On February 6, 2003, the GNSO's New gTLD Committee met for the first time. The Committee endorsed what it correctly perceived to be the preference of Stuart Lynn's action plan: that there should be "some form of structure" for new TLDs, and that that "structure implied future names would be in some way sponsored." The Committee further decided that "there should be a semantic connection between the name and the sponsor," and "a new gTLD should be akin to a franchise, and therefore not owned by the sponsor/registry." We feel strongly that the Committee is heading in the wrong direction: any attempt to establish top-down control over the market-based semantic meaning of strings (and the relationship of these meanings to particular "sponsors") is doomed to failure and will subject ICANN to increasing worldwide ridicule. We urge the GNSO's New gTLD Committee, and the Board, to reconsider the Lynn approach."

1. Must ICANN choose between opening sponsored and unsponsored TLDs? What are the consequences of doing so?

Although the categories of sponsored and unsponsored TLDs (and of "restricted" and unrestricted unsponsored TLDs) have been assumed to provide the framework on which all future "delegations" will be based, these terms are merely figments of ICANN staff's fertile imagination. Here are the key differences:

• Sponsored TLDs receive broad "delegation" of policy-making power (to set prices and registrant qualifications); unsponsored TLD contracts are subject to continuing ICANN control over policies and prices.
• Sponsored TLDs must contractually commit to abide by a charter and risk denial of renewal of their contracts with ICANN if ICANN concludes that they are not adequately responsive to some specified "community"; unsponsored TLD registry contracts also risk denial of renewal, and must re-compete for operation of the registry after five years, but are not considered to act for a particular community.

If ICANN merely assessed the mimimum technical and financial qualifications of prospective registries, and if its contracts merely required compliance with provisions designed to assure technical interoperability (and compliance with consensus policies), there would be no occasion for the distinction between unsponsored and sponsored registries to be made. It would be up to each registry to decide whether to limit eligibility for registration and how best to address and serve a particular segment of the potential market for domain names. Because every TLD ends up serving a community of registrants, ICANN could leave it to TLD delegatees to decide which potential customers to address and how best to serve them.

Such an approach would extricate ICANN from several difficulties. If it removed the "sponsored/unsponsored" distinction, ICANN would not be required to "define" a particular community, much less to seek to prevent competition among registries to serve potentially overlapping groups. ICANN would not have to evaluate the governance mechanisms by means of which internal policy is made for various TLDs. It would not have to evaluate the business models (or profit or nonprofit status) of those it approves as qualified to operate a TLD.

ICANN may believe that the intellectual property community or others would be happier with sTLDs because they will have fewer registrants and be easier to police. But the size of the group potentially served by a TLD is not necessarily related to its status as an sTLD or a gTLD. For example, the group of all mobile device users (who might use the sTLD .mobile) is probably much bigger than the group that is actually interested in .biz (an "unsponsored" TLD). So choosing among the two (or three) existing models for TLD contracts does not and should not be perceived to have any particular relationship to the number of new registrations that will result.

Indeed, the type of a TLD relates only slightly to the nature and magnitude of intellectual property issues that may arise with respect to registrations. For example, a very tightly focused sTLD (e.g., .museum) may entail semantics that make it unnecessary for trademark holders to worry about cybersquatting by means of registrations under that TLD. But the same might be true of restricted gTLDs as well. (Whenever eligibility requirements narrow the potential range of registrants, the potential for confusion or hold-ups will diminish.) And many open gTLDs may also entail semantics that reduce any potential for confusion or cybersquatting. A .sucks or .blog would not present a space in which anyone would expect to find the authentic sites of major corporate brands. In any event, cybersquatting in all ICANN-contracted TLDs is adequately handled by the UDRP. There is no principled basis on which ICANN's division of TLDs into "sponsored" and "unsponsored" models can be grounded on intellectual property concerns. (Moreover, the intellectual property community should note that ICANN's broad delegation of policy-making powers to sTLDs may well make them less subject to ICANN's consensus policies than are gTLDs.)

Some suggest that sponsored TLD operators must be nonprofits, whereas unsponsored TLDs may be profit making companies. But the nature of the business model, or tax status, of a TLD delegatee is none of ICANN's business. Perhaps one form of business may be better able to make investments; perhaps another may be more predictably inclined to establish policies that serve a particular community. ICANN has no right or capability to determine which model is best for which circumstance. The market will decide who succeeds and who fails. ICANN should at most be concerned to impose on all a uniform obligation to take steps to minimize any risks of problems for registrants in the event of registry operator failure. Such steps could include, for example, establishing escrows and encouraging contracts for rapid transitions of responsibility in the event of difficulties.

The "sponsored/unsponsored" set of assumptions falls apart on close examination. Unsponsored TLDs can and should set eligibility requirements if they wish. ICANN simply has no special competence to draw lines between particular communities. If ICANN were to try to protect particular TLD registries against competition, serious antitrust questions would arise. If the concern is cybersquatting, we have the UDRP. If the concern is stability, we have contractual ways to deal with it. The world can carve itself up; ICANN should not be in the business of carving up the world. The GNSO should tell ICANN so.

2. Are there other options available at this time?

Yes. Most of the provisions in the current contracts between ICANN and TLD registries originated with ICANN staff, not from any consensus process. If the ICANN Board were to decide to substantially "slim down" the contracts, it would have the power to do so.

The GNSO could advise the Board to direct staff to:

• remove all express obligations unrelated to preservation of technical interoperability, recovering from registry failure, and compliance with consensus policies
• insert a nearly conclusive right to renewal, to allow and encourage investment
• eliminate provisions that appear to require ICANN staff approval before the introduction of any new optional value added services
• eliminate any requirements to use any particular distribution channels to get TLD registrations to market.

