Re: [wg-c] Proposed gTLDs: The IAHC Seven

[Kent Crispin <kent@songbird.com>]

Pardon the length of this...

On Mon, Apr 10, 2000 at 10:52:21AM -0700, Mark C. Langston wrote:
> PROPOSAL:
> 
>   I propose that we allow no pre-sold TLDs to become part of the
> testbed, and I further propose that we re-evaluate this position once
> the testbed period has ended.

I agree totally.  There can be no pre-sold TLDs.  Moreover, I am TOTALLY
SUPPORTIVE of the fairness concern that underlies this proposal.  The
IAHC names do not have the problem you are trying to address, however. 

Consider:

The instant that any names are publically approved by ICANN people will
start considering what SLD names they may wish to register in the new
TLDs, and making up their internal lists of names they might want to
have.  We obviously can't stop that. 

I, Kent Crispin, as a person who registers domain names for people,
could advertise to my current customers that I will take
"preregistrations" for SLDs in the new TLDs, which would mean that as
soon as the TLDs become active, I will try to register the those SLDs in
those new TLDs.  There is no guarantee that the registrations will
succeed, of course.  There is no way to restrict this activity, either. 

An ICANN-approved registrar can do exactly the same thing.  They can
(and most assuredly will) start taking "preregistrations" on exactly the
same basis, as soon as *any* new names are approved.  They will, as a
practical matter, start taking "preregistrations" as soon as there is a
high probability of a name being approved.  They will be forced to, by
their customers.  [This is exactly what happened with many of the CORE
registrars -- customers would ask to be put on a waiting list, and if 
the registrar said "no", the customer would go to a registrar who said 
"yes".  The competitive forces for waiting lists are *very* great.]

In any case, realistically speaking, and given the international span of
registrars, there is very little that can be done about this, either --
this is a level of business activity that ICANN simply can't afford the
time to police.  Moreover, *nobody* wants ICANN to become an intrusive
international regulatory body. 

And anyway, there is really nothing at all wrong with registrars keeping
lists of names that they will *attempt* to register on a customers
behalf, should the TLD become available. 

Note that all that has been said so far relates directly to ICANN
accredited registrars for *any* new shared-registry TLDs, and it has no
necessary connection to CORE or the IAHC names. 

During startup of *any* new gTLDs there will certainly be an initial
rush, where multiple registrars will try to register the same SLD.  This
will be true whether it is the registrar itself that keeps a list, or
whether it is a list held by an ISP, or whether it is just an individual
registrant desiring a good name.  Some process needs to be in place to
guarantee fair access at the registrar level -- for example, a large
registrar with high bandwidth and a lot of names on a waiting queue
should not be able to swamp the input to the registry when it comes on
line.  Small registrars should have an equal opportunity to register 
names. 

Note once again that this is a *general* problem, completely independent
of CORE.  It is, however, precisely the same problem that CORE had to
address in the design of the CORE SRS (Note that NSI doesn't have a
"startup problem" like this).  CORE settled on a round-robin allocation
algorithm that gave fair opportunity to each CORE registrar.  ICANN will
need to develop a similar algorithm to handle the startup problem, as
well.  [The algorithm can continue to operate after startup, but the
probability of a name collision becomes much much less after the initial
rush.]

Presume, therefore, that before startup of any new shared gTLD, ICANN
will require some fair algorithm -- call this the "round robin"
algorithm (since at some level a round robin allocation will almost
certainly be required). 

CORE currently is an accredited ICANN registrar, and as such would, AS A
SINGLE ENTITY, be entitled to precisely one slot in the round robin
algorithm.  Say that there are 50 active CORE resellers with legacy
queues of various sizes; and 100 active ICANN registrars.  Each of the
CORE resellers has one chance in 50 of being the next round robin slot
at the CORE level, and from there has one chance in 100 of getting the
round robin slot at the ICANN level.  [The same issue applies to Tucows
resellers: Tucows, as a whole, would get 1 slot in the ICANN round robin
(though Tucows may not worry about fairness at the Tucows level).]

