharing the other 6%. Name registration
is quickly becoming a commodity business, and a commodity business, with
commodity margins, will probably not support 100 vigorous competitors.
We are already starting to see some companies wishing to leave the business,
and we need to make as sure as we can that those departures do not impair
the ability of consumers and businesses to rely on names they have registered,
and that departures or even failures do not generate unreliability or
other forms of instability in the namespace itself. So while there are
still issues to be dealt with, I think it is widely recognized that ICANN
has been very successful in changing the retail name registration market
from a monopoly market to a highly competitive market.
2. Creation of a Cost-Effective,
Efficient Dispute Resolution System. A second significant accomplishment
has been the creation of the Uniform Dispute Resolution Policy, a way
to quickly and cheaply arbitrate certain domain name disputes. While domain
names themselves cannot be trademarked, it is certainly possible for domain
names to be confusingly similar to a trademarked name, or in other ways
to be inappropriately used by someone for illegitimate means. Since trademark
and other intellectual property rules differ from country to country,
enforcing those rights is complex and expensive.
One of the policies that was generated from
the ICANN bottom-up process early on was the need for a simple procedure
to resolve the clearest and most egregious violations on a global basis.
The result, after considerable work in a variety of ICANN forums, is the
UDRP, which one commentator recently noted is "widely viewed as a
model of dispute resolution for the 21st Century." The
UDRP is limited to certain very specific claims, is intended to require
only about $1,500 in costs and 45 days to invoke, and is required to be
included in all name registration contracts by all ICANN-accredited registrars,
thus providing the basis for global uniformity in the resolution of this
particular class of domain name disputes. Even though the UDRP is non-binding
(either party may take the dispute to court after an unfavorable UDRP
decision), it appears that has happened in only a few dozen out of over
2,000 decisions to date.
The UDRP is, I would submit, another very
positive accomplishment of ICANN during its short existence to date. As
of this writing, parties interested in further refinement of the UDRP
are already studying its design for possible revisions.
D. The Introduction of New Global
Top Level Domains.
That brings me to the subject of today's
hearings, which is really the third major accomplishment of ICANN in its
short existence: the creation of additional competition at the registry
(or wholesale) level of the namespace. To understand how much of an accomplishment
this was, and how difficult it has been to get to this point, we need
to start with some history, after which I will walk through the general
standard utilized, the criteria that were applied, the application process,
the evaluation process, and the selection process. I will then bring the
story up to date with a description of what has happened since the selections
Background. The Internet
as we know it today was not created with all of its present uses clearly
in mind. In fact, I can safely say (having been very much involved in
the very earliest days of the Internet) that no one had any idea how it
would develop in the hands of the general public, nor even that it would
ever reach public hands. Certainly there was little appreciation of the
increasingly critical role it would play in everyday life.
In those days, we were designing a communications
system intended for military application and used for experimental purposes
by the research and academic community, and not a system for commerce.
Internet addresses are numeric values, usually represented by four numbers
separated by "." (dots). This is sometimes called "dotted
notation" as in 192.136.34.07. In the earliest days, computers ("hosts")
were known by simple names such as "UCLA" or "USC-ISI".
As the system grew, especially after 1985 as the National Science Foundation
began growing its NSFNET, it became clear that a system of hierarchical
naming and addressing conventions would be needed.
At that time, seven so-called "Top Level
Domains" were created: .com for commercial, .net for networks, .org
for non-commercial organizations, .gov for government users, .mil for
the military, .edu for educational institutions, and .int for international
organizations. All domain names since that time (with an important exception
I will mention momentarily) have been subdivisions of those original seven
TLDs. Thus, wcom.com, to pick an example, is part of the .com top level
domain, and all messages sent to Vinton.G.Cerf@wcom.com are routed pursuant to
the information contained ultimately in the .com registry's distributed
database. In particular, that database resolves "wcom.com" into
a 32 bit address, such as 192.136.34.07 [note, this is not the actual
Internet address associated with the wcom.com domain name].
