Comments on new gTLDs

These are the exact comments posted in the ICANN Forum. See also a short discussion of TLD charters, and a short discussion of TLD approval processes.
Answers to selected questions from

ICANN Yokohama Meeting Topic: Introduction of New Top-Level Domains

Kent Crispin

Q36: Should the formulation of policies for limited-purpose TLDs be delegated to sponsoring organizations? In all cases or only in some? In general, this should be decided on a per case basis. In all cases, the resulting policies should be available for public comment and review prior to ICANN approval. Q37: What measures should be employed to encourage or require that a sponsoring organization is appropriately representative of the TLD's intended stakeholders? Applications must at least be up for public review and opposition for an adequate amount of time -- I believe that 90 days might not be enough. Moreover, the public review should be aggressive -- we will not find it easy to be sure that appropriate parties in say India are well informed, for example -- and should involve notification of public authorities through, for example, announcements at appropriate venues in the UN. Moreover, sponsors or other proposers should present documentation of *their* attempts to publicize widely. Q38: In cases where sponsoring organizations are appointed, what measures should be established to ensure that the interests of the global Internet community are served in the operation of the TLD? By definition, a sponsored TLD may be intended to serve a limited population. This is not a bug, it is a desirable feature of sponsored/chartered TLDs -- by catering to a limited audience they are segmenting the domain name space, and operating against the unfortunate practive of getting the same SLD in every TLD. So, one could say that the best way for sponsors to serve the global community is to well-serve their particular community. However, it is also essential that the sponsoring organization be responsive to policies that do apply globally, and thus it is important that there be clear contractual terms that bind the sponsor to such standard policies. An important point, though, is that if sponsors and registry operators are separated, many the global policy requirements would be enforced through the registry operators accreditation agreement, and the sponsor would simply not deal with such issues. Q39: How should global policy requirements (adherence to a TLD's charter, requirements of representativeness, interoperability requirements, etc.) be enforced? Adherence to a charter, and requirements of representativeness, should be enforced through a more or less standard chain of appeal and due process that every sponsor would define as part of its "sponsor accreditation agreement". An entity that proposes to be a sponsor should also be required to propose the mechanism by which its own decisions would be appealed, up to ICANN, if necessary. Such a proposal should treat ICANN as the court of last resort, but be designed to drastically minimize the possibility of ICANN ever being involved. Sponsor proposals should be required to deal with this issue in depth, and to be willing to iterate with ICANN on the matter. Hopefully, if there are several dozen good proposals a relatively standard set of mechanisms will become apparent. Interoperability requirements are a separate matter. Registry operator accreditation should be a separate contract from sponsor accreditation, and registry operators could be fairly strictly bound to follow IETF RFCs, good data preservation practices, and so on. These are details that many sponsoring organizations would have absolutely zero interest in, but more important, little competence. Q49: Does the schedule allow sufficient time for formulation of proposals? I think it is too aggressive. I am more concerned, however, about the short time allowed for development of the call for proposals and especially, the application form. The application form is a serious, complex document, and, as I note below, I think there should be two application forms -- one for sponsorship or creation of a TLD/policy, and another one for registry operators. Q50: Does the schedule allow sufficient time for public comment? Absolutely not. The public comment portion is 7 days, and most of the applications will be filed on the last day. There should be *at least* two rounds of public comment, of at least two weeks, and applicants should have opportunity to revise applications in light of public comment. Q51: Should all proposals be posted for comment simultaneously to maintain equal time for public comment? Should all proposals be posted for public comment as they are received to allow the greatest possible time for public analysis and comment? Realistically speaking, can you imagine that anyone making an application is going to be significantly *early*? You may get a few early birds, but in an immediate publication scenario, everyone will wait until close to the end to see what others will do. I think you would be better off to collect applications for two months, post them for a month of public comment, split into two rounds, with opportunity for modification in light of comments received. Q52: Should the formal applications be posted in full for public comment? If not, which parts of the applications should remain private? TLD proposals (as opposed to registry operator proposals) should be *very* public. If someone has a business plan that depends on secrecy then they should be in a different business. Employee names and addresses, and other such privacy data, would not be required. But anything to do with the