GNSO Initial Report Introduction of New Generic Top-Level Domains

Wellington Meeting Input

A. Summary
B. Process Background
C. Constituency Statements
D. Public Comments
E. Submissions on Call for Papers
F. Summary of Early FindingsRecommendations
G. Analysis of Impact on Each Constituency
H. Relationship to the PDP on IDNs
I. Relationship to PDP FEB 06
J. Next steps

A. Summary

1. This is the second phase of the Initial Report on the policy development process for the introduction of new top-level domains. The key issues under examination are whether to introduce new gTLDs, the selection criteria associated with any introduction on new gTLDs, any allocation methods that could be used to enable the introduction of new gTLDs and policies associated with contractual conditions for new gTLDs.

2. The first version of this report (http://icann.org/topics/gnso-initial-rpt-new-gtlds-19feb06.pdf ) reflected a wide range of input received from interested stakeholders including Constituency Statements, Public Comments and submissions in response to a Call for Papers about the possible introduction of new gTLDs. This updated report benefits from further inputs received at the GNSO Council's new gTLD PDP Committee Meeting held on Friday 24 and Saturday 25 February 2006 in Washington DC.

3. The Washington meeting was attended by a range of GNSO Councilors, Constituency representatives and other members of the community¹. The meeting was recorded and people were able to join the meeting via teleconference. The meeting benefited from the presentation of papers (http://gnso.icann.org/issues/new-gtlds/new-gtld-pdp-input.htm) and a question and answer session with respondents to the Call for Papers.

4.The most important element of the Committee meeting was to expose the PDP's Terms of Reference to further input, in the context of ICANN's Bylaws, Mission and Core Values that constrain the GNSO's policy development activities. Particular effort was made to ensure that lessons were learnt from the previous rounds of new top-level domain expansions in 2000 and 2004 and assignments of the .org and .net contracts.

5.

6. General consensus² has developed around the first term of reference — whether there should be new top-level domains. This "yes" answer is conditional for some constituencies on the appropriate development of robust selection criteria, allocation methods and contractual conditions³. In addition, there was little disagreement from the public comments or call for papers contributors about whether new TLDs should be should be introduced.

7. There is consensus around both the principles for developing selection criteria that map directly to ICANN's Bylaws, Mission and Core Values and the practical impact of providing appropriate policy guidance to the Board about criteria that could be used in further rounds of new top-level domain applications⁴.

8. There was agreement that further work needed to be done with respect to technical criteria. A supplementary call for information from Constituencies was issued on 8 March 2006 regarding four specific areas of interest. Namely:

1. Should the minimum technical criteria for registry operations be set according to the current registry operations of, for example, .NET requirements?
2. Should the minimum technical criteria make some reference to the proposed size of a new registry?
3. Should a separate registry operator's accreditation scheme be established and, if so, what should that scheme look like?
4. Should other business operations criteria continue to be included in a registry operator's application to ensure that any registry operator is adequately funded and professionally managed?

9. There is also some consensus around the treatment of allocation methods in the written submissions. However, this issue requires further examination in the discussion about allocation methods which will take place at the March 2006 Wellington meetings.

10. The consideration of appropriate policy for the development of new contractual conditions needs to be undertaken in the near future, after discussions of selection criteria and allocation methods are completed. This work needs to refer, in part, to the new policy development process which was initiated in February 2006. A public comment period on the issues raised by the new PDP's Terms of Reference is now being conducted (http://icann.org/announcements/announcement-06mar06.htm).

11. The next step is to develop the findings on these issues through further work and consultations within the GNSO itself at the Wellington meetings, in the first instance, and then through further rounds of consultation and public comment periods as defined by the PDP rules.

12. In addition, co-operation with other ICANN Supporting Organizations and Advisory Committees and consultation with the wider Internet community is a necessary part of the policy development process.

13. It is proposed to release the final version of this Initial Report, which will include input from the Wellington meetings, on 2 May 2006.

14. At that time, a twenty-day public comment period can commence. It is expected that the final version of the Initial Report will be sent to other Supporting Organizations, the Governmental Advisory Committee and the ALAC for their formal input.

