Comments on draft-iahc-gTLDspec-00 - Resend (Formatted)

[Alan Sullivan <sully@frontiernet.net>]

Introduction -
        The comments made with regard to the IAHC proposal 
        December 19,1996 are indented. The actually text from the 
        draft-iahc-gTLDspec-00 is included for reference of 
        comments. The general thesis of these comments is: 
        "Shared gTLDs create many market barriers to registries."
        IAHC should adopt a process that allows a free market of
        competing registries to offer exclusive service of a 
        particular domain name space. This does not mean that 
        the name space is not   public - indeed it is a public 
        trust that is being cared for by competing custodial 
        registries.


IAHC has decided to create further gTLDs as a means of increasing the
level of competitive supply and access to the gTLD space for the 
global Internet community.

In addition IAHC has decided to ensure that the new gTLDs are all to 
be operated in a shared fashion across a number of domain operators. 
This is regarded as a secondary mechanism to ensure that each of these
additional gTLDs are operated in an environment of a competitive 
market for access to these name spaces, and the same checks and 
balances of the competitive inter-gTLD access space also apply to 
the intra-gTLD space through this mechanism.

        Another way to ensure that additional gTLDs are operated in 
        an environment of a competitive market would be to allow 
        one registry per a gTLD and not to limit the number of 
        gTLDs. 

        There seem to be conflicting edicts - 

* The .com, .org and .net are classified as existing gTLDs. IAHC 
will define an additional set of seven (7) gTLDs. Any addtional 
gTLDs will be defined under the aegis and policy coordination of 
the CORE, described below..

        and

IAHC has decided to create further gTLDs as a means of increasing the
level of competitive supply and access to the gTLD space for the 
global Internet community.

        By the very definition given above IAHC will define 7 
        additional gTLDs - not the market. By what means has 
        IAHC determined that the gTLDs they decide to add 
        will necessarily increase a level of supply demanded 
        by the market? One could argue this is not
        competitive since IAHC will limit the number of total 
        additional to some arbitrary fixed number (7).   
        Registries can provide  little market differentiation 
        therefore a non-competitive regulated environment akin to 
        public utilities will be created. 

It is noted by IAHC that competitive shared access to any domain
registry is a useful market control mechanism to reduce the risk of
monopolistic trading practises. This is a core principle in IAHC's
decision for administration and management of gTLD space. IAHC further
recommends this policy be considered by the relevant ISO 3166 TLD
national authorities and administrations.

        IAHC should also note that competitive registries that have
        exclusive use of a domain name space also provide a useful
        market control mechanism to reduce monopolistic
trading                 
        practices.  What does IAHC mean by "competitive shared
access"?         
        One could argue that what makes sense - is "a competitive 
        non-shared access" or "a non-competitiveshared access". 
        One could argue that IAHC is really describing is 
        "a non-competitive shared access." This removes any 
        incentive for good service much like what is seen in the 
        essentially monopolistic regulated utilities industry. 
        It should be noted many industries are moving away from such 
        non-competitive models.

It is noted that an immediate large increase of gTLDs does increase 
the risk of impacting the coherence of the overall gTLD space, 
through increasing the level of complexity in the process of 
determining which is the appropriate gTLD for any end consumer of 
these services to choose.

        How does increasing the number of choices a customer have in
        choosing domain spaces increase complexity and impact 
        coherence of the Internet?

Therefore, IAHC will adopt an initially conservative approach to the
number of additional gTLDs, through the initial expansion in the 
first annual round to a total of ten gTLDs. This entails the creation 
of seven new gTLDs.
 
        Let the market forces determine the number of gTLDs. Why do 
        you want ten gTLDs? This seems like an arbitrary number. 
        What not choose 100 gTLDs?



 
A REGISTRAR is the entity which is authorized to enter and modify a
registry's data, based on customer contact.

        This definition for registrar seems straight forward and 
        makes sense a long as one assumes a registrar is THE entity 
        and NOT one of MANY entities as is suggested by:

A registry may have multiple registrars. A registrar may be authorized
for multiple registries.

        This definition of 'registrar' starts to resemble that of
        'agent'.

In determining the numbers of registrars for a registry, a 
fundamental issue is one of trust and cooperation. If multiple 
registrars share a registry and they have a fully cooperative 
relationship, the repository for the registry can be maintained 
using fully distributed data base technology.

If the registrars for a registry have a mutually suspicious 
relationship -- as is typical in competitive business circumstances --
then the repository for that registry needs to be operated by a 
trusted, independent third party, with simple rules of access. 
Particularly appropriate rules include fair use and assigning 
precedence for competing requests on first come, first served basis.

        How do you keep hostile non-cooperative registrars 
        coordinated for any registry? By definition one could argue 
        these are not registrars. The registrars are more akin to 
        ticket brokers or travel agents.

