Second-guessing the ORG process

[John C Klensin <klensin@xxxxxxx>]

An open letter to the ICANN Board: Second-guessing the ORG
process

These are not comments about any particular proposal, nor an
attempt to second-guess the _results_ of the process.  They are
observations about the nature of the rfp, proposal, and
evaluation process to date and its possible implications, both
for ORG and for what it suggests about ICANN more generally. 

ICANN has developed a reputation, even among its supporters,
for having a talent for

	 * Doing the right thing for the wrong reasons,

	 * Doing the wrong thing for the right reasons, and 

	 * Doing the right thing, and for the right reasons, but
	 presenting it in a way that creates the worst possible
	 public impression about the action.

I want to express my appreciation to the Board and Staff for
offering, in the ORG case, to accept comments at this stage,
possibly avoiding another decision that would appear to fall
into one or more of the above categories.

I had hoped to sit out the ORG process, and have made no public
comments about it in the past.  But the materials posted in the
last week or so cause me to feel compelled to speak up.  In
doing so, I am speaking strictly for myself, as a long-term
observer of the Internet, the DNS and associated processes, and
ICANN.  I belong to, or participate in, many organizations; if
this note coincides with the opinions of any of them, it is a
coincidence only.  And, while I have some biases about the
choice of ORG operator, this note is intended to address
history and process, not that choice.

I see the following issues with the ORG selection process so
far.  I urge ICANN to address them in the interest of the
perception of fairness and credibility.

(1) Competition

One of ICANN's stated fundamental goals is to increase
competition in the area of domain name services.  This raises
the question: what is competition and among what groups?  

The definitions I had always learned of "competition", and how
to accomplish it, involve opening up markets to more players
and lowering costs of, and other barriers to, entry.  It
appears to me that the ICANN review processes are biased, in
several ways (some discussed more below), toward "only
incumbent operators need apply".  Spreading TLDs out among an
oligarchy of existing TLD operators who are interlinked by
investments, closed cross-licensing of technology, partnerships
in other businesses, etc., does not strike me as the same thing
as increasing competition and diversity.  Operators chosen from
that pool may ultimately may be the right choice, but, if there
is any bias in the system, it should be toward introducing
competent new operators (and, ideally, new models of operation
that might be uniquely suited to ORG) who can demonstrate an
ability to do the job, rather than shifting allocations of TLDs
among the incumbent actors.

The ability to do the job should, of course, remain paramount.
A populated, operating, domain like ORG is not a good place to
take risks.  At the same time, translating "ability to do the
job" into "only incumbents need apply" seems to me to be
excessively conservative and that it may unreasonably exclude
possible innovations that might be of benefit to the Internet
community as a whole and to those who are registered in ORG.

(2) The nature of ORG and the criteria as provided by ICANN

As originally defined, ORG was intended for registrations for
"any other domains meeting the second level requirements",
i.e., not fitting into GOV, EDU, COM, or MIL at the time (RFC
960) or, later, "organizations that didn't fit anywhere else.
Some non-government organizations may fit here" (RFC 1591,
section 2). That definition evolved into assumptions about real
organizations and non-profits.  As a result of that evolution,
in the period before commercialization registrations of
entities "not fitting elsewhere" gradually migrated toward COM.
After commercialization, an attitude of "register everything
possible in all possible domains" prevailed, but that is a
separate discussion that has little impact on ORG except
insofar as it complicates today's environment.  It is also
interesting to note that neither RFC 960 or RFC 1591 mentions
the term "non-profit".

ICANN initially shared in the oral tradition in this area,
announcing that part of the goal of reallocating ORG was to
move the domain back toward its non-profit organization roots.
At some point, it became clear that doing so would be
impractical: enough overtly commercial operations had
established a foothold in ORG that trying to get them out would
be met with resistance about depriving them of parts of their
identity, and trying to prohibit new commercial registrations
would raise issues about creating barriers to later
registration entrants.  ICANN then changed its announcements to
remove any emphasis on the principle that ORG should primarily
support non-profits.  But it is interesting to note that doing
so either didn't work, or that at least some of the applicants
made assumptions about the selection process that were at
variance with the published statement: virtually all of the
applicants stress their relationships with, support of, or
membership in, the not-for-profit or public service communities
on the Internet.