policy/charter/enforcement must be disclosed, and the details of the sponsoring organization should be disclosed. Registry operator proposals need not meet the same standards. The application should not be secret, by any means, but the same scrutiny is not required. Q53: Should proposals choose a single proposed TLD or numerous possibilities? There could be multiple possibilites. If so, they should be prioritized. Note that registry operator proposals need not specify a TLD at all. Q54: Should ICANN select the TLD labels, should they be proposed by the applicants for new TLD registries, or should they be chosen by a consultative process between the applicants and ICANN? There should be two kinds of proposals requested by ICANN: 1) TLD proposals, which specify TLD labels, policies, charters, and sponsors; and 2) Registry operator proposals, which specify the company and its technical and business characteristics, and optionally, one or more TLDs that it is interested in operating. If a TLD proposal contains a sponsoring agency (and I expect that most in the initial rollout will have sponsors), then the sponsor needs to make a case for its suitability as a sponsor -- stability, representativeness, appropriate expertise in the case of a charter, and so on. Organizations that might be excellent sponsors (The International council of museums, universities, industry consortia, etc) might be lousy registry operators, and vice versa; the desired characteristics are really quite different. An application from a proposing sponsor should also contain details of how it intends to carry out its sponsorship, including an appeals process. It should also define itself in the abstract, and define the metrics by which its performance should be judged. Q55: Should there be minimum or maximum length requirements for TLD codes? Are restrictions appropriate to avoid possible future conflicts with ISO 3166-1 codes? Two letter codes should remain reserved for ISO 3166 codes. Q56: Should there be restrictions on the types of TLD labels that are established (for example, a prohibition of country names)? Yes, there probably should be such restrictions, but they cannot be developed in a vacuum or all at once. Initial TLD names should be very conservatively selected, and subjected to heavy public review. From such substantive review some rules should quickly become apparent; with time a set of "black letter" rules could be developed. But it is too much to expect that such a set of rules can be created immediately. Q57: What should be the criteria for selecting between potential TLD labels? Should non-English language TLD labels be favored? The most important criteria is simply public approval. TLD names should be publicized, and comments solicited, and evaluated on a common sense basis -- some people might find .shop ugly, but .screwgod gets to an entirely different level of offensiveness. In the realm of open gTLDs, a good case could be made for a non-english bias. In the case of chartered/sponsored TLDs, the issue is not as clear, and needs to be modulated through consideration of the charter and policies involved. Q58: How many new TLDs of each type should be included in the initial introduction? The approval process should operate on an individual basis, not on a batch basis. In such a situation there will be a first one approved, then a second one, and so on. It is only important that all different types be in process. Q59: Which types of TLDs will best serve the DNS? The most important criteria from a DNS perspective is how successfully the TLD partitions the user space. A purely open generic TLD like .web doesn't do anything at all to partition the user space; and many people who have registrations in .com will immediately get the same registration in .web. A strongly chartered TLD like .museum, however, has exactly the reverse effect; museums will have little incentive to use .com/.net/.org. Q61: Which types, if any, are essential to the successful testing period? open gTLDs, chartered TLDs, and sponsored TLDs. Q72: In what ways should the application requirements for sponsored/chartered/restricted TLDs differ from those for open TLDs? Assuming that registry operators are separately accredited, purely open gTLDs require nothing more than a name to be approved, and for ICANN to assign the name to an accredited registry operator. However, open gTLDs still must go through the same opposition/comment process as chartered TLDs, because someone might have a proposal for a chartered TLD with the same name. The most important aspect of this is that the application fee structure should reflect the different level of complexity. It shouldn't cost anywhere near as much to check out a pure open TLD name proposal as it would to examine a complex chartered/sponsored TLD. A registry operator application should require a fairly constant amount of work, and be charged accordingly. Q73: Should ICANN require a statement of policy or should a statement of how policies will be made be sufficient? ICANN should require a statement of policy, even if it only says "purely open gTLD". It would be most distressing if the sponsor for the .arts TLD decided on opening day that it would be better to cater to pornography. Q74: What level of openness, transparency, and representativeness in policymaking should ICANN require? A very high level of these things should be the default, with provision for change if necessary. For example, a TLD to support IP telephony might have a very restricted and technically oriented policy process -- that would be reasonable. But the default should be for the TLD to meet the same standards that ICANN must.