Oversight responsibility for a registry rests with a STEWARD. If the
registry is a monopoly, the steward, the registrar, and the operator
of the repository are typically one in the same. When multiple, 
competing registrars exist for a registry, it is appropriate to have 
independent stewardship. This ensures operation of the registry as a 
public trust for the Internet. It assesses performance of the 
repository and the registrars, enacting changes as necessary.

        Who will act as a steward and how will they be compensated?

The ultimate goal for registries operating in the gTLD space is that 
any qualified entity may be a registrar, and that every registrar 
shares responsibility for registering domain names in all gTLDs. 
Noting its concern for managing change to a critical Internet 
resource, IAHC believes that the number of additional Registrars 
authorized in the next year should be limited to between twenty and 
thirty, with additional Registrars added at the rate of twenty to 
thirty per year, subject to an annual review of the efficient 
functioning of the system. The limit on the number of Registrars 
should be completely removed when the capacity to register domain 
names in the present gTLD .com, .org and .net domains become shared 
among all Registrars.

        The ultimate goal suggests name agents around the world. 
        Even though the concern for managing change is noble, 
        the arbitrary limitation of the authorized registrars creates 
        a near term restraint of trade. Let the market determine the 
        number of registrars - not IAHC.

In order to ensure equitable international participation among 
Registrars, IAHC has determined that a fixed number of  Registrars 
will be initially allocated equally to each of the six (6) ITU 
designated geographical world zones (X.121 world zones 2-7); the 
current list being available at 
<http://www.itu.int/intreg/itu-zones.html>.

        These zones do not reflect an equal distribution based upon 
        the current Internet population. This is inherently unfair 
        since more populated domains such as zone 2 and 3 will be 
        under represented on a per capita basis.

The selection of registries in a region will be by lottery among
qualified applicants. Registrar rights are non-transferable to other
entities.

        Why does IAHC limit the choice of registries by region and 
        limit the number of registries by lottery? Why does IAHC 
        not let market forces determine this? 

IAHC will establish the qualifications required of each applicant to
become a registrar. These qualifications will be objective and will be
subject to independent confirmation.

        What are these qualifications? Much more space is devoted in 
        the proposal to the lottery, quantity of gTLDS, geographic
        distribution, and other market barriers rather than the most
        important question that is "What constitutes a qualified
        registry?" More importantly - rather than IAHC 
        concentrating on erecting market barriers (which the 
        current proposal has many), IAHC should determine the 
        criteria of a qualified registry. IAHC should develop a 
        process for qualification, and let the market determine how
        many and what distribution of registrars, registries, and 
        gTLDs exists.

 
5. Applicants must commit to sharing all gTLDs.

        Registrars might have many legitimate reasons for not
        supporting all gTLDs.
 
IAHC delegates stewardship for the set of gTLDs to the Council of
Registrars (CORE), comprising the multiple, competing gTLD registrars.
All gTLDs are shared among all member registrars.

        Does that make CORE the steward referred to by the following? :

Oversight responsibility for a registry rests with a STEWARD.

-----

Registrars for existing gTLDs are encouraged to join CORE and abide by
CORE principles and rules, unifying the handling of all gTLDs and
obtaining equal benefits with other gTLD registries.

        Registrars are not required but only encouraged to join CORE?

CORE will contract with an independent and neutral third-party for
operation of the shared gTLD repository data base (gDB). The specific
details of gDB subcontracting and operation will be determined by CORE.

        Who can apply to be subcontractors?

IAHC will specify the information to be required in all applications 
for SLDs under gTLDs. 

        This is  clearly outside the IAHC charter. The reason this 
        must now be inclusive in the IAHC charter is because all 
        power to determine what is required from SLD applicants has 
        been moved from the registries up the hierarchy to IAHC. 
        This is a good example of the types of problems that exist 
        when trying to implement a shared gTLD. There are probably 
        lots of others. IAHC members will find themselves in an 
        International Not So Ad Hoc Committee if they begin to take
        up SLD issues.
 
To promote accountability, discourage extortion and minimize obsolete
entries, SLD assignments must be renewed annually.

        Again, this is outside if the IAHC charter. This should be up 
        to the registries to determine - but since the IAHC clearly 
        intends these registrars to share the registry - then they 
        have effectively removed this discretion and control from 
        the registries. In a true free market system, each registry 
        would run their domain as they please. The market would 
        determine if their policies make sense. I think the market 
        is wiser than any committee.


There is no single, universal international law of trademark, so it is
not possible to reserve disputes involving trademark and domain names
to an international body applying a globally recognized body of law.

        Exactly - so why has the IAHC devoted a large section of the
        proposal to the trademark concerns for SLDs? Let each 
        registry run their domain as they see fit. If they have a 
        policy that creates lots of problems - they will not stay in 
        business very long. Again - the market should determine this 
        - not IAHC.

