This is intended to be a thought provoking overview of some of the issues facing ICANN as it looks at reformation, in light of the reform proposal promulgated by M. Stuart Lynn, President and CEO of ICANN. It is somewhat structured as a F.A.Q. (Frequently Asked Questions) document, although it is best read straight-through as an essay.

President Lynn details his explanation for why reform is needed in his Proposal for Reform. According to him, ICANN's current structure is deficient in three distinct ways:

• Too little participation by critical entities. He considers critical entities to be the ccTLDs, Root Name Server Operators, Address Registries, "Major Users, ISPs, and Backbone Providers," National Governments.
• Too much process. Lynn feels that the emphasis of consensus and as the goal of ICANN and the resultant focus on building the organization's structure on a "consensus" model has come at the expense of the organization's effectiveness.
• Too little funding. Lynn believes that ICANN's lack of funding is the result of a combination of ccTLDs not donating enough to pull "their fair share" (because most are not under any agreements w/ ICANN) and ICANN's inability to raise additional funds from other sources.

As a result, Lynn believes that without reform, ICANN's very mission -- ensuring the Internet's viability and stability -- faces a number of threats. These threats include decreased international interest and participation (without adequate funds, ICANN will remain dependent on the U.S. Government) and compatibility problems arising from the increased use of alternate root name servers. In sum, "an ineffective ICANN virtually invites the fragmentation of the Internet by those with a parochial commercial, cultural, or political interests into zones that cannot reliably communicate with each other," which he sees as being disastrous for the continued growth and proliferation of the Internet. See Proposal for Reform, "Mission Impossible."

ICANN consists of a nineteen-member Board of Directors and three supporting organizations (SOs). Nine members of the Board are elected on an At Large basis. Another nine members are chosen, three apiece, by the three SOs. The final member - the President - is chosen by vote of the rest of the Board. All Board members are chosen to serve three year terms with staggered appointments (the whole Board does not come up for "re-election" at the same time). See About ICANN web page.

The first At Large elections were held in 2000; five At Large members were elected by Internet users at large, one each from five geographic regions (North America, South America, Africa, Europe, and Asia/Pacific Islands/Australia). See At Large Elections Rules. Analysis of these elections revealed much dissatisfaction with the process within the ICANN community due to issues with administration cost, levels of interest, and verifiability of the general Internet public at large. The Final Report of the At-Large Membership Study Committee recommended that rather than have At Large members control half the Board of Directors, that instead an At Large Supporting Organization be created and vested with the ability to appoint members to the board, the same as the existing three. See the report's Executive Summary.

There are three Supporting Organizations, each with a distinct function:

• Domain Names Supporting Organization (DNSO). This organization is charged with advising the ICANN Board on policy issues with respect to the organization of the DNS in a consensus-building fashion. It is operated by a Names Council, in which seven constituencies are recognized: ccTLD registries, commercial entities, gTLD registries, ISPs and connectivity providers, non-commercial domain name holders, registars, and holders of intellectual property. Through the use of Working Groups and Task Forces, it attempts to create policy that enjoys as much consensus as possible. See DNSO Structure.
• Address Supporting Organization (ASO). This organization manages the system of IP addresses. Its membership includes the three organizations that manage the Internet's Internet Protocol (IP) numbers: ARIN (for the Americas), RIPE-NCC (for Europe and surrounding areas), and APNIC (for Asia and the Pacific Islands). ARIN will soon divide its responsibility with a new Latin American RIR (LACNIC).
• Protocol Supporting Organization (PSO). This organization is concerned with setting standards for Internet protocols and other technical standards that allow computers to exchange information. Its membership includes the Internet Engineering Task Force (IETF), the World Wide Web Consortium (W3C), the International Telecommunication Union (ITU), and the European Telecommunications Standards Institute (ETSI). See PSO Information Page.

Lynn's proposal replaces the Board of Directors with a Board of Trustees who would hold ICANN "in the public trust." The Board of Trustees would consist of fifteen Trustees. Five would be appointed by governments (one Trustee per geographic region -- N. America, S. America, Europe, Africa, Asia/Pacific/Australia). Five would be appointed by a Nominating Committee ("NomCom") that would consist both of Trustees and non-Trustees. The remaining five would be Ex Officio: one would be the President and the remaining four would be appointed, one apiece, by four Councils. The Councils are basically today's Supporting Organizations with one exception: Lynn would split the DNSO into two Councils, one for ccTLDs and one for gTLDs. See this chart for a more visual synopsis of Lynn's proposal.

There are two major functional changes this proposal puts forth that are particularly interesting. The first is the elimination of At Large membership in anyting resembling its current form. The second is the markedly increased role of national governments in the administration of ICANN: governments explicitly control five seats, and have influence over the five NomCom seats, since Trustees make up part of NomCom. Lynn sees the intervention of governments -- in the form of a "public-private partnership" -- as necessary to make sure that the national interest of governments in the Internet's stable functioning are preserved. Lynn sees government backing of ICANN as essential for its success. See Lynn Proposal, "A New Public-Private Partnership is Necessary".