B. Background

1. For clarity, the process by which this updated draft has been produced is useful to set out. The first step was the production of theIssues Report⁵ that was used to solicit formal Constituency Statements, public comments through an ICANN public comment forum and as a guide to individuals to submit papers which addressed the Terms of Reference. The Issues Report was discussed at the 2 December 2005 GNSO Council meeting where the Council resolved to launch a PDP with the following Terms of Reference:

1. Should new generic top level domain names be introduced?

a) Given the information provided here and any other relevant information available to the GNSO, the GNSO should assess whether there is sufficient support within the Internet community to enable the introduction of new top level domains. If this is the case the following additional terms of reference are applicable.

2. Selection Criteria for New Top Level Domains

a) Taking into account the existing selection criteria from previous top level domain application processes and relevant criteria in registry services re-allocations, develop modified or new criteria which specifically address ICANN's goals of expanding the use and usability of the Internet. In particular, examine ways in which the allocation of new top level domains can meet demands for broader use of the Internet in developing countries.

b) Examine whether preferential selection criteria (e.g. sponsored) could be developed which would encourage new and innovative ways of addressing the needs of Internet users.

c) Examine whether additional criteria need to be developed which address ICANN's goals of ensuring the security and stability of the Internet.

3. Allocation Methods for New Top Level Domains

a) Using the experience gained in previous rounds, develop allocation methods for selecting new top-level domain names.

b) Examine the full range of allocation methods including auctions, ballots, first-come first-served and comparative evaluation to determine the methods of allocation that best enhance user choice while not compromising predictability and stability.

c) Examine how allocation methods could be used to achieve ICANN's goals of fostering competition in domain name registration services and encouraging a diverse range of registry services providers.

4. Policy to Guide Contractual Conditions for New Top Level Domains

a) Using the experience of previous rounds of top level domain name application processes and the recent amendments to registry services agreements, develop policies to guide the contractual criteria which are publicly available prior to any application rounds.

b) Determine what policies are necessary to provide security and stability of registry services.

c) Determine appropriate policies to guide a contractual compliance programme for registry services.

2. The call for public comments on these Terms of Reference was announced on 6 December 2005 on the ICANN website (http://www.icann.org/announcements/announcement-06dec05.htm

3. At a subsequent 21 December 2005 GNSO Council conference call it was decided to extend the deadline for public comments and Constituency Statements until 31 January 2006. In addition, a decision was taken to launch a Call for Papers to further inform the process. The Call for Papers was announced on 3 January 2006 on the ICANN website at (http://icann.org/announcements/announcement-03jan06.htm ) To give further impetus to this Call for Papers, it was also advertised in some major international newspapers and magazines in mid January 2006 including the Financial Times, the Asian Wall Street Journal and The Economist.

4.

5. The next sections summarize the Constituency Statements, public comments and written submissions to the Calls for Papers, with input grouped according to the issue areas in the Terms of Reference.

6. In the process, it has become clear that further information is required from Constituencies and the community on certain elements of the PDP. This includes a call for further information from Constituencies about selection criteria that relate to technical standards.

7.

8.

C. Constituency Statements

1. Formal Constituency Statements were received from the Non-Commercial Users Constituency (NCUC), the gTLD Registry Constituency (RyC), the Intellectual Property Constituency (IPC), the Internet Service and Connectivity Providers Constituency (ISPCP) and from a subset of the Registrars' Constituency (RC) by 31 January 2006. A draft statement was received from the Business and Commercial Users' Constituency (BC). The Business Constituency submitted its final position on 8 March 2006. The RC submitted its final position on 2 March 2006.

2. The At Large Advisory Committee (ALAC), observer to the GNSO, also submitted a statement. In the next section, the findings in the Constituency Statements and the discussions at the Washington DC meetings are mapped to the issue areas identified in the Terms of Reference. For the full text of each of the Statements, see http://gnso.icann.org/issues/new-gtlds/new-gtld-pdp-input.

C.1 Whether to introduce new gTLDs

1. The Washington DC meeting confirmed that constituencies support the introductions of new gTLDs. Views diverge, however, on what kinds of gTLDs ought to be introduced; the pace of introduction and the selection criteria for gTLD name strings. Some constituencies make their support conditional upon the nature of the gTLDs envisaged, while claiming that conditions relating to competition, differentiation, good faith, diversity and business certainty must be fulfilled to introduce a new gTLD.
   
