And more comments on "Revisiting Vertical Separation of Registries and Registrars"
My prior two notes discussed the "single-organization" model which I denote as "CRAI-1". In this note I discuss the "hybrid-integrated" model, which uncreatively I denote as CRAI-2. The starting point for this is the 1999 SRS testbed, which made the registrar function to the COM/NET/ORG registries multi-actor, and the subjsequent sale of the NetSol business unit by VGRS to create a "bright line" between "registrar" and "registry". During subsequent new gTLD contract negociations this "bright line" blurred, as others have pointed out, the "15% rule" has some exceptions. On its face the CRAI-2 model is attractive. It does not attempt to prevent registrars, the most experienced parties in the ICANN stakeholder model, other than registries, and the most likely to create non-symbolic "competition" amongst registries, from operating registries. It prevents preferential dealing between registrar and registry functional elements within a single entity. These are laudable goals. However, and why else would I be commenting if there weren't a however, there are potential pitfalls not addressed by the CRAI proposal. We have no "price controls", other than the price cap for some registries. There are registrars that sell capped inventory above that price point, and registrars that sell capped inventory below that price point. We don't have a rule of the form "domains shall not be offered as part of a complex bait and switch scheme, and offered below cost while tied to other goods and services that are overpriced." This is a latent defect. If MicroSoft ever decides to bundle single domain name into its licensed single user product offerings, "for free", and blocks of domain names into its multi-user and site licensed product offerings, also "for free", which it presumably has both the surplus, and the operating margin to do, we're toast. That hasn't happened, yet, though it has concerned me since Working Group C. But suppose something less predatory. Suppose simply that VGRS decides that it will fund a registrar that will specialize in non-VGRS inventories. Absent "price controls", how do we avoid tending towards de facto monopoly for the registrar market in non-VGRS inventories? Suppose that some highly capitalized registrar obtains this de facto monopoly -- some will applaud VGRS (or whomever) for subsidizing and therefore expanding the total market for non-VGRS brands, but what happens if this business unit then decides to reduce the accessibility to, and the value of, non-VGRS brands? I'm not suggesting that the CRAI-2 model should not go forward, only that there are unexamined risks and we really must distinguish between the NCBA starting up a cooperative registrar that sells non-NCBA inventory, and {VGRS|NS|AF} siphoning off the profits in the retail and bulk registrar markets, and in particular, the risk that reducing the retail pool available to independent registrars will reduce the ability of independent registrars to initiate competitive registries. Eric Brunner-WilliamsI'm employed by CORE, a registrar, and a registry operator, so my comments should be read with those interests in mind. |