The exception mentioned earlier is the set
of so-called "country code" (or "cc") TLDs. The original
seven TLDs were once called "generic" TLDs and are now known
as "global" TLDs, meaning that there are theoretically no geographic
boundaries that constrain entries in those databases.(4)
In the early days of the Internet, one of the most important values to
the scientists seeking to incubate and grow this new thing was the spreading
of connectivity to as many parts of the world as possible. To help in
that, individual countries (and some other geographic areas) were delegated
their own TLDs, such as .au for Australia, or .jp for Japan, or .fr for
France. Operation of the registries for these ccTLDs was delegated to
a wide variety of people or entities, with the primary consideration being
a willingness to agree to operate them for the benefit of the citizens
of that geography. These original delegates were frequently academics,
sometimes government agencies, and sometimes local entrepreneurs; the
common thread was that they promised to use these TLDs to provide access
to this new thing called the Internet for local constituents. In this
way, the Internet, which started as a research experiment in American
universities, slowly became truly global. It is worth noting that the
Internet research project was international in its scope almost immediately.
It started in 1973, and by early 1975, University College London and the
Norwegian Defense Research Establishment were involved. Later, sites in
Italy and Germany became a part of the Internet research effort.
The original seven gTLDs were created in
the mid- to late-1980s; no new global TLD has been added to the namespace
since then. There are now some 245 ccTLDs, but as described, these were
intended to be for localized use, not as alternatives for global TLDs.
So as the Internet grew during the 1990s, demand for domain names grew
as well, but as a practical matter the only global (i.e., non-national)
TLDs in which businesses or individuals could freely register a domain
name were .com, .net and .org - all administered by Network Solutions,
Inc. under a contract with the National Science Foundation.
There is a long history about how this came
about, which I don't have time to tell, but suffice it to say that as
demand exploded, NSI could not effectively operate the registry within
the financial framework of its agreement with the National Science Foundation
and sought to remedy this by obtaining permission to charge users for
registration of names in the .com, .net and .org databases. Over time,
there came to be dissatisfaction with the service offered by NSI. In addition
(also for reasons too complicated to relate here), NSI was constrained
by its contract with NSF to charge exactly $70 for a two-year registration
with an annual $35 charge after the second year -- no exceptions, no changes.
As the number of name registrations climbed into the millions, many felt
that the charge far exceeded the cost of accepting the registration and
maintaining the database.
This unhappiness of a significant portion
of the Internet community was one of the driving forces behind a grass-roots
attempt to institutionalize the function of the original ICANN, the Information
Sciences Institute at the University of Southern California, a government
contractor that performed a set of functions known as the Internet Assigned
Numbers Authority (IANA). After almost three years of contentious debate,
the grass-roots effort failed to gel and the US Government (after extensive
public consultation) then called on the private sector to come forward
with a new kind of organization. The private sector responded by creating
ICANN, as a way to, among other things, encourage the addition of competition
at both the retail and wholesale levels of the namespace.
Standards for Introduction of New
TLDs. As described above, ICANN was able to introduce retail
competition relatively quickly after its creation, and this has produced
the expected benefits -- lower prices, more consumer choice, and innovation.
But the introduction of wholesale competition, because it involves actually
expanding the structure of the namespace, presented and continues to present
more risks. While most Internet engineers believe that some number of
additional TLDs could be added without serious risks of instability, there
is considerable uncertainty about how many could be added without adverse
side effects, and very few engineers have been willing to absolutely guarantee
that there was zero risk of instability. Given the increasingly critical
role the Internet now plays in everyday commercial and personal life,
the almost uniform consensus in the community was to be cautious and prudent
in this process.
For example, the White Paper asserted that
"expansion of gTLDs [should] proceed at a deliberate and controlled
pace to allow for evaluation of the impact of the new gTLDs and well-reasoned
evaluation of the domain space." In addition to concerns about the
technical stability of the Internet, many were concerned about potential
costs that rapid expansion of the TLD space might impose on business and
consumers. The World Intellectual Property Organization, which conducted
a study of intellectual property issues in connection with the DNS at
the request of the United States Government, concluded that new gTLDs
could be introduced if done "in a slow and controlled manner that
takes into account the efficacy of the proposed measures in reducing existing
problems." The Protocol Supporting Organization of ICANN (made up
of the Internet Engineering Task Force and other Internet engineering
and protocol development bodies) said it saw no technical problems with
the introduction of a "relatively small" number of new TLDs.