   

2. The Washington meeting notes⁶ indicate that there were a wide variety of reasons to be cautious including " [the] selection and implementation process was time consuming, expensive and unpredictable; [the] limitation on the number added caused problems for other applicants that met selection criteria; some selection criteria were not objective, clearly defined, and measurable enough to allow independent evaluation to be effective...".

3. Multiple reasons for supporting the introduction of new gTLDs were put forward in the Constituency Statements and Call for Papers responses. These included enhancement of competition at the registry level; increased choice for registrants or end-users, innovative new services for both existing and emerging markets and avoidance of the proliferation of alternative roots.

4. The Washington DC meeting (see reference above) showed that there were additional reasons for introducing new gTLDs including "small TLDs are OK if it meets the needs of the community that has put [the idea] forward and doesn't exclude others that are within that community; the new gTLDs introduced so far do not yet cater for parts of the international community that use characters sets other than the limited set from the ASCII character range; a policy is required for the introduction of IDNs at the top level, and [we] need to consider the political and cultural environments as demand for these IDNs is increasing...".

5. There were some common elements articulated by meeting participants which indicated that the following selection criteria "baskets" were useful⁷:

   1. Sound business, technical and operational plans

   2. Operational stability, reliability, security and global interoperability

   3. Simplicity and predictability of domain name registration rules

6. The consistent underpinning of the discussion was that, whatever action is taken, it is consistent with ICANN's limited technical co-ordination mission; that an enabling and competitive environment for the provision of domain name management be fostered and that domain name registration rules are clear.

7. The Registry Constituency (RyC) supports the introduction of new gTLDs as a way to facilitate competition at the registry level, to increase choice for Internet users, to grow the Internet usage, to test user demand for specific TLDs and to increase public benefit by better serving specific communities. The RyC also states that both the depth and range of its members and the experience from previous rounds prove that there is market demand for launching new gTLDs.

8. The Registrars are in favor of a predictable ongoing introduction of new gTLDs in order to promote market dynamism, innovation and competition, to enable services for additional communities and to pre-empt uncontrolled alternate roots. The Registrars propose a limit of new gTLDs "in the hundreds, possibly thousands but not tens of thousands."

9. The IPC lends conditional support to the introduction of new gTLDs, focused on sponsored gTLDs and performed in a slow and controlled manner. Any introduction should be guided by principles of differentiation, certainty, good faith, competition and diversity. Public interest is served by adding value to the name space. To keep an introduction limited would also limit the risk of registry failure⁸.

10. The ISPCP conditionally supports a cautious introduction of new gTLDs provided they add value and competition while promoting the public interest (although this public interest is not clearly defined) in the name space. The ISPCP believes that only sponsored gTLDs can accomplish these objectives. The ISPCP states that guidelines should be adopted on how to establish the need for new gTLDs. The ISPCP further states the same five principles as the IPC.

11. The NCUC is in favor of introducing new gTLDs as quickly and broadly as possible in order to keep the market dynamic, foster competition and facilitate end-user choice, wherever Internet users may be located. A well-defined, fair and efficient process is called for and ICANN should accommodate applications for new gTLDs as long as there are no adverse technical consequences.

12. The BC statement recommends that the priority should be to introduce new IDN top-level domains, while no other gTLDs should be introduced at this point in time. Only sponsored gTLDs should be introduced according to the BC. The BC calls for safe harbor provisions in case of registry failure and re-bidding for existing gTLDs.

13. The ALAC is in favor of an open-ended introduction of new gTLDs governed primarily by the market as expressed by the gTLD applicants. The ALAC recommends that there be no explicit limits on the total number or frequency of gTLD additions other than the processing limits of ICANN. As to the types of new gTLDs preferred, views seem to differ somewhat among ALAC members, with a majority supporting unrestricted gTLDs.

C.2 Selection Criteria

1. It was clear from the Constituency Statements that significant discussion about each element of the Terms of Reference had taken place. For example, the Registry Constituency identified that 11 out of 13 of its members had been involved in the drafting of their Statement. It was also clear that all of the Constituencies had had long discussion over several years about introduction of new gTLDs and had participated actively in the discussion of the 2000 and 2004 rounds.