Again, I do not question the decision to treat ORG as an open
domain.  While counterarguments may exist, the decision itself
appears reasonable.  But, if that is to be the conclusion, then
significantly favoring non-profit, or non-commercial, interests
in either the evaluation process or the selections seems to me
to be unjustified and unnecessarily discriminatory against the
interest of those whose registrations are equally legitimate in
an open domain.  I believe that, if nothing else, ICANN is
under some obligation to explain the combination of a decision
to treat org as an open-registration domain with the aspects of
the evaluation and selection process that seem to stress
non-profit or non-commercial activities.

As a further consequence of this apparent conflict in
objectives, if the organizational/ non-profit/ non-commercial
aspects of ORG are not relevant, then there is no more basis
for an evaluation by the NCDNC than there is for one by the
business constitutency, or the registrar one, or any other
body.

To highlight this conflict in ICANN's statements about what it
is doing, please note that the final, 20 May, version of the
"Criteria for Assessing Proposals" contains criterion 5
("Inclusion of mechanisms for promoting the registry's
operation in a manner that is responsive to the needs,
concerns, and views of the noncommercial Internet user
community") and 6 ("Level of support for the proposal from .org
registrants ...particularly those actually using .org domain
names for noncommercial purposes"), which would seem to be
irrelevant if there is not a clear intent to operate ORG for
the benefit of those groups, to the disadvantage of commercial
or other organizations.  

A distinction can be made from a narrow, economic, point of
view between organizations (whether non-profit or for-profit)
who 

   * have a goal of deriving increased revenues from an
   increase in the number of registered names, or

   * whose operational models depend on adding many more names
   because those models are not viable at the present scale,
   but might be at much larger ones, and

   * organizations who have neither of these goals or
   requirements.

The first and second types of organizations have distinctly
different motivations from the third; the third type has no
inherent motivation to find ways to register ever-more
second-level domains in the TLD.  This distinction becomes
particularly important if alternate models for populating
domains are to be considered (see section 5).

When evaluating [earlier] new gTLD applications, ICANN made the
degree to which the proposal would promote and build the domain
an explicit criterion.  For ORG, it is not obvious that a goal
of increasing the number of names registered at the second
level is desirable.  Presumably because ICANN chose to not
identify those choices, the proposals do not seem to have been
compared on the basis of the two distinct models outlined
above.  Yet that distinction may, in the long term, be
critically important to the community.

I believe that, if ICANN is going to be fair and unbiased about
how it intended ORG to be handled --and to be perceived that
way-- it should be clear about its criteria and intentions, and
that the various evaluation criteria should reflect the
intended approach and no others.  If a purpose-specific TLD for
generic not-for-profit organizations is desirable, and ICANN
doesn't see how to turn ORG [back] into one, then it should
move toward creation of such a new domain, presumably with
arrangements to migrate appropriate organizations from ORG
under favorable terms, and that ORG should then be handled on a
strictly generic TLD basis.  But it is important to make
objectives clear, and then to solicit proposals and do
evaluations based strictly on those objectives.  The current
situation encourages speculation about hidden agendas and side
communications; those speculations are good for neither ICANN
nor for the broader Internet community.

The confusion I am feeling about this extends into the
evaluation process and the proposals themselves.  The proposals
seem to run the full range from "non-profit organization
proposes to operate ORG on behalf of itself and other members
of that community" to "non-profit organization lends its name
and credibility to an essentially commercial operation, with a
separate board that will support the commercial entity" to
"for-profit organization claims that it can do a better job of
supporting ORG than any non-profit or combination with a
non-profit".  It seems to me that those are very different
models.  Preferences for one or another of them seem to pervade
the evaluation processes, but nothing that would support such a
choice of models appears in any of the stated criteria.  The
Staff and Board should make whatever decisions it concludes, in
their wisdom, that they should make.  But, to the degree
possible, they should make their criteria explicit so that the
community can comment on them and, more important, better
understand and trust the results.

Of course, if the criteria are questionable, or some of them
are not explicit, then evaluations based on those criteria
become suspect.


(3) Technical evaluations

The easy way to do a technical evaluation is to assume that
only those who are already doing a given job, and doing it
well, are qualified to do that job.  That approach has several
flaws if one is trying to, e.g., expand the number of actors in
a particular area.  Those of us who were around in the early
days of the Internet, and involved in discussions related to
the OSI model, cannot help noting that if this "only the
incumbents are qualified" assumption had been applied at that
time, it is likely that only then-main-line telcos would have
been allowed to enter the data-network market.  It that had
occurred, the Internet as we know it today would probably not
exist.

