The clever ICANN board

I’ve been forced to rethink my initially dismissive assessment of the ICANN board’s Nairobi resolution on the separation of domain name registries and registrars. As you may recall, I called the resolution “needlessly biased and poorly worded;” and while I recognized that it was “an attempt to clarify things,” I asserted that it “probably did the opposite.” Strange as it may seem, it is possible for both of those accusations to be true and for the resolution to be a stroke of political brilliance in moving toward the formation of a new policy regarding registry-registrar separation.

First, some background for newbies. As the regulator of the domain name industry, ICANN’s contracts impose a structural separation between registries (wholesale operators of top level domains) and registrars (retail providers of the registration service for a particular domain). A domain name registry (say, .com or .mobi) cannot sell registrations under its domain to customers directly; it has to go through a registrar. A registry is supposed to offer all registrars equivalent and nondiscriminatory access to the names available under its TLD. A host of competing registrars with equal access to all TLDs was supposed to foster competition and prevent consumer lock-in, and in the large, established domains such as .com, certainly succeeded in that.

In order to enforce this separation, ICANN has imposed some cross-ownership limits on registries. But when it comes to ownership, the separation policy has been inconsistent and variable over the years. While it prevented registries from participating in the registrar business, registrars were allowed to jointly own registries. While some limits on cross-ownership were applied, they varied across different registries. Back-end registry services (services that don’t own the TLD assignment but provide the basic registry functions for the company that does) seemed to be exempt from the policies altogether.

This messy set of practices, developed through accretion and bilateral negotiations between ICANN staff and registry operators, posed a problem for the new gTLD program. What policy would apply to new applicants?

This triggered an intense, behind-the-scenes debate; and as is so often the case with ICANN, the real policy bargaining and debate took place not within ICANN’s official policy making organ, the GNSO, but between lobbyists and the staff. The staff commissioned studies and reports from economists. Eventually, as part of the Draft Applicant Guidebook, the staff proposed to liberalize cross-ownership restraints in a way that made some registrars happy but some registries very unhappy. Rancorous debate ensued prior to and after the Seoul meeting. This also triggered a debate over whether what the staff was doing was “implementation” of a policy developed and ratified by the GNSO, or whether it was actually the staff making policy. (The Noncommercial Stakeholder Group, apparently the only group interested enough in bottom up policy development to do something about this, took the lead in initiating a policy development process in the GNSO).

Initially, it was my position that the board did not need to pass a resolution on this issue at all. It simply had to wait for the GNSO to develop a policy and then ask the staff to implement it in a new version of the draft applicant guidelines. Instead, it ruled that “within the context of the new gTLD process, there will be strict separation of entities offering registry services and those acting as registrars. No co-ownership will be allowed.”  From my perspective, the board’s intervention was bizarre. This was not a reversion to the status quo ante, it was the assertion of a completely new policy – no co-ownership at all. This policy had no support from the expert economists’ reports, had not been developed or ratified by the community, and constituted a substantial deviation from the status quo and thus did not conform to what anyone expected. Indeed, a set of questions posed to the board by a lobbyist for the registries shows just how profoundly the board ruling threw everyone off balance (see appendix below).

But in throwing everyone off balance, the board may have done exactly what was needed! This is what I did not see in my first blog post. The board seems to have grasped the fact that the GNSO has strong tendencies toward gridlock, and the only thing that would force all parties to come to an agreement would be a baseline position that is unacceptable to all of them. In game theory, this is called the reversion point – the solution that you are stuck with if you can’t come to an agreement. Before the board resolution, the two available reversion points – the staff proposal or the status quo – were favorable to some interests and hostile to others. The board declined to choose either of them and has now moved the reversion point to a place that is downright uncomfortable, if not really painful, to both registries and registrars.

Since that decision there has been a flood of volunteers into the Working Group on “vertical integration.” This suggests that everything is up for grabs and we are indeed starting from scratch. (By the way, “vertical integration” is really a misnomer as the label for this WG. Vertical integration is only one of the issues being discussed; cross-ownership is another, distinct issue – the real overarching issue is registry-registrar structural separation).


Jeff Neuman’s (Neustar) questions to the board following its resolution:

1. The resolution states “there will be strict separation of entities offering registry services and those acting as registrars”

A.  Is that meant to cover just front end registries (e.g., those under direct contract with ICANN), or back-end as well (eg., those subcontractors of registry operators under contract with ICANN).

B.  Would entities offering “registry services”, include those that may be providing a portion, but not all of the registry services?  For example, if an entity is simply providing DNS services for a new TLD, but not the SRS functions, would they be considered under the Board resolution to be “offering registry services”?  If yes, where in your opinion is the line of demarcation?

2. With respect to the phrase “…strict separation of entities offering registry services and those as registrars”:

A. Does this apply to just the new TLD for which a registrar applies to perform the registry function or does that mean if you are a registry for one TLD, you cannot be a registrar with respect to any other gTLD (new or existing)?

B. Does the term “registrars” apply to resellers of registrars as well or literally just registrars

3. The resolution uses the phrase “No cross ownership will be allowed.”

a)  Does this apply to just legal ownership?

b)  What about vertical integration through other means direct or indirect.  In other words, some have advocated that restrictions should apply to those that exercise some form of control over the registry (i.e., by contract or ownership), while others have argued that it should just apply with respect to legal ownership?  Which position has the Board decided to take?

c)  Does the cross ownership requirements apply to affiliates or just the entity under contract.  In other words, if you have Parent Company A that owns Subsidiary B and Subsidiary C.  If Subsidiary B is a registrar and Subsidiary C is a registry, would the cross ownership rule apply?  Will this be allowed in that the Registrar here does not own any portion of the registry, nor does the registry here own any portion of the Registrar despite the fact that they share a common parent.  

4. When the Board states, “no cross ownership”:

A. Does that literally mean 0% ownership as opposed to the 15% used today in the existing agreements?

B. If it literally means 0% ownership, what about those companies that are public where shares are traded on the open market.  Could an entity be prohibited from serving as a registry if a de minimus amount of shares are purchased by a registrar?

C. If it does mean no ownership, will ICANN afford entities a transition period to divest registrar shareholders if a registry elects to do so?  In other words, would the rule be that prior to launching there can be no cross ownership, or is it prior to applying for a new gTLD.


  1. Anonymous

    The decision is what the English call “too clever by half.” The Board should be making clear decisions that are easily understood and communicated, not making obscure chess moves that takes weeks for even seasoned observers (like Milton) to untangle.

  2. Anonymous

    I agree that this is a clever move by the board as it potentially excludes most of the major 'insiders' from the game and will (I hope) force the GNSO to come up with a workable policy very quickly. The Chair of the Working Group has some busy days ahead…