W3C home > Mailing lists > Public > www-tag@w3.org > February 2005

Re: I-D ACTION:draft-hendrikx-wallis-urn-nzl-00.txt

From: Roy T. Fielding <fielding@gbiv.com>
Date: Sat, 12 Feb 2005 14:02:34 -0800
Message-Id: <600f74e92f414cc6d15149b135bb2ce7@gbiv.com>
Cc: W3C TAG <www-tag@w3.org>, colin.wallis@ssc.govt.nz, ferry.hendrikx@ssc.govt.nz
To: Michael Mealling <michael@neonym.net>

On Feb 11, 2005, at 3:49 PM, Michael Mealling wrote:
> On Fri, 2005-02-11 at 15:35 -0800, Roy T. Fielding wrote:
>> This is really disappointing, but entirely predictable.
>> People (and governments) are being sold on URNs based on
>> a false pretense -- that being "location independent" makes
>> an identifier more persistent.  To make matters worse, they
>> assume that NZL will forever remain the three-letter code
>> for New Zealand.  *sigh*
>
> And a .nz ccTLD in an "http" URI will?

At least as long, if not longer -- the number of independent
organizations that already depend on .nz persistence is several
orders of magnitude greater than the only organization that would
care about the NID.

I didn't say that other schemes were *more* persistent than URN.
I said that the persistence of URN is being sold on false pretense.
It is no more persistent than any other registry-based system.

> The assignment policy for URN
> NIDs is permanent. Forever. Full Stop. So that NID will belong to the
> New Zealand government forever.

Only in your dreams.  There is no evidence that NID assignment
won't be influenced by the same factors (social, legal, and competitive)
as DNS names.  What you, and they, are ignoring is that such a name
is vulnerable to political situations.  For example, one wherein a
government is elected based on a platform of New Zealand being
renamed Aotearoa (the maori name) [the current unlikeliness of such
a scenario is not relevant].

Renames are not based on logic or respect for procedure.
They won't care how many systems assume NZL (or nz, for that matter).
They won't care what IANA says of its own procedures -- they are a
national government.  The only reason that events like these
(on a smaller scale) haven't happened to URNs is because they have
not been deployed for anything that actually required persistence.

> The point is that the identifier is
> network independent, not location independent. A URN namespace stays
> assigned to its owner whether or not ICANN, DNS, HTTP, or the W3C 
> exist.

IANA is just another registry, with considerably less influence on
society than ICANN.

> I was hoping we were passed this "my scheme is bigger than your scheme"
> contest....

Maybe if you stopped thinking of objections in such a knee-jerk
manner you would realize the flaws in URNs.  They are just another
form of registry, somewhat more ponderous than DNS but less valuable
to the overall system of the Internet.

The reason not to use artificial names like URN is because they
tend to go out of fashion just a couple years after being defined.
They have less value as identifiers because they present less
value to the people who use them.  They are far more difficult
to validate, require special (under-developed and untested)
software for processing, and do not benefit from the network-effects
of existing information systems.  In short, there is absolutely no
value obtained by preferring

    urn:nzl:govt:registering:dogs:registration:1-0

over

    http://registry.govt.nz/dogs/registration/1-0

and yet the former will cost the people of New Zealand considerably
more, both in deployment cost and the lost value from not making use
of the information system that is already accessible to everyone.

And while you may consider such bad design to be just another
experiment in failed URN deployment, I think of it as the extra
tax money of my extended family living in New Zealand going down
the drain.  I don't like that, and neither will they.


Cheers,

Roy T. Fielding                            <http://roy.gbiv.com/>
Chief Scientist, Day Software              <http://www.day.com/>
Received on Saturday, 12 February 2005 22:02:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:32 GMT