In fact, every entity or organization without
an economic stake in the answer that has examined this question has recommended
the same thing: a "small" or "limited" or "prudent"
number of new TLDs should be tried first, as a sort of proof of concept
or experiment. Once this "limited" number of new TLDs was introduced
-- and the suggested numbers roughly ranged from 1 to 10 -- and assuming
there were no adverse side effects, then additional TLDs could be introduced
if there was consumer demand for them.
The ICANN Structure and Procedures.
Because ICANN is a consensus development body that relies on
bottom-up policy development, the issues of whether and how to introduce
new gTLDs were first taken up by the Domain Name Supporting Organization
(DNSO), the ICANN constituent body responsible for name policy issues.
The DNSO organized a Working Group, which recommended that a small number
(6-10) of TLDs be initially introduced, and that the effects of that introduction
be evaluated before proceeding further. That recommendation was forwarded
to the Names Council, the executive body of the DNSO, which reviewed the
Working Group recommendation and public comments on it, and recommended
to the ICANN Board that it establish a "policy for the introduction
of new gTLDs in a measured and responsible way." The Names Council
suggested that "a limited number of new top-level domains be introduced
initially and that the future introduction of additional top-level domains
be done only after careful evaluation of the initial introduction."
Consistent with the ICANN bylaws, the ICANN
Board accepts the recommendations of Supporting Organizations if the recommendations
meet certain minimal standards designed to ensure that they truly represent
consensus recommendations. Thus, the Names Council recommendation was
published for public comments, and following the receipt of numerous public
comments, the ICANN staff in June 2000 issued a Discussion Draft seeking
public comments on a series of questions intended to lead to the adoption
of principles and procedures to be followed in a "measured and responsible
introduction" of a limited number of new TLDs.(5)
Following several thousand additional public comments, and considerable
discussion at a public meeting in Yokohama in July 2000, the ICANN Board
adopted a series of resolutions instructing its staff to begin the process
of accepting applications for a "proof of concept" for the introduction
of new TLDs.(6)
In early August, ICANN posted a detailed
discussion of the new TLD process it proposed to follow(7),
and in mid-August a detailed set of Criteria for Assessing TLD Proposals.(8)
These nine criteria have been constant throughout this process, and so
they bear repeating here:
1. The need to maintain the Internet's
This speaks for itself. ICANN's overriding
obligation is to protect the stability of the Internet, and all other
objectives are secondary. Thus, any proposal that could be shown to threaten
this stability (other than any risk inherent in any new TLD introduction)
was obviously unacceptable.
2. The extent to which selection
of the proposal would lead to an effective "proof of concept"
concerning the introduction of top-level domains in the future.
This too is largely self-explanatory. The
effort here was not to find the "best" application, however
that might be measured, but to ask the community to offer up a set of
options from which ICANN could select a limited number that, taken in
the aggregate, would satisfy the evaluation objectives of this proof of
concept. This is exactly the same approach that ICANN had previously taken
in the introduction of competitive registrars, and which had worked so
well there. The addition of multiple registrars to the NSI registries
required the creation of new interface software, since before this time
only one registrar had been able to direct new entries in those
registries. Thus, there was some experimental
effort required to make sure that the software was ready for use by a
larger number of simultaneous registrars. ICANN first created a "test-bed,"
asked for expressions of interest from the community, and accredited only
five new registrars for a period of a few months, while they and NSI worked
out the bugs in the interface software. As soon as the test-bed was completed,
ICANN accredited large numbers of registrars, now exceeding 180.
Here, the concept is similar: from options
offered up from the community, create a limited number of new TLDs to
ensure that the DNS can accept, both technically and practically, these
additions without impairing stability in any way. Once that is proven,
additional TLDs can be created as appropriate.
3. The enhancement of competition
for registration services.
Obviously, this is the principal reason for
adding new TLDs, so one criterion for determining which applications to
accept initially is how effective they are likely to be in creating new
competition for the NSI registries. Of course, competition takes many
forms; here, one form would be analogous to .com -- a global, unrestricted
registry focusing on business. To compete in this way requires not only
desire, but the capacity to effectively compete with a competitor with
high brand awareness (.com has almost become a generic term), a very significant
marketing budget, and a large installed base of registered names which
will produce some level of renewals more or less automatically. To compete
successfully on a global basis under these circumstances requires a significant
capital investment, very significant technical expertise (running a database
of several million names that gets hundreds of simultaneous queries every
second is a complicated matter), and a substantial marketing budget to
build the kind of brand equity that will be necessary to compete effectively
with, for example, .com.