One gTLD registrar has attempted to address this problem by inserting
itself as an arbiter of disputes between trademark owners and SLD
holders: The registrar will put an SLD on hold at the behest of the
owner of a trademark registration certificate if the holder of an
"identical" SLD, once challenged, cannot produce its own, trumping
trademark certificate or otherwise establish that its use of the 
domain predates either the effective date or first use date of 
trademark registration. This well-intentioned policy, which has 
generated significant controversy, unjustifiably confers upon a 
non-judicial body the discretion to essentially grant an injunction 
against continued use of a SLD, without any adjudication of the merits 
of the trademark owner's claim against the domain holder. Such an 
approach is inconsistent with basic tenets of trademark law and 
principles of equity and fair play. The dispute policy unfairly 
burdens the domain holder - who may actually have trademark rights
superior to those of the challenging trademark registrant.

        This describes NSI policy. In my opinion - this is bad policy.
        In an open market - most businesses would simply take their
        business elsewhere. Right now that is not possible, 
        since there are no competing registries.

In light of the legitimate interests of domain name holders and
trademark owners, and in the overall interests of consistency and fair
play, IAHC strongly believes all gTLD registries and ISO country code
registries should, therefore, publish applications for SLDs, for a
period of sixty (60) days prior to assigning the requested SLD to the
applicant. Such publication should take place on a publicly available,
publicized web site and include the SLD and the contact and use
information contained in the application (see Appendix A).

However, in view of the fact that an existing gTLD registrar currently
registers SLDs without a waiting period, the requirement that gTLD
registries institute a 60 day publication period will not be effective
until that registrar changes its policy to include the 60 day
publication period. In the interim, other gTLD registries are
encouraged, but not required, to use this 60 day waiting period. Once
that registrar agrees to the 60 day publication period, CORE will
implement the 60 day publication period immediately across the other
gTLD registries.

        In my opinion this is a bad policy. This policy is      
        somewhat like those of US gun control laws, that is to 
        provide sufficient time for a background check. Some US States 
        have even gone to computerized background check systems to 
        REDUCE this wait time. Publication for a 60 day period will 
        be great for intellectual property lawyers looking for 
        business but at too large a cost to the Internet community. 
        Leave it up to the registry that is running a particular 
        domain (which of course assumes NO SHARED REGISTRIES) to 
        determine that policy. I would predict that the market would 
        reject ANY waiting period.

To ensure immediate availability of some domain names, it is 
recommended that gTLD registries offer "random alphanumeric" domain 
names which require no waiting period.

        I would argue that the free market would also reject this
        policy. I predict, if implemented - it  would not be used - 
        unless the need was artificially created by an artificial
        waiting period. I would rather just use my IP address, and 
        I suspect others would too.

In further considering the existing operation of the ISO 3166 second
level domain spaces, IAHC recommends that the administrators of these
spaces add a fixed 60 day notification period between application and
delegation as a means of reducing the subsequent levels of litigation
within these name spaces, as per the related recommendations in 
section 4 of this document relating to gTLDs.

        I believe this to be a bad recommendation due to the reasons
        given above.

IAHC notes that the implication of the recommendation in section 5.4.4
is that there will be a consequent market demand for domain names 
which are available within a more rapid period than 60 days. IAHC 
recommends that the administrators of ISO3166 name spaces consider 
the addition of a functional name space, proposed to be a network 
user identifer: .nui.<iso3166-code>" . The characteristic of this 
space is that it would be filled by randomly generated, meaningless 
alphanumeric strings, available within a short period after 
application, in order to provide an alternative mechanism to
the process of obtaining a specified name which may include a 60 day
notification period.
        
        What a complicated web we weave to accommodate that 60 day 
        waiting period. We are now forced to introduce an artificial    
        name space as well as random artificial names to accommodate
        this artificial 60 day waiting period.

 SECURITY CONSIDERATIONS

The Domain Name Service is essential to the smooth operation of the
Internet. Changes to the structure or style of DNS operation therefore
carry considerable risk. The specification offered here attempts to 
make substantial changes to the administration and operation of 
gTLD names, but limits the scale of those changes during an initial, 
"experimental" period, permitting expansion of the changes as 
experience dictates.

        What does this have to do with security?

ACKNOWLEDGMENTS

The IAHC's efforts have greatly benefited from extensive on-line
discussion, both within the committee and on public discussion lists.
Also, the Committee's open request for proposals resulted in numerous
submissions from which some concepts and details in IAHC's proposal 
were adapted. The current draft specification is the result of 
integrating the discussions and proposals, seeking a fair and 
practical balance.

        Please list the acknowledgments. It is good form to list all    
        participants.

REFERENCES

[Post94] Postel, J. , Domain Name System Structure and Delegation, RFC
1591, March 1994.

        There are clearly more references than this.


-- 
Alan Sullivan
President
Top Domain Registry Inc.