It is also an open question as to whether the [registry]
operation of a large TLD --with or without the complications of
registrar-registry protocols and issues peculiar to the DNS--
is significantly different from that of operating any other
moderately large, multiple-data-source data base and making it
highly available on the Internet and, if it is different, in
what ways.  There is no evidence that either the Gartner team
or the Academic CIOs examined this question or what conclusions
they reached.  If they reached the conclusion that conservatism
requires that only groups that contain incumbent gTLD operators
be considered favorably, then I believe it would be very
helpful to the community, and to trust in ICANN, if they would
say that and explain their reasoning.  It is the appearance of
failure to consider additional alternatives that leads to
mistrust and even assumptions about hidden agendas.  Without
information about internal criteria and how they were arrived
at, it is very difficult to evaluate the evaluations as other
than a "beauty contest".  

In addition, several of the proposals make assertions about the
previous experience and successes of the proposers.  I believe
that many in the technical and user communities would find
these assertions questionable and that ICANN Staff has received
sufficient comments to have a good perspective on those issues
and assertions.  The evaluations would be much more credible if
it were obvious that the evaluators had received staff input,
or conducted verification studies of their own, rather than
taking these vendor assertions at face value.

The choice of the Gartner group is especially relevant in this
regard.  As the staff report mentions, Gartner has a good
reputation for information technology evaluations and
recommendations.  So do their primary peers and competitors,
but I would not suggest that any of those competitors would
have been a better choice.  However, these organizations are
also known to share two properties that are important to this
evaluation: First, their evaluations are no better than the
particular teams that are assigned and the effort that goes
into them.  It is notable that, while the community was given
the names of the people on the Gartner team, it is not supplied
any biographical information about them; they are not
well-known names in the Internet community, but the lack of
qualifying information may be merely a documentation omission.
If it is believed that TLD operations are unique in important
ways, credentials and experience to evaluate capabilities for
such operations are, of course, particularly important.
Second, Gartner (and their peers), are historically comfortable
in dealing with particular types of organizations and
organizational structures.  The recommendations they have made
--and, arguably, the methods they used to score and develop
those recommendations-- are consistent with an hypothesis that
they were less open to other types of organizational structures
and management than might have been appropriate for ICANN.

(4) The General Counsel's report

The materials that ICANN has posted about this evaluation
process, and what I can infer from the transcripts of
discussions in Accra and Bucharest, suggest that the Board is
going to make the decision among these applications after
considering all relevant materials.  With regard to making
those decisions, enlisting staff and outside evaluation support
and reports seems entirely reasonable.  But the General
Counsel's report effectively creates a short list, using
criteria that have not been exposed for public comment and
that, for reasons outlined above, may be dubious.  Only the
proposals "...placed in tier 'A' by at least one technical
team..." were evaluated, and hence the Board lacks a procedural
evaluation on any of the other candidates.

Since there is reason to question both the breath of
evaluations of the technical teams, and the criteria they used,
this may be a narrowing that excessively constrains the Board
(presumably, giving serious consideration in Singapore to a
candidate that was not included in any of the evaluations would
result in a delay in the decision while those evaluations were
being performed.  And that delay might make it difficult to
transfer the domain on the 2003 schedule.  Of course, if the
Board were to make a choice different from those recommended by
the technical teams (or by NCDNHC, for that matter) it would be
necessary for it to explain that choice.  But, if the decision
is to be the Board's, rather than that of a somewhat obscure
set of processes, it should have the freedom to make exactly
that type of decision, freedom that requires that Staff not
create short lists that deprive the Board of critical
information for timely decision-making.

(5) Other criteria that perhaps should have been considered.

Issues with DNS TLD registry management, like most other issues
of Internet operation and protocols, are all about scaling.
Historically, the DNS has an established method for dealing
with scaling, which is deep hierarchy.  From a _technical_
policy standpoint, ICANN should be figuring out how to push
back on every single new DNS registration by organizations or
enterprises that have at least one already.  From a
business/political policy standpoint, that is impossible and it
would be silly to try.  But, if someone comes along with a plan
for ORG that might encourage deeper hierarchy, and a business
model that does not benefit from registration of ever-more
names at the second level, ICANN should be looking at it _very_
seriously.


Thank you for your consideration.

   John C Klensin