Another way to introduce competition into
the wholesale part of the market is to offer a different kind of product
-- not a global unrestricted domain, but various kinds of limited or restricted
registries that might appeal to specific different sectors of the market.
To use a television analogy, narrowcasting instead of broadcasting. Here,
capital and marketing expenses may be lower, but other kinds of service
characteristics may be more important.
ICANN's purpose with this criteria was to
invite a broad range of competitive options, from which it could select
a menu that, taken as a whole, would offer a number of different competitive
alternatives to consumers of domain name services.
4. The enhancement of the utility
of the DNS.
In addition to competition, one must reasonably
consider the practical effects of the introduction of new TLDs. The names
registered in the DNS are intended to be used by people, and sound engineering
requires that human factors be taken into account.
5. The extent to which the proposal
would meet previously unmet types of needs.
If it is assumed that the DNS should meet
a diversity of needs, it would be a positive value if a proposed TLD appeared
to meet any previously unmet needs of the Internet community.
6. The extent to which the proposal
would enhance the diversity of the DNS and of registration services generally.
Here, what was sought was diversity of all
kinds, in the hopes of creating the broadest possible -- and thus most
instructive -- experiment within the limitations recommended (i.e., a
small number of new top level domains). So, the published criteria encouraged
the submission of proposals for different kinds of TLDs (open or closed,
non-commercial or commercial, personal or business-oriented, etc.) The
criteria also sought diverse business models and proposals from different
geographic regions, for the same reasons.
7. The evaluation of delegation of
policy-formulation functions for special-purpose TLDs to appropriate organizations.
For those proposals that envisioned restricted
or special-purpose TLDs, this criterion recognized that development of
policies for the TLD would best be done by a "sponsoring organization"
that could demonstrate that it would include participation of the segments
of the communities that would be most affected by the TLD. Thus, with
this class of application, the representativeness of the sponsoring organization
was a very important criterion in the evaluation process.
8. Appropriate protections of rights
of others in connection with the operation of the TLD.
Any new TLD is likely to have an initial
"land rush" when it first starts operations as people seek the
most desirable names. In addition, every new TLD offers the potential
opportunity for cybersquatting and other inappropriate name registration
practices. This criterion sought information about how the applicant proposed
to deal with these issues, and also how it proposed to provide appropriate
mechanisms to resolve domain name disputes.
9. The completeness of the proposals
submitted and the extent to which they demonstrate realistic business,
financial, technical, and operational plans and sound analysis of market
Finally, this criterion simply emphasized
that, since the effort was a "proof of concept," the soundness
and completeness of the application and the business plan would be important
elements of the selection process. This was not intended to be an experiment
in how well the DNS or the Internet could survive the business failure
of a new TLD operator. Nor was it intended to be clairvoyant with regard
to the outcome of any particular proposal. Thus, to the extent possible,
those applications that appeared to have the soundest business plans,
based on the most realistic estimates of likely outcomes.
The Application Process. The
application process required the filing of a detailed proposal speaking
to all the criteria outlined above. It recommended that applicants retain
professional assistance from technical, financial and management advisers,
and lawyers. And perhaps most controversially, it required a non-refundable
application fee of $50,000. A brief explanation of this particular requirement
may be useful.
ICANN is a self-funding organization. It
has no capital, and no shareholders from which to raise capital. It must
recover its costs from the various constituent units that benefit from
ICANN's processes and procedures -- today, those costs are borne by address
registries, name registries, and registrars. Its annual expenditures to
date have been in the $4-5 million range, covering employee salaries and
expenses (there are now 14 employees), and a wide range of other expenditures
associated with operating in a global setting.
Thus, there was no ready source of funds
to pay for the process of introducing new TLDs, and the ICANN Board determined
that this, like all other ICANN activities, should be a self-funded effort,
with the costs of the process borne by those seeking the new TLDs. At
that point, ICANN estimated the potential costs of this process, including
the retention of technical and financial advisers, legal advice, the logistics
of the process, and the potential cost of litigation pursued by those
unhappy with the results. While obviously all these elements were highly
uncertain, based on its best judgment of how many applications were likely
to come in and what the likely costs would be, and incidentally only after
receiving public comments, ICANN established a $50,000 fee. As it turns
out, there were more applications than expected, and thus the absolute
costs of processing and reviewing them were higher than expected; about
half the application revenues have already been used to cover costs of
the process to date, with considerable work left to do and still with
the potential for litigation at the end of the process. To date, it appears
that the fact of more applications and higher costs of review and evaluation
than expected have cancelled each other out, and so it appears that the
fees adopted were about right in creating the funds necessary to carry
out this process.
