Since Kent invokes my comments during the conference call, I feel compelled to
respond.
The following are my own views only.
They are premised on the assumption that we should be considering the merits of
the drafts, not claiming levels of support from organizations that have various
large numbers of members and customers who have never heard of these issues, much
less debating the "lineage" of drafts that have all obviously evolved from years
of collective thinking about the issues.
The following "Paris FAQ" and "Critique of the BMW draft" are meant as a first
cut at articulating key differences.
drj
--------
An Analysis of the Paris Draft proposal for establishment of a Domain Name
Supporting Organization within ICANN
Why does the Paris Draft call for a DNSO to be an integral part of ICANN?
The DNSO is an advisory body that will be the most significant source of
proposed ICANN policies regarding the domain name system. It does not need and
should not have a role in fund-raising or an elaborate staff. If it were a
separate organization, it would require a separate board and elaborate fiscal
controls, and would likely develop interests and goals conflicting with those of
ICANN. No contractual mechanism could provide adequate assurances that its
procedures would comply with the safeguards established in ICANN's Articles and
ByLaws, because it is not realistic to expect that ICANN could threaten to
terminate a contract with its primary advisory body. There is no assurance that
the needed parties would join a separate DNSO or that a separate DNSO would fully
and fairly represent all of those who demand and need to provide input.
Deliberations within the DNSO will be less vulnerable to challenge if they take
place under the umbrella of governmental endorsements of the ICANN structure. In
short, establishment of the DNSO by means of an amendment to ICANN's ByLaws
provides the simplest route to a structure that will serve the goals and abide by
the principles long debated in the course of creation of the ICANN structure.
Why does the Paris Draft provide for self-organizing non-overlapping
constituencies to select an administrative/facilitative (not a policy-adopting)
Names Council?
The Paris Draft establishes a Names Council that administers and facilitates the
development of broad-based consensus. If the Names Council were defined as the
decision-making body, with a limited number of seats, then interested parties
would seek disproportionate representation in selecting the occupants of those
seats. No pre-defined list of constituencies can adjust, flexibly, over time to
reflect this dynamic medium. In contrast, selection of a steering committee by
constituencies representing a minimum percentage (5%) of an open membership will
assure effective leadership over a long period of time. By providing for
non-overlapping constituencies, and allocating membership to both individuals and
organizations, the Paris Draft makes sure all voices will be heard and given
equal weight in determining whether there is a real consensus in the General
Assembly. Constituencies will behave more like political parties, and less like
"elected representatives" making top down decisions. This will ensure that those
"steering" the research and review processes reach out to include all interested
parties and are not captured by those who can "qualify" as electors for the most
seats. Registries (and, potentially, registrars) are assured a role as a
constituency without regard to the five percent rule because they would not
likely be able to meet that test as the membership grows in size and because
their participation is vital to assure that proposed policies can be implemented
by means of binding contracts between ICANN and the registries.
Why does the Paris Draft provide for an implementation preview by registries?
ICANN has no governmental power to impose policies on registries against their
will. If it is to adopt effective, equitable policies, it must enter into
contracts with the registries. The implementation preview is designed to
establish a basis for contracts that bind the registries to implement policies
with which they may individually disagree, provided that most other registries
are prepared to support and implement those policies. This provision will make
ICANN a standard setting body with some teeth, but also prevent the embarrassment
and futility of policy "decrees" that cannot be broadly implemented. The
implementation preview does not apply to policies, such as adding new TLDs to the
ICANN root, that do not need to be implemented by the registries. It is not a
"veto" power by a particular constituency -- it doesn't prevent communication to
the ICANN Board of the fact (if such were the case) that there is support among
other constituencies for a policy that the registries oppose. It simply
recognizes the fact that ICANN is not a governmental body and must, to be
effective, implement its proposed policies by contract. All open TLDs should be
equally subject to consensus policies, if there really is consensus for policies
that change the business practices of registries or contractual relationships of
registries with third parties. Most registries will not agree to contracts that
require compliance with any and all policies ICANN may decide to impose. Thus,
the best mechanism for assuring enforceability of policies developed by the DNSO
would be for ICANN promptly to seek to enter into contracts with registries that
commit them to implement policies that both have the support of a broad consensus
of DNSO members and that have passed the suggested implementation preview
procedure.
