RE: [idn] An open letter to the IDN WG (long)

["Brian W. Spolarich" <briansp@walid.com>]

| I think two things are mixed up here:
| 
| (a) People seem to want to be able to use non-ascii characters in the 
| identifiers we store in DNS.
| 
| (b) People see to want to store words in DNS, and not use a 
| directory, and they forget that DNS is used for identifiers and not 
| words.
| 
| It is very hard to fulfil these two things, because it might be that 
| people want to do (a) above because of (b), and not because they 
| really need non-ascii characters in the identifiers.

  After hearing the CNNIC presentation at today's WG session, I 
think that this analysis is correct and that John is indeed 
conjoining two issues that should be 'pried apart'.

  The CNNIC folks are struggling with how to make a 'usable
DNS' for Chinese speakers who are struggling with the results of
the Simplified vs. Traditional Chinese script problem.  However
I think that their definition of 'usable' exceeds the mandate of
the IDN WG, and even the DNS in general.

  While the ambiguities between the spoken and written form of
a particular language (and the interaction with the written
form being used as an identifier in the DNS) may be perhaps
worse with TC/SC, these are not problems unique to those
languages and scripts.  If I type in 'www.pain.com' into 
my web browser I might be looking for something about
French Bread or advice on managing migraine headaches.  The
fact that I get information on the latter and not the
former isn't the fault or even the job of the DNS.  Human
language is rife with homophones, cognates (both true
and false), differences in spelling, and other complex 
ambiguities.

  This is why things like search engines and web directories
and portals are so widely-used, and this area of great 
interest to folks involved in NLP and AI.  None of that
belongs in the DNS.

  Paul pointed out that this problem should not be characterized
as an Asian language problem, which I think should be
underscored.  In this respect the DNS 'disadvantages' users
of all scripts equally. :-)

  I think that these are all valid and important concerns
and point to the need for further work in this area.  Indeed
I hope that the BoF on Internationalized Keywords and
Directories tomorrow can form the nucleus for starting that
work.

  One of the things John has said repeatedly is that if we
approach this as a 'two phase' solution (where the Universal
Directory System is phase two) that we'll never get there.
While I do defer to the hard wisdom gained by John and others,
I have to think that this view is entirely too pessimistic.  
We've successfully deployed new services on the network before.  
If a new service sufficiently meets the needs of the user
community, then it has a high likelihood of being deployed
and being 'successful'.

  I realize that John's concerns also involve the increasing
complexity of the network, both in terms of its scale and
implementation, number of interacting elements, and the 
fuzzier problems of the humans who use the thing.  Our 
intuitive models on how to approach problems don't do well
in these contexts.

  At some level this is about managing user expectations.  One
of the jobs of the IETF should be to work with the user 
community to ensure that it is understood what problem a
particular piece of technical work is intended to solve.  If
the user community rejects IDN because it doesn't provide
the answer to Life, the Universe, and Everything, then
we've not done a good enough job of educating the user
community on what it is we're doing.

  Indeed John has the privilege and responsibility of
being one of those educators.  In his work with ICANN
and other organizations perhaps he could focus on fostering
the understanding in the user community on these issues.

  Best,

  -bws