Forced "differentiation" is anti-competitive
These comments are submitted by Wendy Seltzer and John Levine in our individual capacities: We object to the Initial Report TOR 2.15 recommendation: "Applicants must offer a clearly differentiated domain name space with respect to defining the purpose of the application. " We believe this requirement conflicts with several of ICANN's core values, most particularly those of reliance on the market (5), competition (6), and minimization of interference (2). "Differentiation" is a barrier to competition in the domain name market, as the essence of differentiation is to make TLDs less substitutable. Forcing new registry applicants to differentiate their offerings from those of existing registries expands the monopolies of existing registries (which start with a monopoly over the particular strings they run). ICANN's function is not the apportionment of monopoly rents to a few favored database-maintainers. ICANN cannot simultaneously reject new applications that could substitute for existing TLD strings and claim to rely on competition to regulate the behavior of gTLD registries. Since ICANN is ill-suited to regulate registries (as a matter of structure, antitrust law, and history, as demonstrated with the .com reissue and "new registry services"), it should enable a robust market to "regulate" them by means of real, direct competition. If a potential entrant believes it can offer a similar (undifferentiated or generic) service more attractively than an existing registry, it should be permitted to do so. Those offering similar strings will be able to compete on other characteristics (price, service, ancillary offerings) that give the consuming public real options, so that the public can choose those that deliver the best value. While it might be argued that differentiated TLDs offer competition for the market, rather than in it, it is not ICANN's role to choose among flavors of competition. That choice should be left to the market actors and their experimentation. Some experiments may fail, while other may succeed at serving popular demand beyond the imaginings of any central planner. ICANN simply cannot anticipate all the uses the public may make of domain names as the Internet and its uses change. It should therefore use the market to gather information: the willingness of an applicant to try a gTLD string is a signal of that string's potential value. (Applicants might be vetted for minimal technical qualification, or might be able to use a previously qualified Registry Services Provider, as suggested by Ross Rader and Eliot Noss <http://dnspolicy.org:4080/index.php?n=MakingChoices.WhatIsARegistryOperator?>, but not for the beauty or differentiation of their chosen string.) Of course ICANN should be looking out for interests the market may fail to address, but those failures are likely to be fewer than the failures of over-regulation that are evident in the plodding pace of new gTLD introduction in ICANN's eight years. Rather than applying a vague "differentiation" criterion, I suggest that ICANN should accept all TLD strings except those specifically likely to cause typo-confusion based on single-letter variance from or visual equivalence (in standard script) to an existing gTLD. Thus .con and .c0m might be rejected, but .market and .shop approved. If "differentiation" is defined narrowly to mean only this typo and visual distinction, rather than semantic distinction, it would be compatible with a competitive market. We remain convinced that ICANN's rule for new gTLD applications should be a "default accept." Criteria that go beyond investigation of specific harm from a new TLD string put ICANN into a regulatory role to which it is unsuited. --Wendy Seltzer and John Levine (North American representatives to the Interim ALAC), speaking individually |