Public Comment

[Robert Szarka <szarka@downcity.net>]

To: ICANN Board Members 
Re: Proposed Guidelines for the Accreditation of Internet Domain Name
Registrars
Date: 16 February 1999

Thank you for this opportunity to respond to your proposed guidelines,
posted at http://www.icann.org/draftguidelines.html and dated 8 February
1999.  As Managing Partner of Operations at DownCity, LLC (an Internet
Service Provider), I am an interested party, but please note that my
comments are my own and not offered as an official statement by our company.

I will limit myself to the consideration of a few specific questions:

Q4:

What conditions, other than the availability worldwide of a reasonable
number of alternative registrars and easy portability of domain names, are
important to the introduction and maintenance of robust competition in the
provision of registrar services?

A4:

The words "reasonable number" are not defined specifically enough to
provide a useful guideline for your activities.  The "reasonable number" of
taxi services in a large city, for example, may be "one", according to the
owner of Lucky Monopoly Taxi, Inc. and his friend, the mayor.  Some higher
number might be desired by the patrons of taxi services with no alternative
to his poor taxi service.

Were registration service stictly homogenous, the benefit of a large number
of registrars would be the tendency to drive price down to an efficient
level and, consequently, produce an optimal number of registrations.  This
is, in itself, a worthwhile goal.  But there are many dimensions of quality
registrars might choose to emphasize that registrants may value
differently, so that a large number of registrars would also offer more
exactly the services registrants require.  For example, one registrar might
emphasize the security of its service, offering verification options
ranging from the long-unimplemented (by Network Solutions) 'notify after
use' email security feature to validation of signatures using strong
public-key encryption.

The most recent service failures at INTERNIC provide another example:  one
registrar might offer premium service with live service representatives and
instant registration, charging a higher price and requiring payment in
advance, while another might follow policies like those currently in force
at INTERNIC that encourage the abuse of the registration system by domain
speculators.  In that case, the customers of the first registrar would be
isolated from the externalities that might be imposed on them by the
customers of the second registrar under the current monopoly system.

What then is the optimal number of registrars?  If the costs of
participation did not vary with the number of registrars, clearly the
optimal number would be "as many as found it profitable".  Where costs do
vary, registrars should be charged the marginal cost imposed on the system
by coordinating their activity with that of the other registrars.  In that
way, any potential registrar who can offer services at least as valuable as
the cost it imposes on the system will enter and consumers will be offered
the optimal degree of choice.

This begs the question of how those marginal costs are to be determined,
but this is after all already the subject of economic research and in
principle an answer may be approximated.  What is important at this point
in ICANN's organizaion is that this guiding principle be adpoted.


Q7:

Should the requirement (Section III.A.1) that applicants demonstrate
general present or future business capabilities be revised to incorporate
more specifically stated thresholds or characteristics (e.g., levels of
employment or backup procedures)? For which aspects of Section III.A.1?
What should the more specific requirements be? If more specific thresholds
or characteristics should be incorporated for some aspects, should those
thresholds or characteristics be absolute minimum requirements, or should
they be "safe-harbors" that would automatically satisfy certain aspects of
Section III.A.1? (As an example of a "safe harbor," a showing that the
applicant has the equivalent of five full-time employees could be deemed to
satisfy the employee requirement stated in Section III.A.1.j.)

A7:

Thresholds of any kind, unless they are extremely modest and designed only
to eliminate frivolous applications, presuppose that your group can
determine the optimal size, scope, and quality of a registrar's
organization.  But it is precisely because the market can better make this
determination that competition is desireable.

For the reasons outlined in A4 above, successful registrars may well vary
dramtically in size, scope, and quality, just as Internet Service Providers
do where markets are competitive today.  The customer of today's budget ISP
may be the customer of tomorrow's budget registrar, because price may
matter more than, say, the assurance that the registrant will be in
business to service the registration six months later.

Where ICANN can perform a useful service is in providing customers with the
information necessary to choose between providers.  Possible strategies
include the required disclosure of some minimal information (e.g., the
number of employees, the frequency of data escrow, insurance carriage,
etc.) or a standard method for rating registrars according to some
criteria.  If the customer wants to buy a $10 registration from Fred's
Overnight Success rather than spend $100 with Big Registrar, Inc., and does
so cognizant of the risks that Fred will close up shop tomorrow, who are
you or I to deny him the opportunity?

Of course, ICANN may rightly be concerned about the impact of such a
failure on the registration system as a whole.  Potential liability issues
are properly addressed by a modest insurance requirement.  To address other
concerns, a bond might be required that is in proportion to the number of
registrants served.


Q8:

Is it appropriate to have a specified threshold of required commercial
general liability insurance coverage? A requirement of US$500,000 is being
considered. Is that level about right, too high, or too low? Should the
required level vary by the location of the registrar to account for
different liability risks in different parts of the world?

Q9:

Is it appropriate to require that a specified level of liquid capital be
available to the applicant, or is it sufficient to have a more general
requirement such as that in Section III.A.1.i? A specified level of
US$100,000 in liquid capital is being considered. Is that level about
right, too high, or too low? What effect would a requirement at that level
have on the degree of competition? Should the required level vary based on
the country in which the applicant is located? By the type of business
model the applicant follows or proposes?

A8/A9:

It would be inappropriate for requirements to vary by locale.
Registrations should be processed in the most efficient manner possible,
and creating artificial opportunities for arbitrage will be detrimental to
efficiency.  If, for example, the residents of Tonga are best served by a
registrar just down the street, and costs are higher there than in, say,
San Francisco, they will be willing to pay a premium for its services.

However, requirements should, in general, be modest.  For the reasons given
above, a US$100,000 liquid capital requirement might prevent some
registrars with desirable attributes from entering the market, size and
stability being just one attribute among which customers might wish to
choose.  Similarly, the requirement for general liability may be too high.
More appropriate might be a policy that specifically addresses concerns
about the impact of a registrant's failure on ICANN as an organization and
the registration system as a whole.


Thank You,


-- 

Robert Szarka
Managing Partner, Operations
DownCity, LLC
+1 860 823 3000