As we enter an era of hundreds or even thousands of new top level domains, how will the industry evolve? The ICANN Working Group on vertical integration was one attempt to answer that question. In the course of its lively and extensive debates, the policy setting group confronted some fundamental issues about the organization of the domain name industry. Back in March the ICANN Board threatened to impose a draconian ban on any and all forms of vertical integration and cross ownership between registries and registrars if the group failed to reach consensus on a new policy. The working group (WG) will complete its work this week.
First, the bad news: the WG was not able to agree on a single, comprehensive new policy. It did, however, manage to reach consensus on one thing: that the Board’s threatened ban on cross ownership and vertical integration was not desirable. About five distinct reform proposals were developed. The WG conducted a poll about how much support there was in the 50+ person group for each proposal. In that poll, the Board’s Nairobi resolution, and its proposed implementation in version 4 of the draft applicant guidebook (DAGv4), was voted on along with the other proposals. The Nairobi resolution received not a single vote in its favor. While a significant minority (29%) said they could “live with” DAGv4, a two-thirds majority expressed outright opposition to it.
The poll results provide a fascinating glimpse into the political economy of the domain name registration business. Incumbents (Afilias, GoDaddy, some members of CORE) favored a continuation of the status quo. Their proposal, known by the acronym RACK, would force all new top level domain registries to use the same business model that is being used now. This would mean a functional and technical separation of domain name registrars and registries, equal access for registrars across all TLDs, and 15% limitations on cross-ownership. No exceptions would be allowed, not even for .brand TLDs. The RACK+ proposal had a small but significant core of supporters, most of them on the payroll of Afilias in one way or another. Eleven WG members (27%) favored it and another 4 WG members said they could “live with it,” meaning that the RACK proposal was acceptable to 37% of the group.
At the other end of the spectrum, a proposal that became known as “Free Trade,” initally proposed by At Large member Siva Muthusamy, would eliminate cross-ownership restrictions altogether while retaining registry-registrar separation. It would even allow registries to own a registrar that distributes its own TLD, although it would still have to offer equal access to other registrars, providing a substantial liberalization of the current ICANN regulations regarding registry-registrar relations. The Free Trade proposal garnered more support than RACK+, with 16 WG members (39%) favoring it, and another 4 WG members saying they could “live with it.” Thus, the overall acceptability rating of Free Trade was about half (49%) of the WG.
The JN2+ Proposal (name after its proponents Jeff Neuman and Jon Nevett) proposed to restrict Registry Operators and their affiliates from distributing names within their own TLD, but otherwise liberalized cross-ownership. For example, it would allow Neustar to own a registrar and enter the registrar business, but its registrar could not sell .biz or any other TLD controlled by Neustar. Like RACK+, JN2 contains limits cross-ownership among affailiated registrars and registries to 15% or less. Importantly, it allowed exceptions for single registrant TLDs, community TLDs and Orphan TLDs. For the first 18 months, restrictions apply towards back-end registry service providers (RSPs) that control policies, pricing or selection of registrars and resellers affiliated with the Registry Operator or RSP. After 18 months, these suppliers would be allowed to petition ICANN for a relaxation of those restrictions depending on a number of factors. The JN2 proposal was thus significantly more liberal than the RACK+ proposal. It also attracted stronger support. It was second in the number of supporters but more importantly, also had a large number of WG members who said they could “live with it.” JN2 thus had the highest “acceptability” ranking of all the proposals, with 25 WG members, or 61% of those polled, saying they favored it or could live with it.
The Competition Authority Model (CAM) was based in part on the economist reports (Salop and Wright) commissioned by ICANN. It would have allowed any and every business model, including full vertical integration or 100% cross-ownership, but these innovations would be reviewed by a special panel on the basis of market power and consumer protection implications, and referred to national antitrust and consumer protection agencies if issues arose. This proposal was the first choice of very few WG members (2). Its chief supporters were the two noncommercial WG members (including me) and Michael Palage. But another 12 WG members said they could live with it.
Essentially, CAM defined an exceptions process. Seemingly unpopular, as the WG confronted the possibility of the DAGv4 option being imposed a CAM-like exceptions process began to look more and more acceptable to a larger number of members. Indeed, when the question of “exceptions” to the general rule of no vertical integration was posed in the abstract, it obtained overwhelming support, nearing consensus. Twelve (12) WG members favored an exceptions process, 16 said they could live with it, meaning that over two-thirds of the polled members favored some kind of exceptions process.
The problem lay in the details regarding what would qualify for an exception. There was strong support among commercial and noncommercial users for what became known as a “Single Registrant, Single User” top level domain. This would include the .brand TLD, in which a single company or organization ran its own top level domain for its own employees and departments. There was also strong support for an exception for so-called “orphan” top level domains –TLDs that were ignored by registrars and failed to gain distribution channels. WG members favored letting these small TLDs vertically integrate or own their own registrar and distribute their own domain, though they differed on the terms and conditions. Another exception that had strong support was the linguistically oriented domain. Fears that TLDs oriented toward smaller linguistic groups might not find registrars capable of selling registrations using the language of the targeted community led many to support allowing such domains to vertically integrate.
What did all this work add up to? For the Board, the bottom line on vertical integration is clearer than it might seem. Despite the failure to reach consensus on a single alternate policy, the vertical integration working group has succeeded in conveying the message that there is consensus against the complete ban on cross-ownership and vertical integration proposed in DAGv4. Moreover, although vested interest groups prevented the working group from agreeing on policy specifics, the more liberal options had more support than the more restrictive, status quo-oriented ones, and almost everyone agrees that there should be room for exceptions. The surprising level of support for the free trade option; the intense interest in the User House in single registrant, single user TLDs; the widespread agreement on the need for permitting small linguistic and community groups to run their own TLDs without becoming ensnared in a business model designed for .com – all these signs and more point toward a more open market structure in the years ahead. How quickly the slow and politicized ICANN Board can get us there remains to be seen, of course.
Proposals can be found here: https://st.icann.org/vert-integration-pdp/index.cgi?vertical_integration_pdp
Ranked by # supporters
1 Free Trade 16 39%
2 JN2 12 29%
3 RACK+ 11 27%
4 CAM3 2 5%
5 DAGv4 0 0%
Ranked by acceptability
1. JN2 25 61%
2. Free Trade 20 49%
3. RACK+ 15 41%
4. CAM3 14 37%
5. DAGv4 11 29%
Ranked by strength of opposition
1. DAGv4 27
2. CAM3 24
3. RACK+ 23
4. Free Trade 20
5. JN2 15