[IDN-WG] DRAFT: ICANN AtLarge position on new gTLDs

[Lutz Donnerhacke]

The following text is the result of an intensive discussion at
the ICANN Studienkreis meeting in Helsinki, Finland, 5/9/2008

Please discuss it in the content in the approbriate mailing list.
I'd recommend gtld-wg at atlarge-lists.icann.org as the right place.

Please note that I will not take part in this discussion, due to my upcoming
holidays. I feel, it might be better to stay away from commenting the
comments for a few weeks.

******************************************************************************
		 DRAFT: ICANN AtLarge position on new gTLDs

		   gtld-community-reception-draft-Sep2008
			  Lutz Donnerhacke, EURALO


Abstract
~~~~~~~~

The community clearly welcomes the new gTLDs especially the IDN version,
because it will drop the entrance level by providing a more concise view.
The proposed gTLD process is focused on commerical interests only, which
causes substantial fears in the community. But the IDN fasttrack proposal
for ccTLD is explicitly supported.

Futhermore this draft tries to start discussion about medium term importance
of domain names. Based on the current community experience, we expect the
DNS to be faced out in favor of overlay networks which obtain their names
from the dynamically shaped digital neighbourhoods.

Community expectation in new gTLDs
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Internationalization
 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

Most people in the world use languages with non-latin glyphes. If they have
a computer, they use a localized user interface. But if they try to use the
internet, they are faced with addresses they can't read, type, and spell.

The introduction of IDNs for TLDs solves this issue. From the user point of
view, it's initally enough the have a TLD in their own language and script
available. Starting from such consistent user interface, the whole web can
be explored.

Using the fasttrack proposal ccTLD registries are allowed to register common
names for regions in their regional used language. AtLarge clearly considers
multiple names using multiple languages for multiple regions of a single
state as the right solution.

There are still open issues regarding which names should be allowed and who
can object a naming proposal. We do not see a consistent way to assign a
name of an international region to a single ccTLD registry, i.e. the Korean
word "Korea" to North- or South-Korea. Another open problem is to assign
political controverse regions, i.e. the arabic names of Palestinean regions
can't be assigned to the ccTLD registry of Israel. There is a need for
really applicable disput resolution policy even for already existing ccTLDs,
i.e. the historic TLD .eh for the western Sahara which will be used by a
regional community.

Generic top level domain names
 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

The current domain name system requries each community has to choose a
parent domain name to become visible in the network. Usually the parent
domain is the ccTLD of the state, the users of the community live in. For
several communities, that's unacceptable, mainly for political reasons.
Choosing a gTLD might be as difficult due to internationalizing effects.

Using an own TLD for such a community may solve the serious problems and is
the prefered option of AtLarge. There is a need to define strict rules for
such a community TLD, including the immanent conflicts with existing TLDs as
well as deal with objection resolution. Communities of suppressed minorities
should be protected from any veto of there suppressors.

The current proposal for opening the root zone for various new gTLDs deals
mainly with the commercial interests. Of course, the community has the right
to object any proposal and can avoid an auction when applying for own TLDs.
We do not consider this rights to be sufficient, because any application for
TLDs will be prohibited by the high fees for ICANN and consulting companies
in the case of dispute resolutions. That's why We consider the current
propsal as an preference of commericial interests over community needs.


What will be a "name resolution" in next century?
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Up to the late 1980er years data flows using bang addresses naming a hop by
hop route through the Internet, so names does have a local scope. Relicts of
this time are still used in Usenet News, last mile e-mail delivery
(mailertables, forwarding, polling), and remote machine access through
private networks or VPNs.

With the increased usage of FTP, SMTP, and HTTP, the data flow changed to
direct end-to-end delivery. To find the address of the destination, DNS
becomes the dominant lookup mechanism. This delivery method created the
current, common understanding of the Internet. The prominent sites of this
time, like funet.fi, are remembered by the people even if the machines are
vanished since years.

In the last ten years, content overtake the role of DNS. Users are entering
everything they need into a search engine and select from the results. So
the DNS names are pushed into the underground, they are used like IP
addresses. Even complex names does not stop people from visiting the
content. Webbrowsers became a part of the IP stack in the common user
experience: They simply click, but use some prominent domains as shortcuts.

Now social networks attract users more and more. User does spend their time
in digital communties. Static content is degraded to the always available
background (only to reference). Names are scoped by those networks, like
names in file sharing networks: They address the community content, not the
machines the content comes from. Name duplication is solved by neighbour
metrics, "local" names are prefered over "regional" and those over "global"
ones. Search engines are used as final fallback.

With the implementation of peer2peer name resolution into the operating
systems, like Bonjour for Apple OSX or the Resolver in Microsoft Vista/2008,
content addresses are available for all applications on the system: If you
open the word processor to write an letter, you will start with the template
from your current digital network: Your default company template for the
home office id as well as the private template for a letter to the bank when
using the private id. Using multiple IDs at the same time will be easy, they
are just a keypress away.

The definition of neighbourhood for name clash resolution becomes dynamic.
Google condensed the current user expectations into their bowser: Users will
start from the recently used stage. So the upcoming name resolver library in
the operating system IP stack will collect usage statistics per digital id
to define a constantly changing neighbourhood. I.e if you travel, your
expected association of the name "hotel" will be depend on your last hotel
searchs as well as your current geolocation. Your expected outcome of "news"
might stay in your home domicile but changes to the geographic region the
longer you stay there.

So the medium term forecast of Internet usage does not need the DNS system
as the primary user interface anymore. DNS names will be used like IP
addresses today, they will move into network drivers of operating systems.

We do not claim that this forecast is correct, but like to start a
discussion on possible medium term scenarios. Depending on this forecast we
are able to allow new gTLDs liberatly or keep it a restriced ressource.

Lutz Donnerhacke
EURALO

******************************************************************************