RE: [wg-c] voting on TLDs

[Dave Crocker <dcrocker@brandenburg.com>]

At 09:10 AM 3/12/00 -0800, Roeland M. J. Meyer wrote:
>---- Dave Crocker, Monday, March 06, 2000 8:45 AM
>The concern for stability has been present from the start of discussions
>about gTLD expansion, roughly five years ago.  It has covered:
>---- Roeland M.J. Meyer, Sunday, March 12, 2000 0900 hrs.
>What Dave fails to mention here is that most of the triggers for those
>discussions were Dave Crocker and Kent Crispin repeatedly raising the same
>issues over and over again, just as they have done here, although number

These concerns were raised long before my involvement.  The fact that some 
of us have been persistent in reminding folks about the range of factors 
does not mean that we created the factors.  It is odd that the strong 
concerns from the trademark community, and the potential impact of a very 
litigation-oriented community, would be attributed to Kent or me.


>issue with a vetting process, and is not effected by item 1. Item 3 is a
>policy jurisdictional issue which we have no power to effect (nor do the

Taking actions that exacerbate legal concerns of (powerful) trademark 
holders is not an matter of abstract policy.  Choosing to create an 
undisciplined process of addition is quite clearly a strong irritate.  The 
trademark community has been quite clear and forceful on this point.


>The only point Dave leaves off here, and rightly, from his perspective, is
>the natural limit of the marketplace. The most optimistic view is that there
>appears to be a maximum market size of 3000-4000 TLDs, depending on how they
>are defined. This appears well within the technical limits of the current

As always, many thanks for the marketing lesson.  But where DID those 
numbers come from???

In any event, most efforts at estimating new markets turn out to be 
seriously off the mark.  Classic examples are IBM's original estimates of 
the computer market and Sony's marketing folks' estimate of the 
Walkman.  Both were wildly low.

There is no serious basis for claiming any firm estimate of the TLD 
market.  Estimates based on casual surveys of current participants have no 
methodological validity.  In fact they are sure to be grossly 
misleading.  So rather than pull estimates out of the air, responsible 
planning requires taking an approach that can handle the possibility of 
massive demand for new TLDs.


>root servers and gtld root servers <g>. However, arguments of the
>self-healing nature of the DNS is that the Internet, at large seems to be
>able to ignore these outages more successfully, since then.

What evidence is there that average users are able to comfortably "bypass" 
outage of DNS service?  All the evidence I am aware of says that average 
users instantly become non-users, until the DNS is fixed.


At 12:08 AM 3/13/00 -0500, Milton Mueller wrote:
> > incremental cost for adding a new gTLD is close to zero.  Any
> > technical/business infrastructure that can provide a registry service
> > for a single gTLD can provide registry service for a hundred gTLDs.
>
>Agreed again. >chuckle< Kent, have you ever thought about what this fact
>does to your and Dave's scare tactics about the "operational risk"
>associated with adding TLDs? Kind of blows it out of the water, doesn't it?

Odd that operational stability is treated so lightly... While it is an 
appealing debating technique to use emotional language like "scare tactics" 
it is not constructive or professional.  Wasn't there supposed to be fresh 
interest in keeping these exchanges serious and substantial?

As to the real issue, yes adding administrations and operations 
organizations is probably more significant to stability than adding TLDsto 
a data base.  But that is a "probably".  And the distinction, in this stage 
of discussion, is not relevant.  The fact remains that performing any 
change to the DNS, a large number of times and in a short period of time, 
is likely to have unanticipated results.

In systems work, "unanticipated" is sometimes equivalent to "bad" and the 
rest of the time it is equivalent to "very, very bad".


>Ditto my comment immediately above. This is just too good. If there are no
>costs and no scaling problems associated with adding TLDs, then this debate
>is pretty much over, isn't it?

This underscores one of the real difficulties in these discussions.  Some 
people are more interested in winning a debate than in developing 
constructive (and safe) proposals that represent a real balancing among 
competing constraints.

We need the latter.  The former is quite dangerous.



At 02:04 PM 3/12/00 -0500, Milton Mueller wrote:
>Kent is saying that groups could propose, and ICANN accept, TLDs WITHOUT the
>proposer having arranged for specific contracts or operational arrangements
>to actually register names. That position seems unworkable, and antithetical
>to the emerging consensus.

Avoiding, for the moment, the rather curious assertion of emerging 
consensus, a core question is why separating the choice of name strings 
from the administration of names under them is at all "unworkable"?


