-----------------------------------------------------------------------
Comments of Kent Crispin, Chair, gTLD-MoU Policy Advisory Body
In the matter of:
HEARING OF THE U.S. HOUSE OF REPRESENTATIVES COMMITTEE ON SCIENCE
SUBCOMMITTEE ON BASIC RESEARCH
SUBCOMMITTEE ON TECHNOLOGY
Subject:
"The Domain Name System: Where Do We Go From Here?"
Honored Committee Chairpersons and Members:
I wish to clarify two matters mentioned in the Hearing Charter for
the above proceeding:
A. Stakeholder Representation in the CORE Model
From the Charter, section entitled "Internet Governance":
"However, the one criticism of the CORE proposal is that the
POC is too exclusive and does not properly represent all the
key players in terms of governing the Internet. An additional
complaint is that the CORE proposal does not address the
functions of IANA."
The perceived lack of representation of the POC is a matter that
has concerned the POC greatly, and for some time they have been
developing an alternative that will strongly address this issue.
This proposal has been available for public comment since Dec
16, 1997, and may be seen on the gTLD-MoU web site at
http://www.gtld-mou.org/docs/rfcs.html#97-04
Briefly, the proposal is to substantially increase the
membership of the POC to about twice its current size, and for
half of the total membership to be chosen by election. The
elections will be conducted by the members of the gTLD-MoU
Policy Advisory Body (PAB), but anyone may be nominated.
Membership in PAB is very open -- there are no dues, and
organizations of any size or composition may join. Currently it
is a requirement that an organization must sign the MoU to
become a member of PAB, but, as part of this proposal, that
requirement is being relaxed, and a signature on a non-binding
statement of principles will be sufficient to become a member of
PAB.
The net effect is that an open assembly of Internet stakeholders
will elect half of the POC, and that anyone who is interested
may participate in this election.
It is true that the MoU does not address the functions of IANA.
It is believed that the IANA is autonomous from the POC, and
should remain that way. In the MoU model, IANA remains in
charge of the root zone of the DNS, and delegates to POC policy
oversight over the generic TLDs only.
In this model it would be inappropriate for the MoU to dictate
structure for the IANA, or to propose how it should be managed.
IANA has been preparing on its own to transition from US
Government funding -- the POC believes that IANA is capable of
doing this on its own.
B. Competitive Issues and the FTC Paper
From the Charter, section titled "The Registry Issue":
"The CORE proposal, however, would establish one non-profit
shared registry containing all the individual registries.
Under their proposal, NSI would lose the .com, .edu and .org
registries (these registries would become part of the
non-profit shared-registry) and NSI would compete, as a
registrar, for business against all the other registrars
(approx. 100) without the benefit of controlling a registry.
The CORE group argues that registries are largely
administrative, back-office operations that offer little in
added value for customers, and therefore there is no need for
competition at the registry level. The Green Paper argues
that only through the profit-motive and competition among
various for-profit companies will the money be generated to
maintain the registries that will allow the Internet to
succeed."
"The Administration acknowledges that under its
registry-scheme, 'switching costs and lock-in issues' could
arise in instances in which users wish to change from one
registry company to another. Once it becomes widely known,
for example, that a certain business can be found under the
.store Domain Name, the operator of that register could
increase the price charged to the locked-in company.
Moreover, the cost of switching to a different company may
lead to a situation in which it is difficult to move from one
registry to another. On balance, however, the Administration
believes that the benefits of competition among the registries
outweigh the possible burdens caused by this issue. The
Federal Trade Commission is working on this issue and has
submitted comments to the Commerce Department on this topic."
Indeed, the FTC did present comments to the Green Paper.
Unfortunately, the issues are complex and time was short; and
consequently those comments leave significant gaps.
1. The FTC document does not describe the CORE registry
model, and instead speaks of a model of "competing non-profit
registries". Consequently, much of its analysis is at an
oblique angle to the actual proposals under consideration.
The CORE registry model is not, as is commonly supposed, a
monolithic entity. Instead, the CORE model has three
components: CORE itself, which is a non-profit association
whose function is primarily operational; POC/PAB, which
together control policy; and one or more database operators
(DBOs), who operate under competitively bid contracts to CORE.
Each DBO operates a database for one or more gTLDs. These
databases run the CORE Shared Registry System (SRS). The SRS
is explicitly designed so that it is straightforward to move
the data for one gTLD from one DBO to another.
Currently there is one DBO, Emergent. Emergent is a special
case, because their operation of the registry is a phase of
their development contract for the SRS. It is not intended
that they run the database indefinitely.
Having multiple DBOs is desirable from several perspectives:
First, it allows multiple dispersed small entities, each of
which can take over for the other, in case it is necessary.
