Re: [wg-c] Re: Importance of the Registry

[Jonathan Weinberg <weinberg@mail.msen.com>]

	Here's where I think we are.  On the question of what registries should be
like and how they should work, we've talked about two issues:

(1) Should ICANN require that all registries be shared (that is, support
competitive registrars)?
(2) Should ICANN require that all registries be non-profit?

	Those two questions don't have to be decided the same way.  I can imagine
three plausible scenarios that might arise from different answers to the
questions:

A. All registries are non-profit and shared.
B.  Some registries are non-profit and others for-profit, but an ICANN rule
requires all to be shared.
C. Some registries are non-profit and others for-profit, and ICANN has no
rule regarding sharing.

	I'm going to try and talk about the two questions separately.

	The advantages of requiring sharing is that it ensures competition, on an
equal playing field, in the provision of registrar services in any given
TLD. We had an extensive discussion in this thread about whether
competition in registrar services in important.  Most folks argued yes, and
that seems right to me.  (For starters, the bulk of the registration dollar
is going to the registrar, and not the registry.)

	At the same time, there are some respects in which I think Milton's
correct that there's less here than meets the eye.  Consider first a
scenario involving a gTLD with insignificant market position, and no
meaningful market power as a registry.  That gTLD isn't going to get *more*
market power if it operates as its own sole registrar: If it offers its
registrar services at unattractive price and service levels, customers will
use some other gTLD.  Put another way, competition between gTLDs will
impose market constraints on its registrar services.  Consider next a gTLD
with a whole lot of market power.  That gTLD isn't going to get *less*
market power if we require it to support competitive registrars.  Any
market power it can't exert on the registrar level, it'll just exert at the
registry level instead.  (So that it would be pointless to require NSI to
implement an SRS if someone were not also regulating the fee NSI is
charging registrars for registry services).  On the other hand, a sharing
requirement may well provide important safeguards against registry
cluelessness, b/c it makes it much more likely that somebody who is in fact
skillful at providing registrar services is in fact operating in every TLD.
 I think that advantage isn't one to sneeze at.

	I can think of only one situation where requiring sharing might actually
be harmful, and that's in connection with a limited-purpose domain where
the registry operator is using discretionary criteria to decide who can get
an SLD in that domain.  That is, imagine that somebody sought to run
".family", limiting registration to folks operating WWW sites that in its
opinion were family-friendly.  (I've seen this referred to as a
"value-added" domain.)  The success of that domain would likely depend on
the public's view of the care and discernment the operator exhibited in
making those choices.  That business plan wouldn't succeed if any registrar
could add names to the .family domain in accordance with its own views.

	It might not be a terrible thing, therefore, for ICANN to enunciate a
*presumption* that all TLDs must support competitive registrars, subject to
exceptions in cases where the operator shows that there would be
significant advantages to doing it some other way.

	The next question is for-profit versus nonprofit.  The principle advantage
of allowing for- profit registries is that for-profit entities will have
more incentives to compete and innovate, seeking to heighten their
efficiency, lower their prices, and provide additional value-added
services.  The principal disadvantage rests in concerns about lock-in. It
may be expensive for a registrant to change TLDs after a number of years,
because the change will require the registrant to spend money educating
consumers about the new url and fixing broken links; even so, the change
may be disruptive.  To that extent, a TLD registry can exercise market
power over its existing customers, leading to artificially high prices and
other bad things.  It's been urged that a for-profit would be more likely
to exploit that market power than a non-profit would.  (For a good summary,
see <http://www.gtld-mou.org/docs/eco-structure-registries.htm>, helpfully
provided to me by a list member off-list).

	Javier has urged that even without regard to lock-in, a nonprofit
controlled by its customers, working on a cost recovery basis, is the most
likely to operate efficiently and charge low prices, because the registrars
can demand that it operate efficiently, and there is no owner "seeking to
get as much money as possible from the registry."

	I'm not sure, though, that the concern about lock-in is dispositive (even
setting aside the discussion, raised by Chris and others, about long-term
contracts).  At least so long as the market is growing, it won't be a good
strategy for a registry to put the screws on existing customers, because
that will discourage new customers.  Further, it's not obvious that
non-profits are that much less likely to exploit their market power than
for-profits.  I'm aware of some studies comparing non-profit and for-profit
hospitals in terms of the degree to which they in fact exploit market
power; those results were inconclusive.

	And I'm terribly dubious of the proposition that non-profits, working on a
cost-recovery basis, will have lower prices than for-profits in competitive
markets.  The evidence I see around me is to the contrary.  For-profits in
competitive markets have great incentives to reduce their costs and prices;
specifically, in a competitive environment, a registry that wants to
attract new registrants will keep its costs and its prices to registrars
low.  Non-profits (even if they are owned by their customers) don't seem to
me to have mechanisms that are as effective to do either.

	So on that basis, I'd tentatively suggest that it makes sense to allow
both non-profit and for-profit registries to play in this playground (each,
for the reasons I stated above, presumptively would support a large number
of competitive, private, for-profit registrars).  Members of each group
would have something different to offer.  The large domain holders who are
most vulnerable to lock-in would be free to use domains run by non-profits
if they thought that a safer course; other folks could take advantage of
the dynamism and innovation of the for-profits.

	One caveat: This reasoning makes the most sense if (as I'd urge) ICANN
adds a whole lot of new gTLDs.  If ICANN adds only a very small number of
new domains, then it may not be realistic to talk about a "competitive
market" operating at all.

Jon


Jon Weinberg
weinberg@msen.com