(The following is an edited version of comments I submitted to ICANN regarding “closed” generic TLDs.)

On February 5th, ICANN solicited comments on whether ICANN should determine the circumstances under which a generic TLD operator is permitted to adopt “open” or “closed” registration policies. Allowing gTLD operators to make these determinations, as opposed to ICANN, will promote innovation on the Internet to the benefit of consumers.

In order to bring the benefits of a competitive TLD market to consumers, ICANN should generally take as light-handed a regulatory stance as possible, as long as it meets its technical responsibilities. A light-handed regulatory approach is consistent with the policy of relatively open entry into the TLD space that ICANN has adopted.

A benefit of the new gTLD program, in addition to providing competition to incumbents, is the ability of the entrants to develop new business models, products and services. Historically, gTLDs have been open, and arguably that openness benefited the growth of the Internet. But at this stage of the Internet’s development, adding new options to the status quo is more likely to unleash new forms of innovation. Closed gTLDs may be a promising source of innovations that have not thus far been possible to implement or even envision. Closed gTLDs may, for example, be a way to provide services with enhanced security. No one can know what innovations might be blocked if ICANN generally required gTLDs to be open. In short, adding new open gTLDs is likely to create benefits, but the returns to adding completely new types of gTLDs are potentially much larger.

New gTLDs are valuable economic assets. ICANN should adopt policies that assure that these assets are allocated to their most highly valued uses. ICANN’s decision to use an auction when there are multiple applicants for the same gTLD will further that goal. The bidder who believes its business model will be the most profitable will win the auction and the right to operate the gTLD. When there is only a single applicant, that applicant presumably represents the highest-valued use of the gTLD.

The best use of a gTLD can change (e.g., from closed to open) if the initial business model isn’t successful or if economic conditions change. This change can be effected either by the current operator or by a transfer of the gTLD to a new operator, subject to ICANN’s review. In this way, gTLDs can continue to move to their highest-valued uses over time.

The dangers of ICANN dictating how gTLDs should be used are illustrated by the experience with radio spectrum. Historically, the U.S. Federal Communications Commission allocated blocks of spectrum to specific uses—e.g., broadcast radio and television. Over time, the costs associated with misallocation of spectrum under this “command-and-control” regime became very large. The process of reallocating spectrum to higher-valued uses has proven lengthy and difficult. Although the U.S. and other countries have moved toward a more market-based system, the costs of the legacy system are still reflected in the scarcity of spectrum for wireless broadband uses.

Several commentators have expressed concern that closed gTLDs are anticompetitive. No evidence supports this claim. First, we already have experience with generic second-level domain names—e.g., cars.com—which have provided useful services with no apparent anticompetitive effect. There is no reason to expect anything different from a .cars gTLD. If, for example, General Motors (or any other automobile company) were to operate .cars, it is not plausible to suggest it could thereby gain market power in the market for cars. Note also that both operators and ICANN are subject to the U.S. antitrust laws if they use the TLD system in an anticompetitive way. To the extent that ICANN allows synonyms as gTLDS—e.g., “autos” “automobiles”, “motorvehicles”, perhaps even “goodcars”, etc.—the potential competitive problems become even more remote.

In sum, ICANN should provide maximum flexibility for operators to experiment with new business models. This is the best way to promote innovation on the Internet.