>Under Kent's alternative, some group could be designated "policy authority"
>over a TLD without having any demonstrated ability to actually provide a
>service on the Internet. The relationship between its policy authority and

In fact, such a separation is the prevalent form of TLD 
administration.  Most TLDs have a country (or the like) as the registry 
authority, with an operations group assigned a contract for the daily 
work.  That also is the mode NSI was chosen under.  And their track record 
at the time of original assignment was quite limited.

In any event there is nothing in Kent's description that comments on the 
degree of "vetting" that should or should not be done, concerning 
assignment of operations responsibilities.  Having the registry authority 
separate from, and provide oversight to, the registry operator, does not 
abrogate any control over operations requirements.


>the implementation of the policy via the registry operation would not be
>specified. On what basis, then would the delegation of a TLD be made? On the
>grounds that ICANN "liked the idea"? Not good enough. On what basis would

When asking a question, it is best not to then answer it and take the 
hypothetical answer as fact.  The question is a good one and one that 
should be a focus of this working group.  The proffered, hypothetical 
response is silly and unconstructive.


> > Note that I am not in any way restricting a registry from making a
> > proposal for a TLD -- I am simply saying that there is no good reason to
> > restrict proposers to registries only, and many good reasons not to make
> > such a restriction.
>
>Both you and Brunner seemed to be equating "registry" with "existing
>registries."
>This is a logically unnecessary equation.
>
>Anyone who proposes to run a tld is a registry.

That is like saying that anyone who is accused of a crime is a criminal, or 
that world peace is achieved just by wanting it.  Let's try to be at least 
a little careful, and reasonable, with terminology.  A group is not a 
registry unless and until they are given the task of being a registry.


>Thus, any group that wants "policy authority" has to team up with a viable
>operations entity when making its proposal, or do the job itself. Whether

That suggests that a registry authority (that's the term that has been 
around for awhile) is locked into using a particular operations 
contractor.  Doesn't seem like such a good model; does not permit switching 
to a better operator.


>This is where we part ways. There is no difference. For ICANN's purposes, an
>organization that gets a TLD delegation is a registry. This is true
>"Don't have anything to do with" are strong words. Do you mean that ICANN
>would select an operational partner for them? They could not change the
>operational contract when they wanted?
>
>Give me one real-world, actually existing example of such a thing. Certainly

OK:  The original IANA/NSI relationship.  IANA did specify the contents of 
the root, but otherwise "had nothing to do with" daily operations.  That is 
also the model for a number of ccTLDs, such as Tuvalu.

And, please, let's not quibble about the fine-grained definition of "has 
nothing to do with".  The concept distinguishes between authority/oversight 
versus daily activity.  That's an important distinction, while this group's 
debating exactly where the line is drawn is not.


> > Which is easier:
> > 1) Given an approved TLD string, find a qualified registry to operate it
> > 2) Given a registry, find an approved TLD string to operate.
>
>You don't seem to have thought this through carefully.

yeah.  right.

The real distinction is between registry "ownership" of a TLD, versus 
community ownership, with the registry being given (temporary) 
responsibilities.

As long as there is such interest in "emerging consensus" let's be clear 
that there has been a long, strong and very broad consensus about treating 
the DNS as a public resource/trust.  (And let's not get into legal 
semantics games again, please.)

That means that the registry folks do NOT own the thing but, rather, have 
current responsibilities and can be replaced.  As has been done many 
times.  Such as when NSI replaced SRI.


>Also, Your thinking about this is limited by the fact that you think of a
>"registry" as a generic function with no differentiation. But in fact many

Exactly right.  The tasks of a registry are quite straightforward.

There might be other functions such a group could choose to perform, but 
they are distinct from basic registry administration... and therefore need 
not be tied to it.


>of the most interesting new TLD proposals are precisely those which use DNS
>to create some new functionality, such as a privacy-enhanced name space, or

There has been a constant stream of proposals over the 15 years of DNS 
history.  None has panned out, especially none tied to TLD names.

We are again back to the problem is letting interesting, but untested, 
ideas affect policy, versus continuing established practises and 
maintaining stability.

The former is policy-by-whimsy-of-latest-fashion.  The latter is 
responsible operations.

d/

=-=-=-=-=
Dave Crocker  <dcrocker@brandenburg.com>
Brandenburg Consulting  <www.brandenburg.com>
Tel: +1.408.246.8253,  Fax: +1.408.273.6464
675 Spruce Drive,  Sunnyvale, CA 94086 USA