Re: [idn] Chinese Domain Name Consortium (CDNC) Declaration

[John C Klensin <klensin@jck.com>]

Erin,

Let me follow up Vint's note with some thoughts of my own,
with the understanding that they are mine personally and may
not reflect positions I would feel obligated to take as
technical advisor to the IDN WG or as a member of the IAB; the
WG or the IAB might well disagree with what I'm about to say.

I believe it has been fairly well demonstrated by this point
that there are _many_ problems that the protocols specified in
the IDN WG core documents will not solve.  Many of these
problems involve Chinese Domain Names, many others do not.  At
the root of most of both types of problems lies a distinction
between "identifiers" -- character strings used in narrow
contexts and to which many restrictive rules can be applied--
and "user-visible names".  The latter must obey the rules of
the languages in which they are constructed and written.  If
they do not, confusion and astonishment results.

But, while some issues can be addressed with sufficient
ingenuity, the DNS cannot, in general, solve user-visible name
problems within the current design and basic DNS protocols.
Again, this raises especially difficult problems for Chinese
names, but there are problems of one sort or another with most
languages.  The solutions for those user-visible name
problems, as we discussed in Minneapolis, lie in non-DNS
mechanisms, protocols, and database, not in the DNS or the IDN
WG's effort to fit some extra characters into the DNS.

As we look at the IDN WG's efforts, it is important to
remember, first, that Unicode (ISO 10646) is a tool.  It is a
tool that is better suited to some purposes than to others and
it is one that was designed in the context of very complex
tradeoffs.   The WG chose to use that tool, largely because,
given _their_ constraints, there were no plausible
alternatives.

Similarly, we should view the IDN WG's outputs --if, indeed,
IDNA/ NAMEPREP/ STRINGPREP/ PUNYCODE turn out to be the
outputs-- to be tools.  They are not a perfect tool.  It is
easy to find problems they don't solve, and problems with
characters that have equivalent meanings and different
appearances and code points (as in TC<->SC mapping) and
characters that have similar or identical appearances but
belong to different scripts and have different code points (as
in matching characters in Latin, Greek, and Cyrillic scripts)
are high on that list of unsolved problems.  But, again, they
are a tool, and, while it quite drastic, I suggest that there
is a registration-based solution if that tool is
inappropriate.

The IDN WG also has the constraints, ultimately imposed by the
DNS architecture, that it cannot make a decision that solves a
problem for Greek at the expense of Cyrillic, nor can it make
one that solves a Chinese problem at the expense of making
Japanese or Korean unusable.  Perhaps the more subtle matching
and
equivalence problems are sufficient to make the whole notion
unusable, at least for user-visible names.  While I suspect
that might be true, the argument has not been convincingly
made.

This suggests some actions you might consider:

(i) Your colleagues argued quite persuasively in Minneapolis
that the difficulties imposed by the IDN WG approach on
Chinese-Language Domain names would cause serious consumer
confusion and disruption and could be expected to provide an
unreasonably good opportunity for fraudulent behavior.  I
believe that I understand that argument, but many of the
people to whom your statement was addressed probably do not.
I would strongly suggest that you try to produce a document
that explains the problems, in a tutorial style, and at a
level adequate to educate someone with no knowledge of Chinese
and, perhaps no working knowledge of any script other than his
or her own.

(ii) If registrations of CDNs are dangerous, prohibit them in
your own ccTLDs and, with the document mentioned above as a
tool, recommend to ICANN and directly to other domain
administrators (top-level and otherwise) that these codes not
be registered.  That advice will, I fear inevitably, be
ignored in some places.  But, if the explanation of the
consumer risks is clear enough, governments, regulators, and
injured parties will have solid grounds for holding registrars
who promote the use of such names responsible for any problems
they cause.

(iii) Please continue to work with me and with others to
develop non-DNS solutions for user-visible names that do meet
your needs by accomodating matching that uses language and
other information in ways that the DNS cannot.

thanks and regards,
     john


--On Friday, 01 February, 2002 18:27 +0800 Erin Chen
<erin@twnic.net.tw> wrote:

> --------------------------------------------------------------
> -----------         Chinese Domain Name Consortium (CDNC)
> Declaration
> --------------------------------------------------------------
> -----------
>
> Based on current status of Internationalized Domain Name, CDNC
> makes  following declaration:
>
> 1. Currently, IETF IDN WG begins Last Call for the WG core
> documents (IDNA, NAMEPREP, STRINGPREP, and PUNYCODE).
>
> But, the architecture of IDN defined in above four documents
> does not solve the traditional and simplified Chinese
> character variant problem. That will cause serious delegation
> problem in the application of  Chinese Domain Name. This
> technical flaw can't be compensated by  registration policy;
> it'll prevent Chinese Domain Name from being used widely. All
> the future Internet applications based on Chinese Domain  Name
> hierarchy will fail and ultimately Chinese Domain Name
> technology will result in failure.
>
> 2. Under the commercial and social pressure, IETF IDN WG
> issues Last  Call hastily without presenting a proper Chinese
> Domain Name technical solution. It will be detrimental to the
> ethnic-Chinese Internet community.
>
> 3. IETF IDN WG does not solve Chinese Domain Name technical
> problem,  but it is IETF IDN WG's responsibility to sincerely
> consider the  consequences from adopting these drafts. Under
> the current condition,  if IETF approves these IDN drafts,
> registrars will open Chinese Domain  Name registration without
> considering the requirement of Chinese Domain  Name and
> Chinese Domain Name will fall into confusion. This will damage
> Chinese Internet community seriously.
>
> 4. IDN would be flawed by not adopting CDN requirements.
>
> Chinese Domain Name Consortium
> CDNC Founding members: CNNIC, TWNIC, HKNIC/HKDNR, MONIC
> Co-chair of CDNC: Qian Hualin, Professor
> Co-chair of CDNC: Shian-Shyong Tseng, Professor
> 2002.2.1
> --------------------------------------------------------------
> -------------
>
>
>
>
>
>
>