Re: STOP gTLD-MoU -Reply -- Ad Nauseum...

[Dave Crocker / IMC <dcrocker@imc.org>]

At 12:47 AM 6/25/97 -0700, Einar Stefferud wrote:
>The ARPANET was not built to survive Nuclear Bombs, or equivalent.  It
>was simply built to interconnect ARPA Contractors for the purpose of
>sharing whatever resources could be shared by being Network Connected.

	I get frustrated with the hyperbole being used, too and laud your effort
to get the historical perspective back into a reasonable basis.  One, small
clarification:

	As Katie Hafner documents in "Where Wizards Stay Up Late", there really
were some folks who used language about surviving nuclear attack.   This
was a bit of a surprise to me, since it was not language that was in
typical use.

	The usual phrase for characterizing the nature of the technology being
developed was:  "Survive hostile battlefield conditions" and it was very
clear that this presume conventional, not nuclear, weaponry.  The original
packet switches were even built to military specs and were claimed to be
able to survive being dropped from 2 stories up (e.g., by parachute.)

>here is to debunk this notion that academics had some kind of
>stranglehold on the ARPNET to assure that it would not work for
>business purposes.  What a quaint idea;-)...  Actaully, a lot of
>people made very good livings out of working on the Internet, and many
>of them were not academics.

	indeed!

>Over that early time period, lots of business was transacted among the

	This is a very important point.  Commerce was present on the Internet from
the very start of its operation, albeit in limited form.

>Now, after TCP/IP was deployed by ARPA on the ARPANET, it became
>possible to create what became the Internet because it was no longer
>necessary for ARPA to control everything in all IP connected networks,
...
>needed.  Among these changes were conversion from a Government
>Provided Backbone, to our current multi-competitor Open Commercial
>Internet Backbone.

	
	NSF created a second backbone and thereby forced changes to the "exterior"
routing protocols.  This act by NSF had the larger effect of permitting a
truly competitive Internet.  Otherwise, we would have the One True Backbone
provider.  (The step was taken at some expense.  It would seem that every
time we explore moving from exclusive control to competitive service, we go
through some tension...)

>This brings us up to the present mess where suddenly it becomes clear
>that there is no clear line of authority for IANA or anyone else to
>Czar the DNS TLD ROOT, and we are in a big fight about whether there

	Stef, you just moved from 10 years of clear and comfortable authority,
based on community consensus, to "no clear line of authority".  Nothing has
changed about the Internet to create this sudden shift, other than a
position primarily espoused by NSI and a few individuals.  No actual change
in operation fact has occurred.  This, then, forces one to question the
basis for claiming that IANA suddenly has no authority.  What is it about
community consensus has really changed?

>Some of us say YES, and some of us say NO!  And there is no obvious
>source of authority to decide the question for us.  The IANA autority

	There is the same authority there always has been:  development of
sufficient community support to provide legimization.  The growing number
of gTLD MoU signatories is the example.  The lack of viable alternatives
which respond to the pressing problems is the other.

>was derived from community acceptance of administrative actions taken,
>but that community consensus authority did not extend to an IANA hand
>over of it this same community consensus authority to someone else.

	Huh?  IANA is not handing its authority over to anyone else, so why do you
mention this?  Its historical handling of DNS name assignment has always
been based upon delegation of authority for portions of the name space, but
that is different from handing over its own authority.

>None the less, I predict that in the end we will discover that the DNS
>ROOT does not need a Czar to select TLD names for the DNS any more
>than other parts of the Internet need a Czar.  It only needs to avoid

	In the face of scarce resources, unclear scaling effects and the
requirement for unique assignment, we always need a central authority.
That'ts true for IP addresses and it's true for domain names.  It's true in
the past, the present and the future.  One hopes that a lighter-weight
mechanism will be tolerable for domain names, as we find ways to respond to
concerns about scaling effects, but there are far too many and far too
serious concerns about excessive expansion to the TLD space to permit a
free for all at this stage.

>I note that some new TLDs have somehow been snookered into the
>official root servers, and I must say that I see lots of distraught
>complaints about the impropriety, but I see absolutely no indication
>that any aspect of root service was diminished in any way, so I am
>convinced that it is indeed harmless to add new TLDs to the root.  If

	One needs to be very, very careful in asserting a general rule based upon
a single, brief data point.

	Very careful, indeed, especially concerning unrestricted repetition of
such an event.  People who do hands-on operations of systems are very
conservative.  They become that way because they know well the difference
between doing a thing once and doing it on a large scale.  It isn't that
doing it once must be difficult before one must be careful.  It's that the
mere fact of large-scale repetition renders the simplest changedifficult.

>So, all these social theories about a class war between academia, the
>military, and business are just so much foolishness.  What is really

	indeed.

>happening is just that the DNS ROOT is the last of the original
>ARPANET vestigial tails to fall off the Internet skeleton.

	neither the last, nor vestigial.  fix what is broken, not what isn't.

>...  The fact
>that lots of administrators accept Paul's defaults is the critical
>key, and when administrators stop accepting those defaults and
>choosing others, we will all see that the servers do not have the
>ability to control anything about what TLD names are "in the ROOT"

	Consent of the governed has always been the claimed basis for the current
structure.  the fact that this particular piece of technology requires a
single operational authority does not alter the fact that that authority
derives its legitimacy from such consent.

>The fact is that all name resolver administrators have the power to
>point at whatever root servers that want to point to, no matter who
>says otherwise, since they are the people who know the passwords, and

	I'm unaware of anyone saying otherwise.  Can you provide specifics?

	What HAS been said is that we need all those clients to point to THE SAME
root service.  As soon as substantial portions of the net point to
different root services, then access on the net will become fragmented.

>    Internet, and it should be abundantly clear that Dave Crocker
>    speaks only for Dave Crocker, and I speak only for myself.

	at the very BEST, I speak only for myself.  sometimes I'm not even sure of
that...


--------------------
Internet Mail Consortium
                                                    +1 408 246 8253
675 Spruce Dr.
                                                    fax: +1 408 249 6205
Sunnyvale, CA 94086 USA
            info@imc.org , http://www.imc.org