I know there have been complaints by some
that they were foreclosed from this process because they simply could
not afford the $50,000 application fee, and I am sympathetic to these
concerns. But there are three practical responses that, in my view, make
it clear that this is not a fair criticism of the process. First, the
process had to be self-funding; there simply was no other option, since
ICANN has no general source of funds. Based on costs to date and those
projected, it certainly does not seem that the fee was set too high. While
there are still application fee receipts that remain unspent, the process
is not over, and it has already consumed half of the fees collected. Second,
and as importantly, it is highly unlikely that any individual or entity
that could not afford the application fee would have the resources to
be able to operate a successful and scalable TLD registry. The capital
and operating costs of even a small registry are thought to be considerable,
and especially if the goal is to operate a registry that charged low or
no fees for name registrations (many of the persons and entities advancing
this particular complaint are non-profit or public interest bodies), those
fees would not likely cover the costs of operation, much less the necessary
start-up and capital costs. Of course, it is possible that, if an organization
that would otherwise have difficulty managing the costs of operating a
TLD registry were in fact awarded a new TLD, it might be able to raise
the funds through subsequent contributions or grants or the like, but
this leads us directly to the third point.
This effort was not a contest to find the
most qualified, or the most worthy, or the most attractive for any reason
of the various applicants. ICANN is not and should not be in the business
of making value judgments. What ICANN is about is protecting the stability
of the Internet and, to the extent consistent with that goal, increasing
competition and competitive options for consumers of domain name services.
Thus, what ICANN was doing here was an experiment, a proof of concept,
an attempt to find a limited number of appropriate applicants to test
what happens when new TLDs of various kinds are added to the namespace
today -- a namespace that is vastly different in size and in application
than that which existed more than 15 years ago when the first seven global
TLDs and the ccTLDs were created.
Because this was a proof of concept, the
emphasis was on diverse business models, technical capacity, and diversity
of geography and focus -- and not on some weighing of the relative merits,
however measured, of the applicants. Indeed, a serious attempt was made
to avoid otherwise normal business risks, such as limits on capital or
other resources, so that forseeably likely business failures did not interfere
with the data collection and evaluation process of this experiment. Thus,
it would have been impossible to accept any application which relied on
the mere hope of obtaining funding if an application was accepted, and
indeed, several of the applicants not selected in the evaluation process
were thought to be deficient just on that point.
Under these circumstances, it was not appropriate
to encourage applications by those with limited resources, since those
limitations would almost certainly result in their not being selected.
Thus, setting the fee to recover expected costs, without regard to the
effect it had on applications, seemed then (and seems today) the logical
approach. Once this experiment is over, and assuming it demonstrates that
adding new TLDs in a measured way does not threaten the stability of the
DNS or the Internet, I would hope that processes could be developed to
both expedite and significantly reduce the cost of new TLD applications
or, at a minimum, to deal with special cases of TLDs with very limited
scope, scale and cost.
The Evaluation Procedure.
Forty-seven applications were submitted by the deadline established; three
of those were withdrawn for various reasons, and the remaining 44 were
then published on ICANN's website, open to public comments, and subjected
to an extensive evaluation, applying the criteria set forth in the various
materials previously published by ICANN. More than 4,000 public comments
were received. The applications and the public comments were carefully
reviewed by technical, financial and legal experts, and the result of
that evaluation -- a 326-page staff report summarizing the public comments
and the staff evaluation -- was itself posted on the ICANN website for
public comment and review .by the Board of Directors of ICANN.(9) Another 1,000 public comments were received on
the staff report. The Board was provided with regular status reports,
interim results of the staff evaluations, and of course had access to
the public comments as they were filed.