Will the proposed DNSO structure be able to take effective action?
The major argument for a "strong" Names Council elected by specified
"constituencies" is that this will create a small group of "representatives" who
can take prompt action. In contrast, the research and ratification processes
defined in the Paris Draft tends to assure that proposals will not be forwarded
without strong and widespread support among a large group of interested DNSO
members. Any evaluation of these competing models depends on one's view of the
consequences of inaction and, alternatively, of precipitate action. If a broad
consensus and registry agreement are not present, rapid action by a Name Council
(or ICANN) could well be destabilizing. The default condition, now, for the net
is that diverse local players with a substantial stake in growing their
businesses will take actions that benefit their customers and users. With the
possible exception of artificial barriers to the addition of new TLDs to the
root, which are imposed by the US Government, not the registries, there is no
barrier to effective decentralized action to establish good dns policy.
Registries and registrars establish sound policies every day and compete with one
another for customers. And, with the planned opening of the competitive
registration system for .com, .net and .org, there is substantial reason to
believe there will be even more effective competitive forces to constrain any
suboptimal decisions by registries and registrars. In contrast, a small group of
"representatives" that purported to speak for all stakeholders and decided to
"adopt" policies applicable to all registries, registrars and registrants,
without their expressed support, could produce substantial and effective
opposition. In short, there is little need for rapid "regulatory" action by a
small, closed group. And those decisions that are taken on the basis of
widespread membership support (by means of the General Assembly ratification and
the implementation preview established in the Paris Draft) will be more likely to
be effective.
How does the Paris Draft prevent capture or disruption by minority interests?
An open membership of all those who choose to join ICANN cannot be captured
(unless the ICANN were to establish barriers to entry in its own membership, a
course of action contrary to the requirements of its Articles and ByLaws and to
the White Paper). The Paris Draft deliberately ties the two memberships together,
both to avoid complexity and to encourage ICANN to adopt an open membership
model. The Name Council cannot be captured because self-defining constituencies,
meeting objective criteria and obtaining support from a minimum percentage of the
DNSO membership can elect their share of seats. Even if a Names Council were to
show some bias, the Paris Draft structure also prevents capture by requiring
ratification by two thirds of the open General Assembly before a proposed policy
may be forwarded to ICANN -- and by requiring transmission of the full research
record, including expressions of dissent, to the ICANN Board. The research
process is required to be open for comments and participation by a wide range of
interested parties. If the Names Council were to attempt to prevent the
commencement of consideration of a particular issue, then the process may be
started either by petition of 5% of the General Assembly members or by only three
of the potential 21 Names Council members. Fair Hearing processes are designed to
give every disaffected party a forum to present its case -- but procedures can be
developed for summary disposition of frivolous cases. Under the Paris Draft, the
role of the Names Council is to seek consensus among impacted parties -- and that
inherently requires that they act to prevent capture and avoid disruption by
minorities.
Who supports the Paris Draft and where did it come from?
The Paris Draft derives from multiple prior efforts of a very large number of
parties involved in the discussion of possible DNSO structures. It was drafted by
a group involving representatives of the ORSC, AIP and registries representing
the vast majority of registrations from around the world. Its language includes
suggestions that have come from business interests, trademark interests,
registrars, end user groups, and scholars. Sponsoring organizations are now
soliciting additional endorsements. Ultimately, this draft belongs to each and
every participant in ICANN and the resulting DNSO. Everyone interested in domain
name policy is, or ought to be, eligible to become a member of the DNSO. Since
DNSO is an advisory body of ICANN, only the ICANN Board can take the final action
(amendment of the ByLaws) to make it a reality. It is a misnomer to characterize
this as a process involving "applications" to "become the DNSO" -- because no
effective and safe DNSO could be established other than as a functioning subpart
of the ICANN membership. It is clear that the structure of the DNSO raises
uncomfortable questions concerning the membership of ICANN itself (will that be
open or closed?, will it allow individual memberships?, will it be conditioned on
payment of unreasonable and inequitable fees?), about the relationship of ICANN
to registries and registrars (will ICANN policies apply to all open TLDs?, will
all registries agree to implement the policies that evolve from the ICANN
process?), about the relationship between ICANN and governments (will governments
allow ccTLD registries to enter into a contract with ICANN?, will governments
allow ccTLD registries to participate in the DNSO process?), and about the source
and nature of ICANN's own prerogatives and legitimacy. If the DNSO is to succeed,
those questions must be answered to the satisfaction of all who wish to
participate in the evolution of -- and who must implement -- domain name policy.
