[discuss] Internationalization and TLDs (was: Options for root zone)

[Andrew Sullivan]

On Sun, Jan 19, 2014 at 04:23:20AM +0100, Louis Pouzin (well) wrote:

> *« 63. Countries should not be involved in decisions regarding another
> country’s country-code Top-Level Domain (ccTLD). Their legitimate
> interests, as expressed and defined by each country, in diverse ways,
> regarding decisions affecting their ccTLDs, need to be respected, upheld
> and addressed via a flexible and improved framework and mechanisms. »*

First, I'm not actually sure that article 63 means what you think it
means.  I can make out two readings.  It could be about the string
used for the TLD label itself; it could, however, be about the
management of the namespace in that ccTLD.  But in any case, article
63 is about one country being involved in decisions about another
country's ccTLD.  I don't see how that's relevant in this case,
because ICANN is not, _mirabile dictu_, a country.

Second,

> It's none of ICANN's business to invent conditions, rules,
> categories, and silly hair spliting about the string defined by a
> country.

this is quite mistaken, because ICANN has responsibility for the
_parent_ side of the ccTLD relationship: the root zone.  The root zone
is a zone too, just like every other zone, and it needs to have a
common behaviour for all its users.  ICANN is responsible for the
co-ordination of that zone, and it's entirely correct that ICANN has
something to say about it.

> Oh yeah, it's very complicated, too much for ICANN and IETF. The job should
> be delegated to China, where experts in complicated scripts are available.

It doesn't matter whether one knows about Chinese, Devanagari, other
Brahmi-derived scripts, Cyrillic, Latin, Arabic, or even Linear-B.
The problem is that it turns out people use writing systems according
to the assumptions they have about their own writing system; and while
in the majority of cases there is no trouble, in the rare cases there
is potential for difficulty.  The root zone is special because every
single person on earth has to use it, and so the handling of it has to
be even more restrictive than for any other zone.  None of this is
news.  For more recent attempts at the problem, please see the
extensive discussion in the ICANN report on this from a few years ago
(http://www.icann.org/en/topics/idn/idn-vip-integrated-issues-final-clean-20feb12-en.pdf)
and the approach that ICANN is taking to handle some of those issues
(http://www.icann.org/en/resources/idn/variant-tlds/draft-lgr-procedure-20mar13-en.pdf).
Full disclosure: I was involved in preparing both of those documents.
If you want ones I haven't been involved with, I can point you to a
series going back to around 2000.  Some of those, I will note, turned
out to have bad ideas in them.  We learn things over time.

> potential sources of confusing similarities. This results from ICANN's
> policy in constraining IDN specifications so that it would not be possible
> to introduce non ascii scripts. The chinese team who developed IDN was
> smart enough to overcome the constraints.

I don't know what you are talking about.  This has absolutely nothing
to do with any ICANN involvement in IDNA specifications (which were
developed in the IETF and not at ICANN).  It has nothing to do with
the nationality of any contributors to IDNA either (it was not, in
fact, a "Chinese team" that developed IDNA either the first or the
second time.  It wasn't even a "Chinese" team that developed the JET
guidelines -- the J stands for "joint", because Han is used for more
than Chinese).  This issue is, simply, an issue with
internationalization of the Internet, which turns out to be rather
hard when it comes to the details.  This is why ICANN has rules for
IDNs in a zone it maintains (the root zone).  Those rules need to take
into account the fact that everyone on earth has to use that zone.

> It is normal for users to view ccTLD's from their national context,

That may be, but they don't get to keep out all the users who happen
not to share that context.  That's the problem.  You seem to have a
positive-path bias; the root zone rules need to account for the
negative path too.

> E.g. MORPHO could be a perfectly regular new gTLD. Can anyone tell if the
> string is ascii or cyrillic ?

No, and that's precisely the problem.

> We need a new linguistically customized internet.

Good.  Go build it.  Nobody is stopping you.  I hope you are successful.

Yours,

Andrew

-- 
Andrew Sullivan
ajs at anvilwalrusden.com