2. The NCUC argues that "ICANN has no mandate its mission or core values to 'expand the use and usability of the Internet'". The promotion of competition is, however, one of ICANN's core values. The NCUC argues that the best way for ICANN to do that is to make "selection criteria as simple, predictable and content-neutral as possible". The RyC, the ISPCP and the IPC all argue that the selection criteria used in previous rounds are a good starting point for new gTLDs with a focus on compliance with technical standards and network stability. The ALAC stated "ICANN should accept all applications from qualified entities that either benefit the public interest or enhance competition in the registration of domain names".

3. There is very limited agreement across the Constituency Statements on which historical selection criteria ought to be included. However, there is some agreement about principles of differentiation (of name spaces), certainty (of business operations), good faith (registration of names), competition (between different registry providers) and diversity (of usability). The RyC includes a detailed set of questions which could be used to determine what selection criteria could be removed. This analysis is based on whether particular selection criteria meet ICANN's technical objectives, provide objectivity, encourage different users and different uses of the Internet, allow market forces some element of influence and enable policy decisions to be made in the best interests of all stakeholders.
   
   

4. The NCUC argues that the only relevant criteria are those that would determine whether an application meets minimum technical standards established to safeguard against harm to the domain name system.
   
   

5. There are divergent views on whether to introduce sponsored or unsponsored gTLDs. The concepts of "sponsored/unsponsored", "chartered/unchartered", "open/closed" TLDs needs further clarification as it was clear at the Washington meeting that different views are held.
   
   

6. The NCUC argues that there should be as much opportunity as possible for users to determine what new gTLDs should be introduced. The RyC also argues that there is little evidence to indicate that sponsored gTLDs are better than unsponsored in encouraging "new and innovative ways of addressing the needs of Internet users". The IPC claimed that "...The introduction of unsponsored gTLDs such as .info and .biz added little if anything to competition at the registry level...". The ISPCP states that "...any new gTLD proposal should be sponsored". A part of the ALAC submission says that "restricted TLDs would cause unsolvable conflicts" but it is unclear whether there is an agreed position from the ALAC on this and other issues in its submission.

7. The ALAC suggests that gTLD strings should be proposed by applicants, not pre-selected. A public notice period should be launched for each application with a possibility to challenge the chosen string and a dispute resolution procedure should be established to resolve differences. Strings should not indicate a scope wider than the remit of the applicant. The principle of non-discrimination should govern selection. Application fees should be affordable and staggered and should reflect a cost recovery model for ICANN's administrative processes.

8. There is consensus on security and stability as primary objectives although how that could be achieved through selection criteria should be determined through future discussion with other experts. The NCUC suggests a "simple and objective 'registry accreditation' process, similar to the registrar accreditation process". This element is discussed further in the Call for Information on technical criteria.

9. Clarification is required for other selection criteria including those that relate to "adding value to the name space" and selection criteria that would support IDN architecture compatible with IETF standards.

10. To summarize, there is a need to seek further convergence on views on selection criteria. To develop selection criteria that meet the objectives and needs of a diverse user community requires prioritization along the following lines:

C.3 Allocation Methods

1. The Constituencies' views on allocation methods differ considerably and further discussion of allocation methods is needed at the Wellington meeting. The sections below also need further discussion in relation to more detailed selection criteria and proposed policy for contractual conditions.

2. A starting point for some statements is that first-come first-served (FCFS) is a natural choice as an appropriate allocation method. This approach assumes that there is sufficient operational processing capacity and domain name space available for new gTLDs. It also assumes that other allocation methods are only necessary in situations where the number of applications is greater than the available slots or where there is more than one applicant for the same gTLD string (for example, as in the 2004 round where there were two applications for .tel). Logically, the viability of FCFS would primarily depend on the number of available slots for new gTLDs compared to the number of potential applicants. At the Washington DC meetings, it was indicated that ICANN operational capacity for processing applications and providing registry liaison support should not be limiting factors in selecting allocation methods.

3. The IPC doubts the viability of FCFS on the basis of experience with "land rush" in domain name registrations, while, for example, some Registrars see FCFS as viable and regard other allocation methods as only needed for situations where there are two or more applications for the same string.