There has been some criticism of the fact
that the full staff evaluation was not available to the public -- and
thus to the applicants -- until November, only days before the actual
Board meeting. Obviously, it would have been much better to produce this
earlier, and we tried to do so. But in fact the timing of the release
of the staff report was largely the product of the bottom-up process that
ICANN follows to generate consensus. An important ingredient in the staff
evaluations was the substance of the voluminous -- over 5000 -- public
comments produced in the month after the applications were posted. ICANN's
job is to identify consensus, and thus input from the community is a critical
part of any Board decision. Getting that community input, considering
it, and completing the technical and financial evaluations was a massive
It would have been preferable to have issued
the staff report earlier. But on the other hand, in the six days between
the posting of the report and the Board meeting, ICANN received more than
1,000 additional public comments on the staff report, many from the applicants
responding to the evaluation of their particular application. The ultimate
question is whether the Board got sufficient timely information on which
to base its selection decisions, bearing in mind the objective of the
exercise. I believe it did.
At its Annual Meeting in Los Angeles in November
2000, the ICANN Board devoted most of the standard public forum day immediately
preceding the Board meeting to the new TLD issue, with presentations by
the staff of their findings, public comments, and short presentations
from the applicants. Another point of criticism by some has been the short
time -- three minutes -- allowed during this public forum for presentations
by each of the applicants, but oral presentations were never intended
to be the sole or primary source of information for the Board. Voluminous
applications (with many hundreds of pages) had been filed by each applicant;
many of them had received and answered clarifying questions from the staff;
and many of them had provided additional material by filing material on
the ICANN public comment page (every one of the 5,000+ comments was read
by ICANN staff). The Board had access to the applications and to the staff
evaluations well ahead of the public Board meeting at which the applications
were reviewed. The opportunity to make a presentation at the public forum
was simply the final step in an extensive process, available so that any
last-minute questions could be asked or points made.
Since there were 44 applicants, nearly all
of whom wished to speak, and since the time available (given the other
parts of the community who also wished to be heard) was limited to about
two hours, three minutes was simply all the time available. Most used
it wisely, pointing out the particular strengths of their applications.
Some disappointed applicants have also complained
that ICANN staff refused to talk with them, or let them respond to concerns
raised by their applications. This is not accurate; what ICANN staff refused
to do is have private conversations with the applicants, and
this derives from the very nature of ICANN as an entity. ICANN is a consensus
development body, not a regulatory agency; its decisions are intended
to reflect consensus in the Internet community, not simply the policy
preferences of those who happen to sit on its Board at any given moment.
For this process to work, the vast bulk of ICANN's work must be transparent
to the public, and so with very rare exceptions (such as matters dealing
with personnel issues), everything ICANN does it does in public. (In fact,
one of the three applications that were withdrawn resulted from the applicants'
unwillingness to allow significant material in their application to be
posted on ICANN's website.) If the public was going to have a real opportunity
to comment on the applications, the applications themselves needed to
be public, and any substantive discussion of them had to be public as
In an effort to help this process, and still
get questions answered, ICANN staff frequently took email or other private
questions, reformulated them to make them more generically useful, and
then posted them on the website as FAQs. In addition, staff encouraged
applicants to post any information they wished on the public comment pages,
where it would be read by ICANN staff, the ICANN Board and also by any
interested observer. What staff would not do, and what was evidently very
frustrating to many of the applicants that had not previously had any
experience with the open structure and operations of ICANN, was to have
private substantive discussions with the applicants.
It is easy to understand this frustration,
especially for those disappointed applicants who had not previously participated
in the ICANN process and, as a result, did not understand what ICANN is
and how it operates and thus were surprised at the transparency of the
entire process. Still, it is hard to see how any other process could have
been followed consistent with ICANN's consensus development process. Without
access to the entirety of the information about each applicant and each
application that was available to the Board, the Board would not have
had the benefit of public comments on some (often significant) factors,
and it would have been hard to justify its selections as deriving from
a consensus development process.
The Selection Process. To
understand the selection process, we must go back to first principles.
The goal here was not to have a contest and pick winners; it was not to
decide who "deserved" to have a new TLD; it was not even to
attempt to predict the kind or type of TLDs that might get public acceptance.
The goal, articulated plainly from the beginning of the process more than
a year ago, was to identify from suggestions by the community a limited
number of diverse TLDs that could be introduced into the namespace in
a prudent and controlled manner so that the world could test whether the
addition of new global TLDs was feasible without destabilizing the DNS
or producing other bad consequences.
