Yet another analysis of the NSI/IANA deal

["Richard J. Sexton" <richard@vrx.net>]

Comments re the NSI/IANA compromise.

A) THERE ARE PROBLEMS WITH "REGISTRARS"

>(ii) The Domain Name Supporting Organization shall be composed of
representatives from name
>    registries and registrars of top-level domains ("TLDs"), businesses and
any entities that are users of the
>    Internet and any other entities with legitimate interests in these
issues, as determined by the Domain
>    Name Supporting Organization consistent with Section 2 of this Article
and approved by the Board. The
>    Domain Name Supporting Organization shall create a Names Council to
make recommendations
>    regarding TLDs, including operation, assignment and management of the
domain name system and other
>    related subjects; and 

1) There does not seem to be a working definition for a "registrar". In the
degenerate
case, any individual that ever registered a domain for another individual is
a registrar.
There is no technical, legal or financial barrier to entry; in theory
everybody on the
Internet can be a "registrar". Arguably, evry ISP on the planet already is.

2) Registries are a necessity. They provide a unique and mandatory technical
function. Registrars, on the other hand are a service function; there is no
technical requirement for them. It's similar to a car dealership that may
license the car for a purchaser as a matter of convenience, but, anybody 
can go to the license office and register it themselves.

3) the organization is subject to capture by a large group of registrars. A
registry
may "proclaim" any number of registrars; currently there are 230 or so TLD
registries
in the legacy root. If they "created" say 5000 registrars the voice of
registries is
obliterated; this organization is subject to capture.

SUGGESTION: remove "registrars" from this section of the by-laws.

B) COMPOSITION OF THE BOARD

It's unclear how the initial board will be selected. It appears the board
is made up of people appointed by the SO's. But, the SO's are selected by
the board, so there is an apparent bootstrapping problem here.

SUGGESTION: Clarify how the initial board is selected. 


C) INTERNATIONAL REPRESENTATION

Either the ICANN is responsive, and accountable, to the Internet community
or it is not.
If it is, since the Internet community knows no geopolitical boundary, the
notion of
International representation is specious. If the ICANN is responsive, it is
equally
responsive to somebody in Washington DC or somebody is Papua New Guinea. Do
we want the
best possible representation on the board, or do we want a politically
correct board ?

The aggregation of Canada, the US, Mexico (which are all part of *North*
America),
Central America and South America into "the Americas region" would permit
all candidates
to be from the US. This is not expected to be a popular decision with the
Canadian
or other governments. If there are three people from the Americas, select
one each 
from Canada, US and Central/South America. Better yet, ignore the International
Representation requirement

It is easy to make light of this: "I want the best possible board; if I
can't have that
I want a Canadian on the board". 

Additionally, the idea of excluding government people does not mesh well
with, say
the Asia pacific region where it is common for line between government and
business
to be fuzzy at best, ie. Singapore Inc.

SUGGESTION: drop the international representation as a requirement, and make
it a goal.
If two candidates are equal, then select the most geographically diverse
one, but do
not overlook a superior candidate because they come from the same region as
other
board members.


D) IS THERE A TIMETABLE?

The green paper called for 5 new TLDs. The white paper left decision that up
to the new organization
and now, what might be the new organization, seems, in my opinion, years
away from introducing
any new TLDs into the root. CORE members are being bled $500 a month in
their limbo-holding
pattern and the "other guys" have been waiting patiently with operational
tld servers
and registries for almost two years now. Would it be possible to let each
group that
can demonstrate it has been "ready" for, say a year or more, to have one TLD
admitted into
the legacy root? This would also prevent the "shell shock" of having a large
number of
TLDs introduced at one time.

SUGGESTION: Identify some sort of time table; possibly deploy one TLD per
recognizable
entity immediately.
--
Japan, like other countries, had many people trying to
legislate networks in existence while a few people rolled
up their sleeves and installed cable".  - Carl Malamud