4. The comparative evaluation method, of which ICANN does have considerable experience, is the preferred method of the IPC, supported in this view by the ISPCP and by the BC statement. The BC also argues that no new gTLDs should be introduced unless they are sponsored and IDN enabled.

5. The NCUC suggests that comparative evaluations have numerous disadvantages and are, by their nature, at odds with requirements for objectivity. The RyC emphasizes objectivity and predictability from the applicant's perspective as grounds to minimize the use of comparative evaluations. No statements elaborate on whether comparative evaluations (or elements thereof) should be undertaken by in-house staff or external parties. The 2000 round, the .net reassignment and the 2004 round serve as examples of different approaches in this respect.

6. The original statements from Constituencies and others reflect mixed attitudes to auction models as a methodology for allocating new gTLDs. Further discussion at the Washington DC meetings indicated that auction models were not the best way to make decisions about contested applications but this view needs clarification.
   
   

7. Lotteries also meet objectivity criteria and are one of NCUC's preferred solutions. There are, however, downsides to this method including dissatisfaction with leaving important choices to mere chance and the risk for a "secondary market" with undue profits for winners. It is also noted that lotteries are highly regulated and may not be a viable option from a legal point of view.

8. Ballots are mentioned to have superficial appeal as they are based on democratic principles, but doubts are raised as to how to establish an appropriate voting constituency.

9. Both the NCUC and the Registrars mention a tiered approach as a possible solution to the dilemma of choosing allocation methods. The NCUC proposes a combination of auctions for "commercial" applicants and random selections for "non-commercial". The Registrars propose a mix of, for example, 4 auction slots, 4 random selection slots and 2 ballot slots per "round".

10. Closely associated with the discussion of allocation methods is the issue of whether to proceed in "rounds" with defined time slots for application, selection and allocation or to perform allocation as an ongoing process. For example, the RyC preference for FCFS is accompanied by a preference for an ongoing procedure, while the IPC preference for a comparative evaluation by nature is more akin to a "round" approach.

11. The ALAC says that ICANN "should accept and evaluate applications on their merits, against objective criteria, as soon as practicable given the natural constraints of ICANN's time, budget, and available personnel". ICANN's operational limits are seen as a factor in determining how many applications should be considered, how often and against which criteria. However, ICANN staff have indicated that operational capacity for processing applications and providing registry liaison support should not be limiting factors in selecting allocation methods.

12. The RyC highlights the need for timeliness, objectivity and predictability in the allocation process. FCFS is the preferred allocation method and comparative evaluations should only be used to choose between applicants with confusingly similar gTLD strings.

13. The IPC favors improved comparative evaluations to enable due consideration of the advantages and drawbacks of each proposal. FCFS is seen as unworkable in view of land rush experiences. The IPC doubts the usefulness of auctions, in view of risks for dominance, bias and overbidding, but a verdict would ultimately depend on the specific auction method. Lotteries are undesirable as a mere chance instrument and also for potential legal reasons. Ballots raise difficult questions on how to constitute an appropriate electorate. The IPC also states that "...the ICANN Board should not abdicate its ultimate responsibility for gTLD allocation".

14. The NCUC rejects the comparative evaluation model as slow, politicized and unpredictable, as shown by experience. Lotteries and auctions are preferred, for non-discrimination, neutrality and objectivity. Auctions would be appropriate for commercial entities and lotteries for non-commercial.

15. The BC calls for a structured allocation method with assessment by a neutral and professional team, thus implicitly a comparative evaluation, to be guided by experience from previous rounds. Auctions are not seen as a satisfactory allocation method.

16. The ALAC prefers FCFS and finds that auctions are in conflict with public interest goals and undesirable for allocation purposes in this context.

C.4 Contractual Conditions

1. There is agreement on several principles regarding contractual conditions for new registry agreements. In essence, this includes that terms and conditions should be published before the application process and, according to the NCUC submission, "a simple, template registry contract that is uniformly applicable to all registries". This view is in line with the RyC statement which says that "applicants should be provided the base contract in advance." The IPC "recommends policies to guide contractual criteria which are publicly available and go beyond the technical aspects of the DNS".

