[gtld-council] The focus of GNSO Council involvement in IDN

[Cary Karp <ck@nic.museum>]

One of the things we've learned about IDN is that few (if any) of the 
problems encountered in its implementation are truly new. There has 
always been potential for confusion resulting from similarities in the 
way characters appear, for example, between 1 l I, and 0 O. Where there 
is need for differentiating between two domain names in semantic terms 
rather than solely by their graphic appearance, there are plenty of 
problems to be had without going beyond the limits of the familiar 
26-letter Latin alphabet.
What IDN does is make all of this worse; in some cases very much worse.
Certain aspects of this situation require urgent and active 
rectification, for example, through a significant reduction (on the 
protocol level) in the number of characters that are available for IDN 
use. Other aspects of the 'problem' are inevitable consequences of 
accommodating of the namespace equitably to the full diversity of human 
languages and writing systems.
Whatever the extent of GNSO Council involvement in this plethora of 
issues ultimately proves to be, the development of policies relevant to 
IDN in the generic namespace is our prime concern. In keeping with what 
is said above, there are no apparent aspects of this are are truly 
unique to IDN. The differentiation between two candidate TLD names 
become increasingly more difficult to assess as the number of 
dictionaries that might need to be consulted increases, and more 
detailed policies may be needed to constrain the deliberate generation 
of exploitable confusion. These are, however, only matters of degree.
To be sure, the range of new considerations quickly places us in 
unfamiliar territory. One aspect of this that currently appears to be 
causing particular concern is the discussion of providing current TLDs 
with alternative IDN designations. This is as much a matter for the 
ccTLD registries as it is for the gTLDs. Although the government of 
Somewheria might very much like to see the ccTLD label .sw augmented by 
the name of the country explicitly represented in the Somewherian 
script, it is highly unlikely that they would be willing to accept the 
delegation of the IDN alternative to be subject to competitive bidding. 
(Unless of course, they are unhappy with the .sw operator and also want 
that put up for rebid. Either way it remains unlikely that they would 
perceive any advantage in the two domains being operated by different 
agencies.)
Hopefully, it is safe to assume that at some point in the near future it 
will be possible for a request to be made for the localized equivalent 
of .somewheria. It would be equally possible for the operator of that 
domain to use the same subdomain tree as is used for .sw, or for the two 
domains to have separate naming hierarchies. Although it might be 
appropriate as a matter of policy to require that an application for an 
iTLD (dusting off the abbreviation initially used for what was to become 
'gTLD') include a description of any intended parallel use of a 
pre-existing name tree, the delegation itself can be made using the 
conventional procedure for adding a new TLD to the DNS root. 
Corresponding conditions pertain to gTLDs, although the issue of 
language nexus would need to be addressed from a different perspective.
The aliasing mechanism currently being considered for technical testing 
permits the direct linking of a new label to a pre-existing TLD. This 
would make it possible to consider the introduction of localized TLD 
labels separately from the creation of new TLDs. It thus provides 
potential means for being able to address situations where localization 
 is a particularly urgent concern, without locking the timetable for 
that process to the one specifically focused on the establishment of new 
TLDs.
We seem, however, to be stuck in a discussion of, (a) whether there is 
any point in proceeding with testing of an aliasing mechanism prior to 
determining that it would be harnessed in any impending policy 
statement, or (b), whether there is any point in developing policies for 
aliasing IDN labels to existing TLDs prior to determining that there is 
a suitable technical mechanism for doing so.
The President's Advisory Committee on IDN, subsequent to intense debate 
on that very point, is now proceeding with the planning of a technical 
test under the assumption that knowledge of its outcome will be useful, 
in any case, in the development of iTLD policies. I believe that it 
would be beneficial if Council would address this matter on the same 
unconditional basis and without delay. I cannot see any objective basis 
on which we could issue a categorical statement against the application 
of an aliasing mechanism, but now is the time to consider and articulate 
our position. Whatever this ends up being, the purposeful structuring of 
the technical test would be furthered by our input.
If the GNSO treats this primarily as a cc concern, we will end up in the 
paradoxical situation where the generic -- and therefore international 
-- segment of the TLD namespace remains fettered to the Latin alphabet, 
while the localized facets of the namespace acquire access to the full 
IDN repertoire. If there is any justification for the frequently 
expressed concerns about IDN fracturing the global namespace, the robust 
internationalization of its generic component is all the more a vital 
and urgent an issue.
/Cary