The issue changes somewhat if CORE is also selected to be a registry for
one or more of the new domains, and the TLD involved was one of the IAHC
TLDs.  Should the queues that the CORE registrars maintain have any
precedence over the queues that an ICANN-accredited registrar might
have? No, absolutely not.  EVERY registrar must have fair and equal
access to the registry, whether a CORE member or not, or indeed whether
the registrar is CORE itself, or not.  This would probably require some
kind of serious split between CORE as a registry and CORE as an ICANN
approved registrar.  That's life.  It is, in my opinion, an absolute
requirement that a registry give all registrars precisely equitable
access, and that indeed, there be *no* presold registrations. 

But there is no way whatsoever to prevent registrars and others from
keeping private queues, and, in fact, they are relatively harmless. 

> The IAHC 7, on the other hand, were of dubious value, were not in the
> roots, and were only reachable by those willing to configure their
> systems to handle multiple independent roots.

Nope.  They were never reachable at all.  CORE never set up alternate
roots (except for internal testing purposes); and it never sold
registrations in any alternate roots. 

> They should not have
> been pre-sold as though they were going to be added to the roots
> any day now.  However, if I recall the ad copy, that's how CORE was
> pushing them.

Nope.  CORE has *never* presold any registrations, at all.  Individual 
CORE registrars have kept private queues, as described above.  There is 
nothing that can be done about that; it will happen with any new names.

[...]

> Introducing new gTLDs where the namespace has been presold does not
> solve this problem.  It simply introduces new gTLDs in which all the
> "good names" have been taken before the public as a whole has a chance
> to register anything in them.

There is, as described, an intrinsic startup problem.  But as far as 
CORE is concerned, it is most emphatically the case that there are *no* 
presold registrations; *every* ICANN-accredited registrar will have a 
precisely equal probability of being the one to register any given name.

[...]

> Anyone who registered such a name knew, or should have known, the
> risk they were taking by giving money to a registrar selling SLDs under
> TLDs not in the sanctioned roots.  There should have been -- indeed,
> could not have been -- any guarantee that those registrations would
> carry over if the TLD was ever added to the roots.

On the whole, no guarantees were made.  CORE as a whole doesn't have any
realistic way to control the exact wording of an individual registrars 
web site, of course, and that is true of ICANN in dealing with 
ICANN-accredited registrars.  [It is my personal opinion that there 
were indeed some CORE registrars with lousy ethics, just as it is my 
opinion that there are some ICANN-accredited registrars with lousy 
ethics.  That is completely to be expected.]

> If people want to keep operating TLDs outside the authoritative roots,
> fine. 

CORE has *never* done that.

> if people want to sell SLDs within them, fine.

CORE has *never* done that.

> If people are
> willing to give these people money for those SLDs, again fine.

Some CORE registrars have charged processing fees for maintaining a
waiting list -- but that's a completely reasonable business practice. 
And at the beginning, there were a very few CORE registrars who made
outlandish claims that were, in my opinion, borderline fraud.  But CORE
is a membership organization that has no realistic way of policing such
nonsense, and as the probabilities and the realities became obvious, the
idiocy went away. 

The fact is that preregistrations don't guarantee anything of
significance to the "preregistrant", and at this point most people 
realize that.

> But
> don't presume that you're doing a public good by arguing to add a
> presold gTLD to the root to alleviate the current shortage of names
> within namespace.  It's painfully obvious when you're thinking with
> your wallet.

The public good comes from the fact that there are users out there who
know the IAHC TLDs -- POC regularly gets mail from people asking when
they will be able to register in one of the TLDs or another, and the
names are very well vetted.  These names have been out in the public eye
for two years now, and they are not bad names.  Aside from a couple of
obvious examples, there are no conflicts.  It may well be that some of
the old CORE registrars make a few dollars when they shake the dust off
of their old web site and find some people on their lists who are
actually still interested, and who manage to make it through the
probabilities, but it would not be a giant windfall -- maybe they could
treat the employees to pizza or something?

Please note that I am not thinking with my wallet.  CORE has never paid 
me a dime.  In fact if I were thinking with my wallet I would be doing 
something else right now.

Kent

-- 
Kent Crispin                               "Do good, and you'll be
kent@songbird.com                           lonesome." -- Mark Twain