[gnso-vi-feb10] Some random notes

[Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>]

Some random notes, for which I blame Ron for suggesting that we should
think.

i) For applicants which have no equity to structure, in registry
service providers or registrars, either because they are similar to
the original sTLD applicants (SITA's .aero, NCBA's .coop, and
MuseDoma's .museum), or similar to PuntCat's .cat, or are municipal or
other public authorities, this issue is relevant only in so far as it
(a) delays the acceptances of their applications, and (b) alters the
location of marginal profit for registry-registrar combines and
reduces the incentive for registrars committed in such combines to
pursue lower marginal profit inventories.

Of these, (a) is clearly a greater concern than (b), as the
registry-registrar combines are unlikely to consume the entire
inventory of non-shell registrars, and the creation cost of registrars
is under five figures, so non-shell registrars are (like brands) an
inexhaustible resource.

With very few exceptions, all the disclosed linguistic and cultural,
municipal, geographic, and treaty organization applications known to
me are in this situation. To use rough numbers, that is approximately
100 projects in pre-application-wait-state, of which fewer than 10 are
even potentially private equity-based and capable, in theory, of
seeking a registry-registrar combine as an incidental element of their
policy and operations.

ii) For applicants which have equity to structure, in registry service
providers or registrars, (a) until we consider the currently
contractually protected status exclusive dealings between the registry
operator and the registry service provider, this is a non-issue, and
(b) applicants formed by registrars for the purpose of forming a
registry-registrar combines are distinguishable from applicants formed
by non-registrars which seek to acquire equity (or control) in a
registrar to effect a registry-registrar combine.

Brand holder goals appear to view the formation of registry-registrar
combines as secondary to the purpose of control over the distribution
of the associated inventory.

Registrar goals appear to view the formation of registry-registrar
combines as primary to the purpose of extracting the maximum profit
from the distribution of the associated inventory.

iii) Unlike the circa-2000 collection of actors, there appears to be
an insufficient interest in the pooling of resources sufficient to
meet one or more six or seven figure capitalization goals, by
registrars, for the purpose of competing with the existing Latin
Script market, or competing with the com/net brand, for core market
share. There is no evidence that any broad competition policy issue
can arise from the presence, or absence, of registry-registrar
combines, whether formed for the purpose of gaming micro-inventories,
or formed for the purpose of rational distribution of micro-inventories.

CORE's recommendation that community-based registries, and perhaps
even standard registries, pool resources to form a shared registrar is
motivated more by a concern for the recurring cost of different rights
of others policies and supporting mechanisms, and similar long-term
recurring costs, than the frequently argued need to find or form a
registrar for a registry lacking a registrar. CORE views cooperation
to be in the interests of the cooperating parties.

Some characterize this as being in CORE's interests, which is true if
and only if the applicants use CORE as a cooperative service provider,
but it is as true that simply getting community-based registries, and
standard registries in the IANA root is in CORE's interest.

iv) Elsewhere, in the IETF DNSEXT WG list, and to some VI WG
participants, I've pointed out that the development of a CLONE
resource record will allow cooperating brand registries to share
infrastructure, should their applications ever proceed to contract and
delegation.

v) What percentage, what quota, what criteria are available to
allocate to brand holders? The percentage has been, and remains zero,
the quota is zero, and the criteria is don't even ask. Whether brand
holders seek to form brand-registrar combines or not.

vi) What percentage, what quota, what criteria are available to
allocate to registrars? The answer was unlimited, though in practice
limited to the existing 900+ registrar (shell included) inventory, and
probably to fewer than 100, and at Nairobi became the same as for
brand holders. Zero. Whether the registrars seek to form
registry-registrar combines, or merely hold minority investments in
the associated registry vehicles.

vii) Subscription based service abounds under the .com brand. All are,
at present, restricted to Latin Script as their primary brand. In the
near future this restriction will be eliminated.

Independent of the choice of Script, distribution of domains through
existing subscription based services will continue, under second level
brands, and applications will be made for the same model using first
level brands. CRM functions for subscription based services are
routinely outsourced, yet rarely use an equal access competitive
subscriber selection model. A structural constraint on the ICANN
contracted parties is that second level domains are distrubuted under
an equal access competitive registrant selection model.

An interface between walled gardens and registrars is lacking, this is
our industry's own private bit of the network neutrality struggle, if
our application is worth having, it will be replicated within the
walled garden, or the gardener will create an opening in the wall.

viii) One or more allocation of benefit among cooperating registrars
models have existed in the secondary market for most of the past
decade, as eNom has pointed out, it controls 135 registrar
accreditations for the purposes of goal maximization in the secondary
market, and competitors to eNom use revenue sharing mechanisms to
benefit independent registrars to participate in competing goal
maximization systems in the secondary market.

The primary arguments offered for "single registrar" in the context of
"single registrant" are cost and complexity.

ICANN has experience with price capped registry contracts. ICANN does
not have experience with price capped, registry specific, registrar
contract addenda. Nor does ICANN have experience with registries
placing more than nominal access control restrictions on registrars.

If the brand, and also the subscriber control and issues are to be
addressed without wholesale exception to Recommendation 19, then the
control mechanisms, and the allocation of benefit to registrars needs
to be addressed.

At present we have a system in which benefits for a registration
function in the secondary market are shared across pooled registrars,
shell and non-shell, the choice of which form is a secondary market
actor implementation choice.

Related to the walled-garden and competitive registrar item above,
we've the opportunity to consider how to virtualize a single registrar
out of the existing registrars, less the shell registrars, so that no
registrar which seeks to enter the brand, or subscriber registry's
registrar market is equitably benefited without creating the control
problem.

I'm stopping here because I've got other stuff to write, for someone
who really wants a gTLD, and is utterly unconcerned with vertical
integration.

Eric