Selection of TLD Names/Policies and Registry Operators

Approval of TLD names and the policies associated with TLD names,
and approval of TLD registry operators to run TLDs, are two very
different problem domains  that should not be mixed. 

The first question  requires answering the question "would the public
interest be served by delegating tld X with policy Y?" The second
question requires answering the question "does registry operator X
exhibit all the technical and business qualifications required to run a
TLD registry?"

The second question is moderately complex, but straightforward.  The
technical and business qualifications for a registry operator are not
great; the network and operations infrastructure required are modest 
when considered against the scale of a modern corporate data center; 
and there are literally thousands of companies that are qualified to 
run a registry.

The first question, however, is much more complicated, and requires more
discussion.  Before getting into that somewhat complex discussion, let
me state the conclusion of this document:

    ICANN should separate the issue of selection and accreditation of 
    registry operators from the issue of selection of TLD names and 
    policies.  There should be two separate solicitations of interest 
    from ICANN -- one for technical entities interested in being 
    registry operators, and another one from entities proposing TLD 
    names and policies.  The same entity may of course enter both 
    tracks, but approval of one should not be coupled with approval of 
    the other.  TLD names/policies should be independently approved, 
    and only after they have been approved should solicitation of 
    registry operators for the TLD commence.


Now, back to the discussion of how to select TLD names and policies:


            The "Aggressively Libertarian" Approach

Some actually argue that selection of TLD names and policies is *not* a
matter of interest to the general public at all, and that instead,
registry operators should be allowed to pick TLD names on a
first-come-first-served basis, and that they should be able to set
policies for TLDs without restriction.  Any other approach, it is
argued, would open ICANN up to anti-trust action. 

This has an appealing simplicity, but from a political point of view is
completely unrealistic, and it would be irresponsible of ICANN to follow
this course.  First, the existence or non existence of certain TLD
names/policies is heavily controversial.  An obvious tiresome example is
the .sex or .xxx TLDs, devoted to pornography.  But human creativity in
coming up with offensive names cannot be underestimated, and it cannot
be anticipated.  Second, and more important, perhaps, is the fact that
the technical requirements for being a TLD registry are really not all
that great.  There are currently many many entities with the technical
infrastructure required -- corporations, universities, ISPs, and so on. 
If registry operators are allowed to pick their own names and policies
there is a very real possibility that there will be a gold rush into the
TLD name space that parallels the .com gold rush.  That is, we move the
problems of .com to the root. 

Not only is this aggressively libertarian approach politically
unrealistic, it is, interestingly enough, a total contradiction and
refutation of all the democratic principles that people argue so
strongly for.  It takes management of the TLD space away from the
Internet community, and completely turns it over to the registry
operator community.  That is, we insist on bottom-up, transparent DNS
govenance -- except when it comes to the most significant component of 
DNS governance, selection of TLD strings and policies.  We might just 
as well do away with ICANN entirely, turn DNS over to a consortium 
of registries, and completely eliminate any possibility of regulation 
by the Internet community.


              The "Community Policy" Approach

ICANN is pioneering what might be called a "Community Policy" approach
to policy determination.  I use this term to indicate policy developed 
by a broad public, but without the connation of government action that 
is implicit in the term "Public Policy".

This approach seeks to gain input from the widest possible cross-section
of the community, and to come to a consensus on a policy.  Sometimes
consensus cannot be achieved, but a strong attempt is made, and if
action must be taken, the attempt is to cover the largest position
possible, consistent with actual creation of the required policy.

In the context of TLD strings and policies, this means that ICANN must
publicize individual TLD proposals, seek wide public comment, and
synthesize a response based on that community input.

In the context of DNS policy, the arm of ICANN that has the broadest 
input is the DNSO.  Therefore, it should be charged with collecting the 
necessary community input into all TLD name/policy decisions.  With 
this in mind, I propose in outline the following process:

  Those interested in the TLD would petition the NC to form a WG of the
  DNSO to work out the details.  Creation of the WG would be publicized,
  and critics and supporters alike would join.  It would be the job of
  that WG to come up with a concrete proposal, to work out the details
  of the policy including a charter, if appropriate, and a sponsor or
  sponsors, if appropriate.  It would be the job of the WG to address
  the concerns of critics, and to finally produce either 1) a very
  concrete proposal to be sent to the NC, and then to the Board; or 2) 
  in the case of controversial TLDs, a report on the controversy, with 
  arguments from the various sides.

The overhead of any community comment process will encourage the
measured introduction TLDs, and wide publicity will help guard against a
faux paus such as the introduction of a name that is offensive in
another language.

Of course, we have been discussing new TLDs for 5 years, and some names
have been in the public awareness for quite some time.  In the initial
short term it may be useful for ICANN to select a few of these names for
the first few TLDs.  But in the long term selection of TLD names should
not be left to the ICANN staff -- TLD names and policies are a minefield
of potential conflicts, and those conflicts should be carefully explored
in public view.


                            Summary 

Selection of registry operators is a much different problem than
selection of TLD names and policies, and should be separate activities
within ICANN.  TLD names and policies are very complex and vary with
each case, and therefore need to be widely discussed and publicized
before any action is taken.  But once registry operator standards are
devised, they can be applied as an administrative procedure.