Vertical Separation

[Amadeu Abril i Abril <amadeu.abril@xxxxxxxxxxx>]

CORE has been a strong proponent of vertical separation between registry and 
registrar functions since before the existence of ICANN. We have even 
voluntarily implemented such functional separation for the TLDs where we 
provide backend services (not acting for .museum, .cat and .aero at the time we 
were its registry service providers). We have encouraged ICANN to establish 
such rules. We therefore welcome the correct step in providing clear rules in 
this direction. Nevertheless, we consider that ICANN has greatly missed the 
reasonable point in the current language. 

To be clear, we propose the following structure:

* A general rule limiting cross ownership (and control) between registries and 
registrars. But 2%, as written in the proposal, is well beyond reason. There is 
not a single rational point we can establish with certitude, and, given the 
lack of experience in this market with sudden addition of a large number of 
TLDs, we submit that a figure well below what represents “control” in 
traditional antitrust analysis makes sense (and could be raised if experience 
shows the fears proved exaggerated). 15% could be such a figure, as it  allows 
for many workable, non-controlling situations, fails below usual control 
thresholds, and, finally, is present in all current 17 TLD Registry Agreements 
(five also contain additional provisions, but the 15% cross ownership limit is 
present in them all). Therefore, up to 16% ownership/control, per se legality

* As CORE experience with .aero, .museum and .cat proves, many of the harms we 
associate with cross-ownership between registries and registrars (unfair 
competiton regarding other registrars; frontrunning; undue access to 
registrants’ data...) are prevented or greatly mitigated if the 
registry/registry service provider does not act as registrar for the same TLD, 
even if it does so for other, unrelated TLDs. But some risks remain, inclduing 
the ability to achieve some anticompetitive effects through corss-marketing and 
toher tools.But this is only possible if the markets for the different TLDs are 
really connnected, and the players have a certain size that allows them to 
really use such tools. This is why we propose a prseumtive acceptance of 
greater than 15% cross-ownership (up to 100%( provided that the entity/group 
does not act as both registry and registrar/reseller under the conditions that 
they have relatively low market relevance (well below market power standards).

* Finally we can see cases where there might be need for an exception to the 
principle, allowing a registry to act as a registrar for its own TLD, and we 
would propose a mechanism and guidelines for such exceptions (in some cases not 
just for the vertical separation sule, but also for the need to use 
ICANN-accredited registrars).

More details and proposals should emerge from the current Vertical Integration 
Working Group (where we will contribute more detailed propsoals), but in the 
meantime we urge ICANN to seriously think about the unreasonable and 
overreaching consequences of their current proposal. Not only 2% makes no sense 
at all (better say 0%, at that point) but language as the one copied below 
should be either aligned with the principles stated or simply removed:

“Further, applications where the applicant has engaged an ICANN-accredited 
registrar, reseller, or any other form of distributor or any of their 
Affiliates (or any person or entity acting on their behalf) to provide any 
registry services for the TLD will not be approved.

Why can registries own up to 2% of a registrar, but independnet registry 
service providers, having much less influence, let alone power, on TLD 
policies, cannot be present at all, when the corss control here is 0%, and, as 
it is the current case with CORE’s services, it does not act as a registrar for 
the same TLD? This amounts to disqualifying a whole class of actors because of 
their nature, with no rlelation to their acts, actual or potential. We cannot 
support such an approach.

Amadeu Abril i Abril
CORE Internet Council of Registrars