Proposal for IDN Label Length Requirement

[Wil Tan <wil@xxxxxxxxxx>]

In Section 2.1.1.3.2, String Requirements, Policy Requirements for Generic
Top-Level Domains:

  "Applied-for strings must be composed of three or more visually distinct
letters or characters in the script, as appropriate"

As many have raised before, this is an unnecessarily restrictive rule for
IDN labels that will harm the CJK community if it is not resolved in a
culturally sensitive manner.

In its analysis of the gTLD guidebook (1st draft) public comments, ICANN has
provided a well-grounded explanation for the requirement and committed to
working with the community to produce a solution. I'm glad to see that ICANN
staffs have made good on their promise and have been working with the IDN
community throughout and after the Mexico City meeting.

I believe that the consultation work is still going on, and would like to
propose that the requirement be lifted from string whose writing systems
employ basic building blocks that have generally accepted semantic
associations. In such writing systems, single and two-character sequences
generally represent concepts in their own right, without the need for
abbreviation. Restricting the strings in these writing systems to three
characters or more is akin to forcing someone to pick ".canine" as a gTLD
string because ".dog" is too short, if not worse.

Incidentally, these writing systems generally do not remotely resemble Latin
so visual confusability will not be an issue (there's string review for
that.) The character repertoire for these scripts is also orders of
magnitude larger than that of alphabetic or syllabic scripts. To put it in
perspective, there are 71,442 Han characters in Unicode version 3.2 versus
26 English alphabets.

Wil Tan