[wg-c] Comments on the 8 principles

[Kent Crispin <kent@songbird.com>]

In general, I think the idea of principles like this is potentially 
useful, but that the particular implementation is inadequate.

>Based on new input received here follows amendment 6 to the
>principles (which now number eight).
>-------------------------------------------------------------
>Criteria for assessing a gTLD registry operator application,

"registry operator" is a technical function.  A "registry operator
application" should only be evaluated on the ability of the registry to
provide the appropriate service, namely, support registrations for
domains.  Any registry operator that can provide service for one domain
can provide service for multiple domains at a very tiny incremental
cost, so linking TLD name approval to registry operator approval is
pointless.  

Someone with a great idea for a name would simply contract the registry
operation services to a qualified registry operator -- it would just be
stupid to require that everybody with an idea for a new TLD must come up
with a unique new chunk of technical infrastructure to support it.  

Therefore, registry operators should be approved completely
independently from TLDs.  Approval of a TLD name has no technical
component -- it's essentially all policy; approval of a registry 
operator is fundamentally on technical, stability, and business 
viability grounds, and should be as objective and policy independent as 
possible.

>subject to
>current technical constraints and evolving technical opportunities, should
>be based on all the following principles :
>
>Principles affecting the relationship between a gTLD Registry operator,
>Registrars and
>those who may register in the domain
>1. Certainty: a gTLD should give the net user confidence that it stands for
>what it purports to stand for.

In practice, a principle like the above can only cause people to argue
about whether a string gives confidence to the average net user, or what
something "purports to stand for".  Neither of these arguments have any
possible definitive resolution.  Therefore, this principle is in
practice useless.  This principle is a glittering generality.  Ditch it. 

>2. Honesty – a gTLD should not unnecessarily increase opportunities for
>malicious or criminal elements who wish to defraud net users.

Of course, of course.  But this is simply another glittering generality
that will send people down endless ratholes arguing about what
"unnecessarily" means.

>Principles effecting the relationship between Registries (and with relevance
>to Registrars)
>3. Differentiation – the selection of a gTLD string should not confuse net
>users and so gTLDs should be clearly differentiated by the string and/or by
>the marketing and functionality associated with the string.

Confusion to net users is not something I look forward to arguing about,
either.  Moreover, as has been pointed out many times by people as
diverse as Milton Mueller and Dave Crocker and Paul Vixie, the DNS is
not a directory service.  More directly, there is no obvious mapping
from an entities characteristics to the domain name that entity might
use.  Therefore, this principle is largely irrelevant to the actual use
of the DNS. 

>4. Competition – new gTLDs should foster competition in the supply of domain
>names and in the provision of Internet applications and services such that
>the authorization process for new gTLDs should not be used as a means of
>protecting existing service providers from competition.

Sounds good, but the arguments concerning .com2, .com, and .biz don't 
bode well for this principle actually having any use for discriminating 
between TLD applications.

>5. Diversity - new gTLDs should foster the expression of views, both
>commercial and non-commercial.

Sounds good.  Of course, every new gTLD would meet this criteria.

>Principles with query resolution and character encoding implications

Eh?  I don't see the relationship between "semantics" and "query 
resolution and character encoding".  Especially the latter...

>6. Semantics – registry applications for a gTLD should explain what meaning
>will be imputed to the proposed gTLD string and how the new gTLD will be
>perceived by the broad population of net users.

As has been pointed out repeatedly, the meaning or perception of a
string across the broad population of net users is almost useless as a
criteria.  The TLD ".red" would mean something totally different to a
Spanish speaking person than it would to an English speaker.  Even more
telling, there is work underway to expand the character sets supported
by DNS.  This leads to the possibility of TLDs like 

                .<Korean characters>

[I'm told, incidentally, that "naa" in sound would most closely 
represent the Korean word for "myself".]

We tend to forget that the ascii character set used in domain names is
unintelligible to a large fraction -- perhaps a majority -- of the human
race.  Therefore the idea that TLDs should be chosen on the basis of
their meaning to the broad user community is fundamentally flawed.  The 
broad user community has no common view.

>Other principles
>7. Multiplicity - new gTLDs should become available as needed to meet the
>needs of an expanding Internet community.

Great.

>8. Simplicity - adherence of the above principles should not impose an
>overly bureaucratic procedure on a registry.

What would be a meaningful definition of "overly bureaucratic"...

-- 
Kent Crispin                               "Do good, and you'll be
kent@songbird.com                           lonesome." -- Mark Twain