The fact that substantial numbers of registries have supported the Paris Draft is
a good sign that its structures can lead to policies that can be implemented. The
fact that it has an open membership, not carved up into artificial or inflexible
constituencies, deprives any holdout opponents of the argument that they are not
"fairly" represented in its processes. The fact that it calls for a facilitative
Names Council, and for ratification of consensus proposals by the General
Assembly, deprives any opponents of an argument that the policies it forwards
don't represent a true consensus of the interested community. The fact that it
requires an implementation preview lays the groundwork for broad and fair
application of the policies that are developed. In short, while the Paris Draft
represents only a process for potential evolution of new dns policies, it holds
out the hope that the end product of that process will be widely supported and
implementable -- avoiding destabilizing "top down" pronouncements by
self-appointed regulators and embodying the best aspects of the "rough consensus
and running code" methodology that has made the internet so successful to date.
---------
February 13, 1999
A critique of the BMW draft
By: David R. Johnson
Overall, the BMW draft differs from the Paris draft by (1) rejecting an open
membership for the DNSO, (2) rejecting the need to assure that proposed policies
can be implemented and the opportunity to create the basis for contracts that
require all registries, despite their individual opposition, to accept policies
supported by 3/4ths of the registries under contract with ICANN, (by means of
the implementation preview), (3) vesting decision-making power regarding proposed
policies in the Names Council rather than a General Assembly of DNSO members, (4)
selecting the Names Council and ICANN Board members representing the DNSO through
pre-defined constituencies (rather than letting DNSO members decide which
constituency best represents their interests, (5) allowing particular members to
join multiple constituencies (thereby granting large companies with diverse
interests disproportionate say in the nomination process).
The most fundamental differences between the BMW draft and the Paris draft
appears to stem from differing views regarding the role of the DNSO and the
consequences of any lack of broad consensus on particular proposed policies.
Under the BMW approach, the absence of consensus is viewed as a problem that
should be overcome by executive action of constituency representatives. In
contrast, the Paris draft seeks an accurate reading of the extent of consensus
among a broad membership -- and it contemplates that, in the absence of
consensus, policies and practices of the dns system should be made by diverse
local actors, all of whom have a stake in adopting policies that best serve their
members and customers. If the DNSO is viewed as an "expert" body, as the BMW
draft appears to assume, then the lack of widespread involvement by those
impacted by policies (such as domain name holders) is not seen as a problem. In
contrast, if one basic role of the DNSO is to give ICANN an accurate
understanding of whether those impacted by policies will view them as acceptable,
either on the merits or because they have had an effective means of participating
in the policy development, then open membership and General Assembly voting on
key questions become much more important to the DNSO's success. Are we creating a
top down "expert" regulatory body or a "bottom up" standard setting body for a
diverse industry and community? That is the question posed by the differences in
these drafts.
