[discuss] Options for root zone

[Andrew Sullivan]

On Sat, Jan 18, 2014 at 06:48:37PM +0100, Louis Pouzin (well) wrote:
> Another case of crooked interference is the bulgarian cyrillic ccTLD, *бг*,
> chosen by the bulgarian gov and persistently rejected by ICANN.

Yes.  Because it looks like "br".  It looks _remarkably_ like "br" in
the font I happen to be using on my screen at this very moment.  And
it's worth observing that the so-called internationalized country
codes, quite _unlike_ the ISO3166 codes, are not standardized, which
is precisely why "the Bulgarian government" is choosing the string
rather than some completely disinterested 3d party like ISO.

Indeed, this kind of mess is exactly why Postel so cleverly threw that
hot potato into the ISO country code list.  It got him/IANA out of
having to decide who was a country and what the TLD for it should be.
One "uk" was, I suspect, enough (I wasn't around).

One of the problems that arises in internationalzation of a global
resource, no matter how much Daniel and others like it, is that there
will be troublesome corners where different cases bump into each
other.  Things that are obviously not a problem in one language are
potentially a serious problem when every language is taken into
account -- not to mention every script, which is the other problem
here.  Worse, of course, not everyone using some of the scripts on the
Internet (e.g. Latin) actually knows how to use that script, becuase
their own language may be written in some other script.

I think the ICANN procedures for visual similarity resolution are very
far from perfect.  But we must not conflate that and the IDN issues
with "country code TLDs" in the traditional sense.  The latter are
completely determined by an external standard, and there's never been
any evidence of fooling around with that.  Indeed, ICANN has been
remarkably patient in some cases, like with .su.

It is a nifty rhetorical trick to pretend that the internationalized
country TLDs have the same external foundation as the traditional
ccTLDs do.  And ICANN has been rather careless in the way it has
referred to these things, which make the problem quite a lot worse.
But in fact, these two different kinds of labels are based on
different policies.  Because one of them derives from an external
standard, there must be no variance; this is why all two-character
ASCII labels are reserved.  But the other derives from a country's
preferences, and those preferences are not the only consideration that
should be taken into account for the safe and stable operation of the
root zone.

Best regards,

Andrew

-- 
Andrew Sullivan
ajs at anvilwalrusden.com