Re: [gnso-vi-feb10] Niche TLDs

[Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>]

On 7/1/10 2:45 PM, Roberto Gaetano wrote:
> The theme is the following:
> Under which circumstances would people feel safe in allowing vertical
> integration for a TLD that serves a small community, like a cultural,
> linguistic, geographic community?
While feelings are relevant to the members of the VI PDP WG, the 
feelings of the members are not equivalent to the informed judgment of 
applicants for linguistic and cultural TLDs.
Is self-sales better than finding a registrar? In theory, yes.

In practice, in 2001, yes.

In practice, since 2001, no.

Will it be an insurmountable obstacle in the approaching (and receding 
at an equal rate) new gTLD round?
The cost of forming a registrar is an order of magnitude less than the 
application fee, all other costs ignored.
In the abstract then, we are looking for one entity capitalized at 
$185k plus some operating expenses (see the rest of the DAG for just 
how big that is) and one or more independent entities capitalized at 
less than $10k, and possibly using any one of a number of 
registrar-in-a-can solutions.
It is probably worth exploring if any of the shell registrar herders, 
who in steady state simply have to make annual renewal fees plus a 
margin, and exist simply to provide threads to the .com drop pool, 
want to "lease threads" to would-be registrars for a fraction of the 
cost of accreditation. It may be that when, or if, applicants begin to 
seek registrars, ICANN accredited registrar status may be offered on 
very attractive terms (isn't the drop game diminishing anyway?).
So, the net-net of this is we're considering a rule change to prevent 
community-based applicants from having to find parties within their, 
or other, communities, with the business sense to become registrars 
for peanuts, and obtain a near monopoly on the retail sales of one or 
more inventories, because of a necessity claim.
But that's not all...

> Let me start.
>
>     * There should be declared criteria to delimit the community, and
>       adherence to these criteria will be audited by ICANN.
You want to make this harder than scoring a 14 of 16? You expect 
rational business planners to make their plans success conditional on 
scoring more than a 14 of 16??
>     * The registry will be operated as a non profit for the benefit of
>       the community.
That may be how I do things but I don't think it is good policy to 
insist that no community, presently ignored by ICANN and probably more 
underserved than just "ignored", decline every for-profit proposal, 
including those brought institutions within the community itself, in 
the pursuit of non-profit purity.
>     * There is a maximum number of SLDs, passed which the registry
>       loses its "niche" status..
Fine. Both numbers and months have been done previously, in .museum 
and .coop, respectively.
> The point is that if a registry does fulfill these requirements, they
> will be granted an exception, and will be allowed to operate without
> giving equal access to all registrars.
Your first condition, community membership criteria, is an admission 
control criteria, and it is a continuous test. You appear to posit a 
one-time grant conditional upon a continuous, and therefore 
prospective property.

> There might be interesting questions, like:
>
>     * Will they be allowed to use the services of one registrar,
>       selected by them, or not?
The registrants of the Native American serving registry are going to 
use GoDaddy whether I tell them to or not. There will be preference, 
for a variety of reasons, e.g., there being only two registrars in the 
entire MENA market (Morocco to Iraq). Not many more in all of Latin 
America. Three in all of Africa...
>     * What is the maximum number opf names under the TLD?
How about 5k, with a firm recommendation that each of .aero, .coop and 
.museum, even .cat, get this grant of aid. We could see the use of the 
exception in action while waiting for 2011 or 2012 or ...
However good this feels, it doesn't reduce the greater costs. Back in 
Cairo, CORE met with ICANN to discuss the impact of the recurring 
costs, which ICANN subsequently reduced significantly. What we've yet 
to address is the cost of performance, now at SLA levels higher than 
the COM/NET/NAME operator is held to, which is a non-trivial cost to 
the entrant operator.
Further, this proposal ties the applicant, eventual RO, and the RSP, 
as a single unified, or paired entity. If, after building out the 
infrastructure sufficient to meet the requirements in section 6 of the 
DAG, the community-based registry sees offering RSP services to other 
applicants, the "non-profit" only and hard limit per RO constrain this 
grant.
At some point the real issue could be addressed more by the Board 
constitued WG Avri and Evan chair, though that group is constrained to 
just the fee issue, and not the costs implicit in non-fee conditions.
As an applicant, my costing would be far more benefited by removing 
the absurd SLA statements, the ipv6 requirement, the DNSSEC 
requirement, ... than the requirement to offer terms to one or more 
registrars.
Eric