This is generally considered a superior model technically,
when compared with a single monolithic entity. Second, there
are the obvious benefits that accrue from having several
entities competing to provide service. And third, there are
undeniable political advantages to having DBOs located in
different countries.
On the other hand, there are natural economies of scale that
are involved -- the incremental cost of adding a new TLD to an
already existing database are very small. It is unlikely,
therefore, that there would be a large number of DBOs -- a
balance will be struck between the economies of scale and the
various advantages mentioned above.
To summarize, CORE operates a registry that can be segmented
on gTLD boundaries, and can also be distributed on that basis.
Each segment is run by a DBO, under contract to CORE. Those
contracts are let on a competitive basis, and thus the DBOs
definitely compete. Each segment (database) is run on a
non-profit, cost-recovery basis, as far as CORE is concerned,
but the DBOs can be for-profit companies.
In addition, explicit and direct control over CORE policy
comes from POC/PAB. These bodies are not composed of CORE
registrars, have no financial interest in CORE, and are
explicitly designed to give representation to users and
stakeholders in the DNS. Thus the goals of POC and PAB,
especially as far as pricing and service are concerned, are
far more likely to be aligned with the end-users of the DNS
than they are with the DBOs, or even the CORE registrars.
2. Instead of dealing with the CORE model, the FTC paper
makes a strawman contrast between "competing non-profit
registries" (non-profits) and "competing for-profit
registries" (for-profits).
In this hypothetical and theoretical context the paper first
argues that the competitive environment surrounding registries
may be sufficient to avoid "lock-in" effects, and then draws
two conclusions:
First, that competing for-profit registries would be more
likely than competing non-profit registries to bring the
"potential benefits to customers from enhanced competition
-- such as price reductions and quality improvements..."
There is essentially no argument given, however, to support
this conclusion, and, given that competition would be
operating in both cases, it isn't clear why the benefits of
competition should not apply to both.
Second, that competing non-profit registries would be just
as likely to exploit "lock-in" as competing for-profit
registries. This is a curious argument, given that the
first part of the paper goes to some lengths to minimize the
probability of "lock-in". More than that, however, the
paper gives no explanation of why, if competition can
mitigate lock-in, it wouldn't mitigate it for competitive
non-profits.
However, neither of these conclusions is really germane to the
CORE model.
3. The FTC paper, understandably enough given the FTC's
mission, makes in implicit assumption of a regulatory regime
that remains under the control of US law. However, this
clearly will not remain the case -- the Green Paper itself is
operating under a mandate to internationalization, and
promises that the US Government will be completely out of the
picture in the year 2000.
This assumption is most clearly spelled out in the following
sentence from footnote 14: "If significant lock-in problems
develop, NTIA may wish to revisit these technical issues and
consider means for enhancing portability." Since lock-in by
definition is a long-term problem, the NTIA will long be out
of the picture by the time it becomes a major problem in
practice, and will be in no position to revisit the issues.
While the FTC paper minimizes concerns about lock-in, it
concedes that it may indeed be a problem. It seems fairly
clear, therefore, that the international regulatory issues
must be explored by the NTIA before it puts anything in place.
Note that the CORE model, with the contractually enforced
oversight of the POC/PAB, does address this issue.
4. The FTC paper considers switching costs and lock-in as a
"plausible" theoretical possibility. However, its analysis
vastly underestimates the extent of this problem. As has
repeatedly been pointed out, domain names are parts of the
hyperlinks that make up the World Wide Web, and those links
can be embedded in tens and hundreds of thousands of web pages
all over the world. A presence on the web is valuable to
precisely the extent that the site is reachable. A change in
domain name invalidates all the links to the page, and thus
destroys almost all the value of the web presence. For a
web-based business this is not an inconvenience, it is a
catastrophe.
5. The FTC paper properly focused only on a narrow set of
economic issues, and did not address other issues that may
have overriding significance. For example, the FTC did not
address the very real costs to consumers of inconsistent
policy (for example, in dispute resolution) that multiple
competing for-profit registries would engender. (It is of
course also true that this cost would be born in the
hypothetical case of competing non-profits -- the CORE model
explicitly addressed this issue by providing a consistent
policy framework for all registrars and database operators.)
In summary, the FTC paper does an excellent job describing
several theoretical economic2 results. However, it does not
compare the actual competing proposals, it ignores the problem
of regulation in an international context, and gives only a
cursory analysis of the monopoly problems inherent in DNS.
Thank you very much for your consideration
Kent Crispin, Chair
gTLD-MoU Policy Advisory Body
-- Kent Crispin, PAB Chair "No reason to get excited", kent@songbird.com the thief he kindly spoke... PGP fingerprint: B1 8B 72 ED 55 21 5E 44 61 F4 58 0F 72 10 65 55 http://songbird.com/kent/pgp_key.html