The drafts differ in their approaches to achieving the flexibility necessary for
this dynamic medium. The BMW proposal seeks flexibility by reciting general
principles and leaving many details regarding process to be worked out later by
"self-organizing" constituencies and the Names Council they select. In contrast,
the Paris Draft spells out specific procedures for policy formulation and allows
flexibility by means of encouraging members to form and join constituencies
representing at least some minimum percentage of the membership. (In the Paris
draft, each member may join only one constituency, so multiple inputs by large
companies are avoided and there is reasonable assurance that most members will be
able and will choose to play a role in selecting the Names Council.) By selecting
a particular set of named constituencies, the BMW draft seeks to balance
interests represented on an executive Names Council in a manner that, even with
our current understanding of the array of impacted interests, leaves out or
relegates to minority status many types of participants who have a stake in the
policies ICANN will make. A flexible constituency model will, instead, both allow
and require each impacted party to choose which group best represents its
perspective.
The following comments relate to particular provisions of the BMW draft:
Introduction. "The undersigned hereby commit to fulfill the undertakings and
policies set forth in this application"
The mechanism for assuring that selected DNSO processes are implemented should
be the adoption of provisions in ICANN's Bylaws. No set of private parties can
bind the diverse constituencies whose voices need to be heard in determining
whether there is broad consensus for proposed policies that will impact
participants in the domain name system. The BMW draft is ambivalent regarding
whether the DNSO should be a separate organization with members, funding and
staff independent from those of ICANN itself. The ICANN board should resolve this
ambiguity by forming the DNSO as an advisory body within ICANN, subject to the
many protective provisions of the ICANN Articles and Bylaws. That will eliminate
the need for a separate group, which may not speak for the enumerated
constituencies, and which surely will not speak for all impacted constituencies,
to purport to speak for an as yet unformed membership.
I.A. "The DNSO membership shall consist of constituencies�"
This provision prevents any impacted stakeholder from joining the DNSO other than
through a constituency. It will prevent flexible adaptation of the Names Council
and effectively eliminates the voice of minority interests within the named
constituencies. What will be the remedy for a company or individual dissatisfied
with "its" constituency? with the association or organization that purports to
speak for that party? The Paris Draft provides a safety valve for those who don't
agree with the actions of their "representatives" and care enough to represent
themselves.
I.A. "Each constituency shall self-organize and determine its own criteria for
membership."
There is no clear need for criteria for membership in a constituency other than
membership in the DNSO (which should be a subset of membership in ICANN) and the
willingness to exercise one's vote for a representative on the Names Council
through that particular constituency. The BMW does not and cannot explain what
would happen if the named constituencies do not "self-organize" or if they face
substantial internal disagreements. The Paris draft automatically handles
intra-constituency conflict by allowing groups of DNSO members to reallign
themselves over time. The BMW proposal creates a risk that criteria developed for
membership will be exclusionary or unreasonable -- whereas the Paris Draft
eliminates such criteria and self-corrects if the "leadership" of a particular
constituency fail to represent the views of substantial numbers of constituency
members.
I.B. Membership in the Constituencies. "The DNSO and the Names Council will
develop fair and open procedures for the creation, deletion and merger of
constituencies; and adjustment of the representation of constituencies on the
Names Council."
The BMW draft does not provide concrete procedures for members of the DNSO to
form new constituencies. It is unreasonable to expect the initial group to
welcome the addition of new groups that will dilute their voting power in a Names
Council that is itself the source of a declaration as to whether a particular
proposal is supported by "consensus". Adding words that promise that future
procedures will be "fair" and "open" does not build trust, because the details of
constituency formation could instead be selected now and can be tied to the
decisions of members regarding who should represent their interests and serve as
their voice on a Names Council that merely facilitates presentation of proposals
to the membership body. A later provision contemplates the payment of dues to
join a constituency, which will create a further barrier to participation by
small businesses and individuals.
I.B. Membership. "The constituencies are�"
The BMW list of constituencies has many flaws. Businesses are specially called
out, even though this category overlaps with many of the other categories. No
account is taken of potential conflicts between small and large businesses. Even
the non-commercial group excludes individual registrants and, indeed,
non-commercial interests that are not registrants. The registrar group includes
both true registrars (those with a contractual right to write to the TLD zone
files) and resellers, even though their perspectives and interests may differ.
The trademark group excludes those who oppose treating domain names as identical
with trademarks, but provides no other clear "place" for the latter in the
organization. In contrast, the Paris draft leaves selection of constituencies up
to the constituents.