Perhaps the best way to get community review of ICANN's course with respect to new TLDs would be to invite and permit any applicant to propose any type of TLD and any form for the contractual relationship between ICANN and the TLD registry. The Board and community would then be presented with a full range of options, and the Board could approve contracts with public input based on detailed information regarding the benefits to be provided by any particular combination of provisions (subject to a "floor" of minimum provisions with respect to the UDRP and technical interoperability). The key point is that there is no reason to have identical contracts with each operator, provided that minimum technical and financial standards are met by applicants and stability is assured.

Similarly, ICANN staff's approach to the RFP process could be molded by advice provided by the GNSO in the short term. Even if ICANN staff is committed to issue an RFP that contemplates rolling out only sTLDs, the GNSO could advise the Board promptly to establish the minimum technical and financial qualifications that could be used to qualify all applicants. For example, a registry might be required to show that it can, directly or by means of subcontract, operate in compliance with industry standards. In addition, a minimum financial qualification test might require each registry to have some specified sum (e.g., $1 million) available via bond or unencumbered assets. Once these minimums are set, the GNSO could advise the Board to move promptly to invite applications by new registries and to condition acceptance only on the applicant's willingness to sign a thin contract shorn of previous over-regulatory provisions. The sooner the Board adopts such an approach, the sooner existing registries and registrars would be able to request re-statement of their own contracts -- so as to become free to define their own products and business models in a manner that best responds to market conditions.

3. Shouldn’t ICANN insist that a registry serve a specified community?

No. While it is natural to want to use ICANN's leverage to assure that a registry provides service and policies that a targeted community finds valuable, the marketplace will do a far better job of ensuring that TLDs have incentives to provide valuable services.

Use of a regulatory model assumes that there will be only one TLD serving any particular community. But registrants should have choices, and ICANN should not be -- indeed, may not lawfully be -- in the business of protecting TLDs from competition. ICANN's current approach places it in the position of creating mini-monopolies, without any showing as to why that would be beneficial to registrants.

As long as new TLD providers can seek to serve any particular group of potential registrants, the incumbent registries will need to compete for the business and loyalty of those they serve. Additional regulatory intervention by ICANN would overextend its resources and place it in the position of reaching judgments on questions it does not have the institutional ability to decide.

4. How should the taxonomy question be resolved?

Quickly, through deciding to let the market decide which potential registrants will find value in particular strings.

If the GNSO were to recommend adoption of a taxonomy, it would have to recommend a particular taxonomy -- and ICANN would then have to ban all other strings or develop rules regarding which strings are "confusingly similar." If it chose a list of permissible TLD strings, ICANN would then have to choose among competing applicants, based on criteria that would take it into subjective assessments of how "representative" a particular TLD registry operator might be of some group ICANN had determined should be the beneficiary of its artificial semantics. No one is smart enough to foresee all the different ways in which a TLD might be used or valuably named, and it is not clear that the GNSO has any particular institutional competence to make recommendations in favor of reserving particular TLD strings.

Close analysis suggests that a large number of intractable problems will be posed if the Board goes down the "taxonomic" path of creating specific entitlements to TLD strings, overseeing the governance of "sponsors," and making centralized decisions regarding permissible "communities" and markets ahead of time. Because no sound technology-based rationale could be advanced to support imposing a taxonomy, the imposition of such a rule would arguably violate the antitrust laws. Such a decision would also conflict with obligations in ICANN’s MOU and Bylaws requiring it to promote competition to the maximum extent feasible.

The posing of the "taxonomy" question itself could lead to endless delay. But the GNSO should be smart enough to see that the entire inquiry is a waste of time. The GNSO could promptly advise ICANN to proceed with a new TLD rollout plan that establishes minimum qualifications but leaves the marketing questions up to those with the highest stake in getting them right -- the registries themselves. If all qualified applicants were allowed to enter the market (on some reasonable pacing schedule), no registry would expect to be protected from competition and all applicants would likely be able to find a non-conflicting string that would allow them to serve their customers.

5. Can we end the "review" process of the "proof of concept" TLDs?

The lessons learned from the previous launch of seven new TLDs are already available for all:

• complex agreements and long negotiation periods do not enhance the ability of new registries to serve their customers
• sunrise periods create more problems than they solve
• registries should be permitted to seek whatever distribution channels best serve their markets, and
• no one can predict with any assurance in advance how well a particular string will be received by the marketplace.

6. What's the best thing for the GNSO to do?

The newly formed GNSO should solidify its influence by telling the Board to take action promptly to rationalize the new TLD process. Rather than wading into the taxonomic swamp of semantic meanings and representativeness, or waiting for an endless and fatally fuzzy "review" process to end, the GNSO should tell the Board to:

• promptly facilitate the creation of minimum, objective technical and financial criteria for registry operators
• announce a technically-based upper limit for the number of new TLDs to be added to the root each year
• open up applications for new TLD operators
• announce a schedule by which new TLDs (selected at random from qualified applicants during any particular period) will be recommended to the Department of Commerce, and
• offer technically-based, lightweight contracts to new and existing registry operators.

The way to maximize ICANN's effectiveness, and the value created by the DNS for existing and potential markets, is to get back to basics. ICANN needs to ensure that it is erecting the minimum possible barriers to any financially and technically sound business to seek to create new value. The internet itself was built because no one had to get permission to proceed to try a new business model or develop a new application. ICANN cannot lead the market. It should simply get out of the way. As its first action in the new year, the GNSO should tell the Board that it is time to act constructively to open the name space to all qualified providers.

David R. Johnson is teaching a course at Yale Law School about internet governance and policy; Susan P. Crawford is teaching cyberlaw at Cardozo Law School. We are writing in our personal capacities, and not on behalf of either of our institutions or clients.