This was not a race, with the swiftest automatically
the winner. It was a process that was intended to enable an experiment,
a proof of concept, in which private entities were invited to participate
if they chose to do so -- and those who did choose to participate did
so voluntarily, knowing that the odds of being selected were not high,
that the criteria for being included in this experiment were in some measure
subjective, and that the goal was the production of experimental information
that could be evaluated. Of course, when many more applications were received
than anyone had suggested should be prudently introduced at this stage,
some evaluation was necessary to attempt to identify those suggestions
that might best fit the experimental parameters that had been laid down.
But this was never a process in which the absolute or relative merit of
the particular application was determinative.
Many applications with likely merit were
necessarily not going to be selected, if the goal was a small number (remember,
the entire range of responsible suggestions for introducing new TLDs was
from one to 10 new ones). And since one objective was diversity -- of
business model, of geography, of type of registry -- it was highly likely
that some qualified applications would not be selected -- both because
prudence required the addition of only a small number of TLDs, and because
our proof of concept required data from a diverse set of new TLDs. This
was especially true of those applications seeking open, global TLDs; while
two were selected, about half of the 44 applications sought such a charter.
But it was also true of others; .geo received a very positive evaluation
from the staff, but the Board felt that, at this proof of concept stage,
there were in fact potential risks to the operation of the DNS that could
not be fully evaluated without consultation with the technical support
organization(s) associated with ICANN.
Thus, the Board considered every one of the
44 remaining applications at its meeting on November 16, 2000, measuring
them against their collective judgment about how well they would serve
to carry out the test that was being considered. In a meeting that lasted
more than six hours, the Board methodically reviewed, and either set aside
or retained for further evaluation, application after application, until
it was left with approximately 10 applications that seemed to have broad
consensus support. After further, more focused discussion, that number
was pared to the seven that were ultimately selected, and which had almost
unanimous Board support: .biz, .info, .pro, .aero, .coop, .museum, and
.name.(10) In the aggregate, the Board
concluded that this group provided enough diversity of business models
and other relevant considerations so as to form an acceptable test bed
or proof of concept.
The various TLDs have very different intended
purposes, and that is the strength of the group in the aggregate. Two
-- .biz and .info -- were advanced as essentially alternatives to .com
-- global, business-oriented registries aimed at capturing millions of
registered names around the world. In order to compete with .com -- which
has a recognized brand, a large installed base that produces a regular
stream of renewals, and a very substantial marketing budget -- these particular
applicants assumed they would need a significant investment in both capital
equipment and marketing. The Board felt that these applicants seemed most
capable of bringing the necessary resources to bear to test whether anyone
can effectively compete with .com after the latter's significant head
Two other TLDs -- .pro and .name -- were
aimed at individuals rather than businesses, but in very different ways.
.pro was aimed at licensed professionals, while .name was aimed at any
individual. The other three -- .aero (aerospace industry), .coop (for
cooperatives), and .museum (for museums) -- were all restricted TLDs,
aimed at an industry or a business method or a type of entity, and added
to the diversity of this experimental collection of TLDs.
ICANN's objectives -- and by that we mean
to say the objectives of the general Internet community, which ICANN tries
to represent -- were to introduce a small number of various kinds of new
TLDs into the namespace in a prudent fashion, see what happened, and then,
if appropriate, based on those results, move forward with additional new
TLDs. It is certainly conceivable that some different subset of the applications
it had before it would have met that objective as well as those chosen,
but the real question is whether the choices were reasonable, and likely
to produce the necessary information on which future introductions could
be based. It is also possible, as some of those not selected have complained,
that those selected will have a head start (to the extent that matters)
over future TLD applicants, but this would be an inevitable consequence
of any selection of less than all applicants. Those who were not selected,
no matter who they are, were predictably going to be unhappy, and those
who were selected were predictably going to be glad, but neither was an
ICANN goal. ICANN's goal, and its responsibility, was to find a limited
collection of diverse new TLDs that could be prudently added to the namespace
while minimizing any risk of instability. While time will tell, at this
point we believe we faithfully carried out that responsibility.
The Post-Selection Process. Since
November, we have been in the process of drafting and negotiating agreements
with the selected applicants. Since these agreements will hopefully be
templates for future agreements, we are taking great care to make sure
that the structure and terms are replicable in different environments.
