ICANN ICANN Email List Archives


<<< Chronological Index >>>    <<< Thread Index >>>

Price caps are service caps, not just profit caps

  • To: crai-report@xxxxxxxxx
  • Subject: Price caps are service caps, not just profit caps
  • From: Eric Brunner-Williams <ebw@xxxxxxxxxxxxxxxxxxxx>
  • Date: Mon, 15 Dec 2008 10:45:20 -0500

The discussion of price caps, both by existing operators of unrestricted registries, seeking "equal terms", and prospective operators of unrestricted registries, seeking fewer restrictions, both seeking their modification, as well as by registrants of unrestricted registries, seeking their preservation, misses two points.

The first is that the annualized cost of the registry service -- provisioning, database entry persistence, and publication, and the annualized cost of additional registry obligations -- generic to corporations and specific to the ICANN contract, is vastly lower than the current set of price caps. As we pointed out in CORE's proposal for the annual fee, there are a range of gross registry revenues where the registry is in "survival" mode, and a range of gross registry revenues where the registry is "stable". Restated, at some point along the registry volume growth curve, the "margin of profit" transitions from negative to positive, and at some subsequent point along the registry volume range, the marginal profit becomes greater than the marginal cost. CORE's proposal used the familiar $6 cap and under that price model we define that point as the point in time when the registry has between 100,000 and 200,000 registrants. Of course, for registries with volumes an order of magnitude greater, or more, than a million registrations at the familiar $6 "cap", we're really talking about a dollar of cost and five dollars of profit, per registration/year.

From the registrant point of view, this is the unreasonable gouging by registries, and from the registry point of view, this is the return on investment. For registrars this is simply the base unit cost.

The cap defines the costs the registry may reasonably incur. As a rule of thumb, $6/domain/year model assumed only for the purposes of discussion, the cost is no greater than a fifth of the cap, and were the 1999 cap of $50, the value of the services provided by the registry, to the registrar and to the registrant, could be as much as $10. The services might be postal provisioning of the registrant of a second-factor security device, publishing the registration's zone data through a distributed, heterogeneous, secure, 4+6 server constellation, and so on.

The point is, we're not simply discussing price caps, we're discussing service caps. The service offered originally by SRI, and presently by the COM, NET and ORG registry operators, is a business model. The RSTEP functional block in the GFA anticipates other registry services, sufficiently complex to merit expert review during the application evaluation process. For those applicants who simply offer an alternate namespace via FCFS, a cap on the registrant exploitation, on the return on investment, whether identical to, or more generally consistent with, the current (ever increasing, but that's another issue) cap, forming a coherent narrative of ICANN's policy towards unrestricted minimum service registries is better for the registrants, and the investors, than an incoherent ICANN policy narrative, or one that keyed to points in time, even personalities "at the table", that only those of us who are "at the table" are likely to recollect a decade from now.

There is an additional service which arises from base unit cost and one additional factor, which is easily discerned when we examine the fraud and related exploits of the DNS. The Srizbi system uses domain names of the form aeedaute.com, aeggdwyu.com, aeoytgrr.com, ... ywiiuoar.com, yworpaww.com, ywwuyrew.com for reconstitution, and during the recent multi-week period when the Srizbi system was partitioned (the McColo shutdown), the system attempted to acquire on the order of a thousand domain names per week. Clearly these names are not ment to be presented to users, they are for infrastructural use only by the Srizbi system. We don't know why the authors of the Srizbi system chose one capped inventory provisioner over another, but we know that those authors of this particular instance of a system that has critical dependency upon the DNS, for infrastructure, for content delivery (spam, phish, and so on), for network resource acquisition (malware installation and so on), and for economic value capture (money, identity, and so on), picked the lowest priced inventory.

As a general principle, bad actors avoid high prices, even when the ROI is several orders of magnitude greater than the base unit price for domain names, and even when the bad actors don't actually pay the price, whether exploiting the AGP or stolen credit cards. There is a second form of this "price alone determines bad actor inventory preference", which is present in the use patterns by the authors of the Srizbi system. All their infrastructure buys went through highly automated registrars. Restated, the degree of difficulty for a completely automated (and therefore scalable) system determines the system's registrar selection, and the base unit price determines the system's registry selection.

There are registrants who share the lowest cost easiest acquired goals of the Srizbi system's authors, and there are registrants who have goals other than lowest cost easiest acquired, goals such as reliability, reputation, reachability, and so on. This is the second point, that "price" itself is a "service".

CORE's position is that innovation is important, and we separately will address the penalty the RSTEP fee places on applicants seeking to do more than simply copy the unrestricted FCFS model with a suffix that isn't confusingly similar to an existing operationalized unrestricted FCFS model. The minimum service model appropriate to the SRI period, prior to the USG forced transition of the NSF backbone to a competitive market with no consideration of economics or security, and exploited very profitably COM, NET and ORG successor, is not what ICANN should be mandating as a maximum service model, nor is it what ICANN should be forcing through RSTEP fees.

The discussion of price caps should be fully informed by our goals. Is a 500% profit margin for Verisign and a smaller profit margin for Verisign's competitors consistent with the principles of the White Paper? How is profit structured?

Eric Brunner-Williams

<<< Chronological Index >>>    <<< Thread Index >>>

Privacy Policy | Terms of Service | Cookies Policy