I.C. The Names Council. Initial Names Council provisions.
The BMW draft calls for creation of the first Names Council by means of a
meeting "of all entities that have subscribed to the application in the form
accepted by the Corporation." This ties participation in the DNSO formed by ICANN
to required agreement with a particular proposal. This provision also
contemplates an "assessment of fees for the start-up process" -- creating another
barrier to participation. It is not made clear how the geographic diversity
requirements will be met, either in the initial Names Council or thereafter.
I.D. Applications for Membership in Constituencies.
The BMW requires submission of applications to "the relevant constituency" but
doesn't make clear who will be able to claim to represent such constituency for
purposes of accepting such applications. To whom would a university submit its
application to become a member of the "Non-commercial registrants" group? What if
two groups form? The answer appears to be that those who signed the BMW proposal
will set themselves up, somehow, as the initial constituencies. But the parties
supposedly represented by the constituencies may not agree that the supporters of
the BMW draft should play this role. In contrast, under the Paris draft, each
party impacted by the domain name system merely has to decide (1) whether to
join the DNSO, and (2) which unique group of fellow members, if any, to join with
for purposes of selecting a Names Council representative. Under the Paris draft
approach, those groups that attract support from an adequate number of fellow
members (or who, like registries, play particular roles defined with reference to
direct or indirect contractual relationships with ICANN) will naturally form and
fit into the structure. Even some delay in the formation of some particular
constituencies would not interfere with effective operations, under the Paris
Draft, because the General Assembly of the members has the key decision-making
powers and it is possible for members to participate without being a member of a
constituency.
II.A. Methods for Developing Substantive Internet Policies� "�the Names Council
shall make recommendations�" "A consensus recommendation is one that is supported
by the affirmative vote of two thirds of the members of the Names Council and is
not opposed by the votes of all of the representatives of any two
constituencies."
The BMW draft uses the word "consensus" to refer to the state of voting within
an 18 member board that is deemed to represent the interests of members. But,
aside from the likelihood that the defined constituencies do not fully cover or
adequately balance the diverse interests of the broader membership of impacted
parties, there is no assurance that the votes of these "representatives" will
actually reflect the state of consensus among the members themselves. Various
members will have different views, held with widely differing levels of
intensity, on different questions. There is a simple means of determining the
overall level of support for a proposed policy -- submission of the proposal to
the membership for a vote, as contemplated by the Paris Draft. The Names Council
can play a useful role in framing questions, encouraging research and compromise,
and assessing the likely reactions of the members. It (and ICANN) can assess the
relative weight and importance of any dissenting opinion. But it cannot arrogate
to itself the ultimate decision-making power without running serious risks of
inaccurately reporting the degree to which particular policies, if adopted by
ICANN, would be viewed as wise or legitimate by the many diverse parties who
would be impacted. Only an open membership and General Assembly voting can
provide a process the availability of which will disarm attacks against ICANN
that claim it is a "top down" regulatory body potentially captured by those with
the resources and will to seek disproportionate influence.
II.B. Selecting Nominees to the Board of the Corporation.
By allocating selection of the DNSO representatives to the ICANN Board to the
Names Council, the BMW draft effectively gives the selection power to the group
of pre-selected constituencies that can obtain half of the votes on the Name
Council. That selection may well not represent an outcome that best reflects the
overall views of the DNSO membership, especially if that membership (if not
limited by artificial constituency constraints) were considered to include all
stakeholders in the domain name system.
Miscellaneous.
Various provisions of the BMW draft calling for audited financial statements and
DNSO self-funding will lead to additional layers of expense and erect barriers to
participation by small companies and individuals. The BMW draft ends with
reference to a decision by the "general assembly of the constituencies" regarding
a "fair means of allocating membership dues among its membership". This provision
contemplates the existence of a DNSO membership, but provides no rationale for
restriction to the stated constituencies. If the "general assembly" of all
members can allocate dues, why cannot it determine the state of consensus
regarding substantive policies? Why is it necessary to collect dues at all, when
the role of the DNSO is simply that of an advisory body and the members will
already be expending substantial time and funds in order to participate? Why
should only certain constituencies have a right to assess all members for dues
money that will apparently be spent to subsidize the expenses of Names Council
members that may not support the positions of the members who must pay such dues
assessments?
