[council] More IDN policy quicksand

[Cary Karp]

The quotes below are from Bruce's remarks in the real-time captioning of 
the GNSO Public forum. I am not being at all critical of the way 
anything was expressed in that context, but as we proceed with the 
development of formal IDN policies we will need to become very precise 
in the terminology we use.

> at the moment, the IDN guidelines suggest using a single script to
> the left of the dot, so you don't have a mixture of, say, English or
> Latin scripts and Cyrillic scripts in the same name or mixtures of
> different scripts.

The guidelines say not to mix scripts in a single _label_, unless the 
language being represented is normally written using multiple scripts. 
This allows for situations where there may be arguable need for having 
multiple scripts in a domain name even when the intended language is 
normally written using only one of them. (Examples of such cases have 
been provided by representatives of the Unicode Consortium, if memory 
serves, primarily via their own e-mail discussion list.) This 
consideration is entirely separate from the current restriction to the 
exclusive use of Latin script in top-level labels.

> if I have English script, then that connects to an English script
> string in the root. And then if you wish to use Chinese script, then
> that's associated with Chinese script TLD in the root. Or do you
> allow mixes? And we have today, you can have English or Chinese -- or
> you certainly can have today a Chinese script dot English script.

The English _language_ is written using the Latin _script_. A large 
number of additional languages are written using that script. Some of 
these languages are more reliant than others on the ability to indicate 
diacritical marks in addition to the shared repertoire of basic 
characters (with it being a common misconception that English is 
entirely free from such need), but ASCII can comfortably be used for 
quite a few of those languages. Although we do urgently need to get 
beyond the constraints that it imposes, we should keep in mind that 
ASCII is not a synonym for English.

Many of the individual languages that share the Latin script are used in 
more than one country. The same can apply to the different languages 
that are written -- listing only a few of many possible examples -- with 
  the Arabic script (which, unlike the term English, does designate both 
a language and a script; Latin script is frequently called Roman script 
precisely to avoid this type of ambiguity), the Chinese script, the 
Cyrillic script, the Devanagari script, and all the other scripts that 
either are already under consideration or are going to appear in short 
order as we proceed with the internationalization of the namespace.

There is certain to be some discussion of the extent to which countries 
may "own" the scripts used within their borders, but there are 
astonishingly few cases where there is any unique relationship between 
script and country. Corresponding conditions apply to languages.

The first version of the IDN Guidelines was marred by its having been 
written under the assumption that the terms "language" and "script" were 
synonymous, with the preferential use of the former to avoid confusion 
(despite clearly differing opinions about the likely consequences of 
doing so). The 2.0 revision was prepared specifically to rectify the 
difficulties to which this led. I hope that I am not alone in feeling 
that the terminological stringency now reflected in the Guidelines 
should also be conveyed by the material that Council prepares.

/Cary