Lynn's proposal leaves open the possibility of the world's governments dominating the Board of Trustees, assuming that the five governmental Trustees can exert sufficient influence over the Nominating Committee to sway their appointments to friendly appointees. Were that to occur, a government takeover of the Domain Name System would become quite conceivable: the outcome Lynn finds least desirable (see Introduction to Lynn Proposal, Paragraph 4).

The Berkman Center believes that there is another possibility that should be explored that could salvage the independent, private operation of the Domain Name System. The simplified version is this: through the introduction of several topic-related gTLDs, ICANN could be administrated by duly-appointed representatives of each of the gTLDs in a quasi-Federalist system of governance.

We envision a series of gTLDs that cover a variety of interests (various industries, individuals, non-profit organizations of different types, educational institutions, connectivity providers, governments, etc.) governing themselves. We already have precedent for doing this with our current Restricted gTLDS -- .aero, .museum, .coop, .edu, .gov, and .mil. The notion of self-governance in this fashion is entirely compatible with the libertarian goals and policy orientations that have always guided the development and governance of the Internet.

Next, we envision an ICANN Board that contains representatives elected from the membership of those gTLDs. The most apt analogy would be to a Parliament or other legislative body: each representative represents a particular district. On the Internet, we believe, it is valid to consider gTLDs as "districts" for representative purposes. After all, just as a particular geographic zone is a part of a national government's "space," so too is a gTLD part of the Internet's name space.
5. What are some of the benefits of a Federalist approach to ICANN governance?

First and foremost, it brings ICANN as reasonably close to true democracy as possible. It is pretty much universally agreed that true At Large membership is not administrable. However, the same argument does not apply to gTLD-based representation, as Restrictive gTLDs have a history of success in managing their own affairs. In addition, since gTLDs based on specific categories of Internet domain holders would necessarily include a much smaller piece of the Internet "pie", more robust verification would become practicable.

Second, by empowering the Internet's "major users," it encourages participating in ICANN by a larger body of interests that had previously neglected it. Recall Lynn's second criticism of the current ICANN board; namely, that various "important" interests on the Internet were underrepresented. By giving different types of domain name holders a specific voice in governing how ICANN works, it enables them to influence ICANN policy, thereby adding to its legitimacy in a way that Lynn's own proposal may not.

Third, by creating many differentiated gTLDs, we stop the leakage of brand identity that has plagued the traditional gTLDs. The most obvious case is .ORG which, due to its Unrestricted status, contains such a heterogeneous group of registrants (individuals, non-profit organizations, for-profit corporations, politicians, and more) that Internet users are not able to rely on the goodwill of ".org" when determining whether a particular web site is actually that of a not-commercial entity. By splitting up the name space into multiple, topic-oriented gTLDs, we will give Internet users increased confidence in the identity of the organization whose web site they are accessing. In the long term, this may bring with it the increased benefits of less litigation over domain trademark disputes and an elimination of "cyber-squatting."

a) How can we be sure that the various constituency groups on the Internet can be rationally and cost-effectively categorized into multiple gTLDs?

This question comes up, perhaps most significantly, in the current discussions over the disposition of the .org gTLD. As previously mentioned, .org contains such a heterogeneous mix of registrants that it is virtually impossible to know, by looking at domain name alone, what type of user or organization a particular registrant is. In theory, a gTLD just for non-profit organizations would alleviate this problem. But, who is considered the "non-profit/non-commercial community?"

Governments and individuals have drawn many lines with respect to the question of what type(s) of organizations may be considered non-commercial. Many different types of organizations and individuals can come under the heading of the "non-profit, non-commercial community." In America, the most commonly thought-of notion is a charitable organization like the Red Cross or United Way. Indeed, the United States Government designates several classes as being exempt as part of the Internal Revenue Code. (See I.R.C. section 501(c)) Internationally, standards besides a tax exempt designation may come into play in determining whether an organization is not-for-profit.

The point of this example is not to express a judgment as to how .org, or any other gTLD, ought to be organized. It is simply to point out that in most cases, cost-effective methods of differentiating categories of registrants exist (e.g., requiring educational institutions to submit proof of their status from a governmental body before granting them an .edu domain name). Just how cost effective these screen methods are is a subject for further study.

Lynn's complaint about ICANN being policy-obsessed stems from the built-in desire of ICANN to be extremely consensus oriented. The system that we envision is necessarily majoritarian. We believe that a democratic or quasi-democratic system of ICANN administration is essential for its long term legitimacy and impact. ICANN's current model has shown that operation-by-consensus is not practicable. While a small number of appointed directors may be able to administer ICANN somewhat more efficiently than a democratic process, that does not necessarily lead to the conclusion that a Federalist system of ICANN governance would suffer the same bloat and inaction associated with the current consensus-oriented ICANN.

We believe that ccTLDs belong at the table, whether with seats on the Board or through an advisory panel. However, the increasing globalization of the world dictates that multinational interests (be they corporations, not-for-profits, individuals) require significant independent representation before a truly democratic and representative ICANN. The Internet is too broad and unique to be viewed in strict geographic terms.