[Editors Note: This is the second installment in our series looking at the ongoing ICANN reforms. If you haven't already, be sure to read Part 1, Part 2, Part 3, Part 4

[View/print PDF]

The current round of ICANN reforms started in September 2006, when the London School of Economics Public Policy Group released a report assessing the GNSO. To understand the implications of this report one must first know more about the GNSO. While it is not well known to the people involved in Internet governance, the GNSO is the core of ICANN politics and consumes most of the organization’s time and resources.

The GNSO has two main powers. It appoints two members to the Board. And its Council initiates the policy development process for generic top level domain names. (“Generic” is the awkward word used by the ICANN community to denote any top level domain names that are not two-letter country codes. TLDs such as .com, .org, net, .info, and .mobi are considered generic. About 2/3 of the world’s domain name registrations are under so-called generic domains.)

The GNSO Council delegates policy making tasks to subgroups, which develop draft policies. These are opened to public comment and voted up or down by the Council. A policy that achieves a 2/3 majority of the Council is considered a “consensus” policy. Once the GNSO Council approves a policy it is sent to the ICANN Board for approval. A “consensus” GNSO policy must be approved by the Board unless 2/3 of the Board members vote against it.

The GNSO was “multi-stakeholder” long before that term became fashionable. Political scientists would call it a system of corporatist representation, because it divides up the world of domain name politics into distinct kinds of “interests” and assigns a place at the policy making table to each of these recognized interests. But how big a slice of the voting pie do these interests get? The problem of distributing voting power has always been the key to GNSO politics. Let’s take a closer look.

The GNSO is composed of six constituencies. Two of these constituencies are domain name supply industry interests, registrars and registries. They are often called “contracting parties” because their very existence depends on having contracts with ICANN, which can enforce regulations on them through the contracts. Since 2003, registrars and registries have had weighted voting; for every seat on the Council they get two votes.

Three of the other constituencies are basically large business entities: commercial users, trademark holders, and telecommunication companies. There is, finally a Noncommercial users constituency, which has been composed mainly of public interest advocacy groups who take an interest in Internet and domain name policy, but also education and research networks.

Each of these constituencies elects a certain number of representatives to a Council. (Table 1) The GNSO Council is the legislative body in ICANN’s policy making process. It identifies problems that need to be addressed, collect information and accepts public comment on how to address them, formulates policies to fix the problems, votes on them, and, if successful, passes the policy on to the Board and staff for approval and implementation. If the GNSO is not a legislator, then the Board is, and the Council is like a legislative subcommittee.

In theory, the voting distribution of the GNSO Council creates a rough balance between “user” and “supplier” interests, with 12 votes assigned to registries and registrars, and 12 to the users. In theory, the appointment of three members to the Council by the ICANN Nominating Committee adds some independent voices who can break up deadlocks among the Constituencies.

Right from the beginning of this Council in 1998-9, however, there was a glaring imbalance in its structure. Trademark holders, commercial users and telecom firms are deeply overlapping categories. Thus the GNSO Constituency structure gave commercial interests three times as many votes as noncommercial interests. These “three” constituencies were, for all practical purposes, one big Constituency: they met together frequently and voted together 90% of the time, reflecting the fact that they were composed of essentially the same interests – big corporations interested in brand protection.


Council seats

Voting power





Domain name supply




Intellectual property



<![if !supportEmptyParas]> <![endif]>

Trademark defenders

Commercial users



Internet service providers



Noncommercial users



Public interest, nonprofit

Nominating committee







<![if !supportEmptyParas]> <![endif]>

Table 1: The GNSO Council


The LSE report on the GNSO was commissioned by ICANN itself, in line with its commitment to continually assess and improve its performance. After several months of data collection and interviewing, the LSE released its report in September 2006.

The LSE report praised ICANN’s Supporting Organizations because they “make possible the policy development processes which provide the foundations for ICANN's legitimacy as an open and global policy-making body for the Internet.” Nevertheless, the report laid bare some obvious flaws.

The report documented the incredibly small amount of attention paid to GNSO by the ICANN Board, and showed that there were few institutionalized channels for communication between the Board and GNSO. It also showed that decision making in the Council takes an inordinately long amount of time. The report’s researchers wrestled their way through the tangled skein of WHOIS task forces in order to calculate the number of hours spent, and estimated the cost at $7 million in person-hours.

But it was the Constituency structure of the GNSO that came in for the most criticism. “The current pattern of Constituencies is relatively complex and no longer seems well-adapted to the needs of all stakeholders in the rapidly changing Internet community,” the report said. In particular, it noted that participation in the user constituencies was narrow and tended to be dominated by a small core of people: “The Constituencies show a mixed pattern of participation, with relatively high levels of involvement in two Constituencies covering Registries and Registrars, but relatively narrow participation in four others, covering business users, intellectual property, internet service provides and non-commercial users.” “There are some worrying signs,” the Executive summary went on, “of dominance of some constituencies by a small core of people and of low participation rates in policy development work by Constituency members.” The LSE report documented and legitimized longstanding complaints that there was no major distinction between the three business constituencies.

In its recommendations, LSE suggested that GNSO be restructured into three basic groupings: Registration Interests (registries and registrars), Businesses, and Civil Society. The Registration Interests group would merge the registry and registrar Constituencies into one. The Businesses group would merge the trademark, Commercial users and ISP constituency into one. Civil Society would be populated by the kind of organizations that formerly went into the NCUC. The report suggested that this would be a cleaner and more flexible division of the world. The GNSO Council would be reduced in size to 16 from its present 21. And the threshold for “consensus” policy would be raised to 75% of the votes. In another important recommendation, the report told ICANN to financially support the participation of the members of this smaller Council.

There was no way around it: in addition to some sensible but less controversial changes, ICANN’s own hired experts were proposing a major redistribution of power in the GNSO. This was a bold and long overdue proposal. But for progressive reformers the report contained a major disappointment. LSE proposed that Registration Interests and Businesses be given 5 votes on the policy making Council, while Civil Society be given only 3. It also proposed to raise the threshold for passing a policy to 75% Council vote. That voting distribution proposal was intended to give supplier interests and business interests a veto over any policy. If only 4 of their representatives opposed something, it would prevent a 75% majority from forming and thus block any policy. The public interest advocates in civil society, even if they were completely unified, could not exercise such a veto. Civil society, in other words, was considered a second-class citizen in the representational game. Nothing in the factual findings of the report justified this discrimination. The report did not even mount an argument for it.

Defenders of the LSE proposal pointed out that the existence of three “at large” GNSO Council members appointed by the Nominating Committee compensated for this inequality. This aspect of the LSE proposal flagged the existence of two competing approaches to the representation of public interest voices in the GNSO. One approach, represented by the NCUC, allowed public interest advocacy groups and other noncommercial entities to organize themselves into a sector. The other approach relied on a detached, secret Nominating Committee process to choose “good people” in a closed, deliberative process, to represent the general interest on the Council. These two contrasting approaches still are at play in the ICANN reform process. And, as often happens in the complex world of international institutions, the representatives of these two distinct approaches started to view themselves as competitors or enemies.

In our next installment, we explain how the LSE recommendations were used by ICANN’s Board to propose – and begin implementing – the comprehensive reforms we are now experiencing.

Next: Field Guide to the ICANN Reforms, Part 3