Re: [alac] New gTLDs analysis -- IDN

[Thomas Roessler <roessler-mobile@xxxxxxxxxxxxxxxxxx>]

On 2003-05-03 16:13:53 +0800, Dr Xue Hong wrote:

> Is it purporting to serve those non-English speaking users, for
> current TLDs are inconvenient to them, or even impossible to be
> used by them?  Or, is it simply to create a new market for those
> current *stakeholders*?

The point I'm trying to make is that the IDN and latin TLD markets
alike should be open to any party which is: 

      (1) technically capable -- that's what Ross Rader's proposed
      accreditation scheme is about.

      (2) not inappropriately harming others in the market:
            - through confusing similarity.
            - anything else?

Everything else should be left to market forces to sort out, and
should not be subject to regulation.

> If the mission of the "At Large" is to let those users who have
> no other channel to participate the ICANN process to get in and
> get heard, then the (interim) ALAC should, of course, speak for
> those inarticulate non-English users. No need to worry for those
> commercial companies, for they have numerious channels and
> connections to the ICANN.

I'm not worried for the commercial companies: I'm just a strong
believer in market principles yielding optimal results for
consumers, and I'm opposed to anything which amounts to unnecessary
regulation of this market.  I'm even more opposed to anything which
moves the evaluation of a TLD proposal (i.e., the *application* of
criteria) into the sphere of political considerations.

> P.S. I attach a small part of the "Table for Chinese characters"
> to show why the LgTLD evaluation process shouldn't basically be
> applicable. Without the participation of the language community,
> do you think the table can be effectively used for "no harm"
> evalution?

Unless I'm mistaken, what that table is doing is to define
equivalence classes of Unicode code points, in the Character
Variants column.  It´s then simple to define an equivalence relation
not just for characters, but for strings (just map every character
class to the lowest code point, and compare these "normal forms").
Any two strings which are equivalent according to this relation
would have to be considered confusingly similar.  More generally, as
long as you can map possible TLD strings to a unique normal form
according to rules which come out of the language communities, you
can even cast the confusing similarity test in code.

But maybe I'm missing some other aspects a "no harm" evaluation of
an IDN TLD identifier should take into account...  If so, I'd very
much appreciate if you'd tell me about these.

-- 
Thomas Roessler                 <roessler-mobile@xxxxxxxxxxxxxxxxxx>