Here is a revised statement.


Noncommercial Users Constituency statement on vertical separation of registries and registrars

version 1.1, 26 August 2009

Opening observations. 

Registry-registrar separation was a regulatory response to the dominance of the entire gTLD market by one vertically integrated provider (Network Solutions, Inc., now VeriSign). By separating the retail side of the market (registrars) from the wholesale maintenance of the list of unique registrations (registry), capping the wholesale price of the registry, and giving any number of registrars "equal access" to the opportunity to register available names in the .com, .net and .org domains, the U.S. Commerce Department introduced vigorous retail competition in the domain name market. 

The introduction of new gTLDs raises many questions about this model. It is not clear that new TLDs need to be price-capped, given their competitive disadvantage relative to established domains; it is possible that new TLDs are handicapped by the registry-registrar cross ownership and integration restrictions; it is clear that the separation seems inappropriate for certain kinds of TLDs, such as self-provided TLDs confined to a single organization, or very small nonprofit communities.

Although debate over this issue has been sparked by the introduction of new gTLDs, the policy associated with introducing new TLDs is conceptually distinct from the issue of cross-ownership and vertical integration. One could change the policies regarding cross ownership and vertical integration without introducing new TLDs; one could introduce new TLDs without changing the cross ownership and vertical integration policy. We note that Recommendation 19 of the GNSO policy authorizing the new gTLD process states: "Registries must use only ICANN accredited registrars in registering domain names and may not discriminate among such accredited registrars."

Process issues
1. Where is the GNSO policy?
Vertical separation of registries and registrars is a policy issue - one of the most fundamental policies underlying ICANN's regulation of the domain name industry. And yet this important policy change is being handled as if it were an "implementation" decision that can be inserted into new gTLD contracts. Although ICANN's management has commissioned economist reports on the topic, there has no GNSO process to make a policy change. We fail to see how a policy as important as this can be changed without a GNSO proceeding. We are deeply concerned by what appears to be yet another case of staff-made policy.  

Policy issues
Opponents of vertical integration (e.g., http://registryregistrarseparation.org/ ) have made the following arguments. The NCUC response is inserted below each argument. 

1. A vertically integrated service provider would have an unfair pricing advantage
We see nothing "unfair" about this advantage. If an integrated provider has lower costs, and there is competition, it will benefit consumers to be served by integrated suppliers. If they have lower costs the new, integrated competitors will be able make their new TLDs attractive to consumers by offering them at lower prices. 

2. A vertically integrated service provider would have unfair access to competitors' confidential data. 
This may be detrimental to competing suppliers, but not necessarily to consumers. If the vertically integrated provider made attractive competing offers to consumers, it could benefit them. 
However, we would insist on privacy protections that would not allow registries or independent registrars to spam domain name registrants. We do not want to see the vertically integrated registries make use of personal or private customer data to send unsolicited emails - nor do we want to see independent registrars and registries doing that. This problem may be best addressed through privacy/data protection policies rather than through vertical separation per se. 

3. We should be cautious about changing the policy because once the vertical separation is removed it will be difficult to go back. 
This is correct. That is why we advocate a two-stage process (see below) that goes through the GNSO, which is supposedly the basis of ICANN's policy development process. 

4. Domain name front-running. Registrars and registries have an inherent conflict of interest because a registrar can detect consumer interest in an unregistered name and raise the price or pre-register the name.
This is a valid concern. Front-running of domains is a disruption of the market process (due to asymmetrical information) and can be considered extremely discriminatory and detrimental to consumers. However, we note that problems with front-running have occurred under the existing separation regime.  Registrars have significant capability to detect interest in a name without owning or operating the registry.

Thus, separation of registries and registrars may not be the best way to avoid this problem. It may be possible to eliminate front-running through contractual regulations and better enforcement of existing contractual conditions, just as front-running is regulated in stock and commodities markets. GNSO must determine whether a structural separation between registries and registrars is the best way to address this. 

5. Registrars are intrinsically evil. A paper by a consultant has asserted that vertical integration is a bad policy because registrars have in the past ten years committed a number of abuses, including evasion of UDRP, cyber-squatting, typo-squatting, offering domain name pre-registrations in new, as yet uncreated TLDs, identity theft in second-level domains of the .INFO top level domain, and abuse of the redemption grace period. But these arguments are illogical. The cited abuses took place when the vertical separation policy was strictly enforced. It is unclear how such abuses would be increased, or affected in any way, by the presence or absence of vertical separation. They cannot, therefore, be considered arguments against vertical integration. 

Recommendations of NCUC
a) Issue must be resolved through the GNSO
Our primary recommendation is that the issue of changes to the registry-registrar separation be submitted to the GNSO as a policy matter. The debate over this issue has sparked several economic studies, policy analysis papers, mobilizations for and against by registrars and registries, and comment by users and consumers. It is ridiculous for anyone to contend that this is not a significant change in policy. Policy changes of this magnitude applicable to gTLDs must go through the GNSO; that is required by ICANN bylaws.

b) One thing at a time
The addition of what will certainly be dozens, and possibly hundreds of new top level domains over the next few years puts an enormous burden on ICANN staff, its policy development processes and ICANN's monitoring and enforcement capabilities. We think it unwise to link the addition of new TLDs - which by itself involves enormous policy changes - to a major change in ICANN's approach to market structure and competition policy in the industry. 

c) Support for one of the two CRA recommendations
The Charles Rivers Associates (CRA) report made two very cautious proposals for making exceptions to the separation of registries and registrars. Both, in our opinion, were clearly supported by economic analysis; one of them is justifiable under current rules without a new policy proceeding.

1.	Recommendation 1 was that single organization TLDs (for example, .ibm or .bbc) should be permitted to operate both the registry and the registrar that sells second-level domain name subscriptions. 
Because single-organization TLDs are basically a new phenomenon, we do not consider this to be a major policy change and thus we favor making this exception and incorporating it into the implementation of the new gTLD round. There might be substantial demand for internalizing a major corporation's or organizations' domain names under a single, self-provided TLD. It is not realistic and serves no public interest to force these organizations to use third-party registrars. Indeed, such a policy might compromise the security of these organizations. There are no competition policy issues raised by this change, as long as the organization's use of the TLD is confined to its own internal departments, employees and units.
2.	Second, CRA proposes that a registry may own a registrar so long as the wholly-owned registrar does not sell second-level domain names subscriptions in the TLDs operated by the registrar.
This, in our opinion, is a reasonable recommendation. Nevertheless, it is a policy change (it alters the policy governing the commercial terms and conditions applicable to existing gTLD registries and registrars) and should therefore be part of a new policy proceeding in the GNSO. Thus, action on this should be deferred until the GNSO resolves it.