Since these agreements will contain the promises and commitments under
which the applicants will have to live for some time, the applicants are
being very careful. The result is slow progress, but progress. We are
hopeful that we will be able to complete the first draft agreements within
a few weeks. The Board will then be asked to assess whether the agreements
reflect the proposals that were selected and, if so, to approve the agreements.
Shortly thereafter, this great experiment will begin. We are all looking
forward to that time.
Of course, it cannot be stressed enough that
no one knows for sure what the effects of this experiment will be. Since
there have been no new global TLDs introduced for more than a decade,
the Internet is a vastly different space than it was the last time this
happened. Of course, there have been a number of country code TLDs introduced
over that period, and since some of those have recently begun to function
in a way quite analogous to a global TLD, it may be that we will be able
to conclude that the DNS can readily absorb more new global TLDs. But
there has never been an introduction of as many as seven new global TLDs
simultaneously, with the possibility of a land rush that is inherent in
that fact. There has never been a highly visible introduction of multiple
new TLDs in the context of an Internet that has become a principal global
medium for commerce and communication. We do not know whether the introduction
of a number of new TLDs -- especially combined with the relatively new
phenomenon of the use of ccTLDs in a fashion never intended (after all,
.tv stands for Tuvalu, not television, no matter what its marketers say)
-- will create consumer confusion, or will impair the functioning of various
kinds of software that has been written to assume that .com is the most
likely domain for any address.
In short, it is not absolutely clear what
effects these introductions will have on the stability of the DNS or how
to introduce new TLDs in a way that minimizes harmful side-effects, and
that is precisely why we are conducting this experiment. The results will
guide our future actions.
One of ICANN's primary missions
is to preserve the integrity and stability of the Internet through prudent
oversight and management of the DNS by bottom-up, global, representative
consensus development. Like location in real estate, the three most important
goals of ICANN are stability, stability and stability. Once there is consensus
that stability is not threatened, ICANN is then charged with seeking to
increase competition and diversity, both very important but secondary
goals. A competitive Internet that does not function is not useful. An
Internet in which anyone can obtain the domain name of their choice, but
where the DNS does not function when someone seeks to find a particular
website, is also not useful.
In its short life, ICANN has some real accomplishments
-- made more impressive by the inherent difficulty of developing global
consensus on anything, but especially on issues as complex and contentious
as those facing ICANN. It has achieved these accomplishments by hewing
to its first and guiding principle -- to maintain a stable, functional
DNS -- and within those limits by seeking to increase competitive options
and efficient dispute resolution. This same principle has guided the careful,
prudent way in which ICANN has approached the introduction of new global
TLDs, really for the first time in the history of the Internet as we know
ICANN's processes are and have been transparent.
The goals and procedures were derived from public comments, clearly laid
out at the beginning of the process, and all decisions were made in full
public view. Given the importance of care and prudence in the process,
and the potentially devastating results of a misstep, ICANN has and will
continue to err on the side of caution. This may mean slower progress
than some would like, but it will also reduce and hopefully eliminate
the potential for the catastrophic effects on business and personal use
of the Internet that malfunction or other instability of the DNS would
1 My curriculum vitae is attached.
2 I have attached to this testimony
a time line that describes the chronology of the debate over new Top Level
3 The White Paper was a policy
statement published by the Department of Commerce on June 10, 1998. See
Management of Internet Names and Addresses, 63 Fed. Reg.
4 Of course, in fact entries in
.gov, .mil, and for the most part .edu relate only to the United States,
but the other global TLDs are open to entries from all over the world.
5 See generally ICANN Yokohama
Meeting Topic: Introduction of New Top-Level Domains, at http://www.icann.org/yokohama/new-tld-topic.htm.
6 See Resolutions of the ICANN
Board on New TLDs, at http://www.icann.org/tlds/new-tld-resolutions-16jul00.htm
7 See New TLD Application
Process Overview, at http://www.icann.org/tlds/application-process-03aug00.htm
8 See Criteria for Assessing
TLD Proposals, at http://www.icann.org/tlds/tld-criteria-15aug00.htm
9 See Report on New TLD Applications,
10 See http://www.icann.org/minutes/prelim-report-16nov00.htm#00.89