A final word on the Risks of Inaction and Capture
Proponents of the BMW draft have argued that its provisions are needed to assure
that some small group can take rapid action -- and that the Paris Draft
procedures may lead to inaction. First, it is important to question whether
action should be taken -- and recommendations should be made in the name of the
DNSO community -- in the absence of a broad consensus that would manifest itself
in the vote of a DNSO membership. In the absence of consensus, the domain name
system will continue to operate. And diverse participants will continue to make
decisions that best serve the needs of their customers or constituents. ICANN
will continue to have the ability to seek contractual commitments from
registries. Governments around the world will still have the ability to regulate
those within their jurisdiction. There has been no showing of any need for rapid
action by the DNSO on particular issues. Second, even if there were some issues
as to which some parties are impatient for action, their advocacy should be
channeled into the useful task of convincing others, or formulating policy
proposals that receive widespread support, rather than towards the divisive goal
of seeking to influence a small and not necessarily representative Names Council.
Third, ICANN will itself have power to obtain expert opinion quickly, from
multiple sources, and to take unilateral action to set policies for the root
server operations on matters, such as the addition of new TLDs, that do not
require implementation via alteration of contracts with third parties, in the
event that it considers such issues to present an emergency. Correspondingly,
ICANN does not have and will not obtain the power unilaterally to impose on third
parties with which it contracts (such as registries) policies that it considers
needed on an emergency basis but which such third parties do not necessarily
consider to be "emergency" matters. Finally, the Paris Draft procedures would
allow for rapid submission of consensus provisions when there really is a
widespread consensus among the membership, as there might well be in the case in
which emergency measures were really required. If two-thirds of the members of
the DNSO support a proposal, the Paris Draft would move it forward as a
recommendation, despite the opposition of some hold-outs. In contrast, the BMW
draft allows any small group of individuals constituting half of the membership
of the Names Council to hold up a proposal.
Everyone agrees that capture is to be avoided, but there is disagreement on what
mechanisms best avoid this vice. The BMW proposal creates various clear risks of
capture. Each constituency will, in effect, "capture" the minority voices within
the constituency (preventing their effective influence on policy) and will
capture the entire DNSO away from those who are not admitted to the specified
constituencies. If a few powerful interests gain disproportionate influence on
the nomination of a relatively few Names Council board members, they could defy
the will of the majority of the DNSO members. Just a few Names Council members
can block action. In contrast, the open membership and General Assembly
decision-making contemplated by the Paris Draft would be much more difficult to
capture. No small group could hope to generate the required ratification vote,
because any actions to sign up large numbers of "sham" members would be
transparent and could be just as easily countered by opponents. The main fear of
capture regarding the Paris Draft proposal appears to be that a small group could
artificially assemble the one third vote of the General Assembly necessary to
prevent ratification. But, assuming relatively low barriers to entry as a member
of the DNSO, any such move could be countered by others (representing a more
mainstream view) by recruiting additional participation on their side of the
issues. The research committees would be selected by a Names Council that itself
includes the representatives that the members themselves have selected, an
inherently diverse group. The Paris Draft prohibits formation of constituencies
defined with reference to company membership. Any large group of "sponsored"
members would have to disperse into the natural constituencies of
interested/impacted parties, lowering their proportionate impact. In short, there
is a much greater risk of capture if power is concentrated into an executive
Names Council (which declares "consensus" rather than testing the actual views of
the membership) selected by a finite number of specified constituencies (which
set their own rules for admission and may not collectively be inclined to admit
new members to the club). And there is much greater opportunity for perceived
legitimacy if the DNSO is open to all who care to spend the time and resources to
participate -- and whose participation is direct and consequential.
David Johnson