2. The IPC provides detailed commentary on policy compliance arguing that "self-regulation is not the complete answer" and that ICANN should "increase staffing and funding resources to its contractual compliance section in the event registries fail to meet their contractual obligations". Commentary in the ALAC statement says that there should be closer attention to ICANN's Bylaws in developing explicit contractual conditions "including but not limited to those provisions concerning openness, transparency, procedures designed to ensure fairness, and independent review".

3. One particular aspect introduced by the IPC is to bring a subset of the WIPO-2 Recommendations, notably protection of IGO names and abbreviations, into modified UDRP provisions. This area relates to contractual conditions, but could possibly be allocated to a dedicated task force to finalize as a separate track.

4. The NCUC provides a summary of approaches to contractual conditions which says "...We believe that the GNSO should set general policy guiding the contracts . . .The addition of new TLDS should be predictable in timing and procedure, transparent and rule-driven".

5. The RyC argues that "the terms of the latest ICANN-Registry/Sponsor agreements that invoke the GNSO consensus policy recommendations of the process for the approval of registry services fill the void of previous ambiguities with regard to security and stability of registry services. There is minimal, if any, need to develop additional policies to guide the contractual criteria of registry services".

6. The BC calls for policy being developed regarding registry fees, for refinement of the public consultation for proposed contracts and for fair treatment of registries in proportion to their demands on ICANN resources.

7. The ALAC suggests introducing a binding reference to ICANN Bylaw provisions in all gTLD registry agreements with delegated policy-making.

8. In summary, there seems to be converging views calling for predictable and published standard contracts which conform to existing policy on the delivery of registry services.

D. Public Comments

1. Public comments on the Terms of Reference for the PDP on introduction of new gTLDs were sought as required in the PDP procedures. The public comment period ended on 31 January 2006 and comments received are posted on the ICANN web site at http://forum.icann.org/lists/new-gtlds-pdp-comments/ . An overview of public input received has also been posted on the GNSO web site at http://gnso.icann.org/issues/new-gtlds/new-gtld-pdp-input.htm. This overview includes discussions on the General Assembly (GA) list posted at http://gnso.icann.org/mailing-lists/archives/ga/ . Entries on an ALAC wiki web page are posted at http://www.icannwiki.org/ALAC_on_New_TLDs .

2. The following sections analyze the public comments. In addition, the Call for Papers presenters at the Washington DC meeting added texture and diversity to the views expressed, by email, to the public comment period. The presentations are available at http://gnso.icann.org/issues/new-gtlds/new-gtld-pdp-input.htm .

3. Additional public comment periods will be conducted throughout the PDP process and commentary will be included in further iterations of any reports.

D.1 Whether to introduce new gTLDs

1. The public comments generally support the initiative to introduce new gTLDs. Of the seven public comments that appeared to directly or implicitly address the question of whether to introduce new gTLDs, five favored the introduction, while two said the need was not sufficiently strong. Comments by Matthias Jungbauer, Jeff Williams, Chris McElroy, Danny Younger, Elmar Knipp and Thomas Lowenhaupt expressed varying degrees of support for the introduction of new gTLDs.

2. George Kirikos, on the other hand, said the need for new gTLDs is not yet evident, and the current gTLDs could not be characterized as 'full'. He criticized the previous introduction of "hobbyist" gTLDs and proposed that actual use of the existing gTLDs (as opposed to 'parked' names) be measured to indicate how necessary new gTLDs really are. Kirikos also proposed a system of allocation he called the 'Ascension Allocation Method'.

3. Paul Tattersfield also opposed the immediate introduction of new gTLDs, arguing that introducing new gTLDs "could actually decrease competition by reinforcing .com's dominance".

4. Danny Younger, summarizing discussions on the GA list from 5 - 31 December 2005, pointed to interest in establishing new gTLDs amongst various communities, institutions and groups. He said the artificial limitation of new gTLDs has limited the opportunities of small business, non-profits and individuals and entrenched the "dominant corporate players". From this point of view, new gTLDs are necessary to increase opportunities for different types of player.

5. Thomas Lowenhaupt said it was important for cities that these geographic entities be recognized by the DNS. New gTLDs — presumably those using geographic terms such as city names — would, he said, strengthen local economies, create a sense of community, improve safety and give better access to local Internet resources.

6. Other commenters that favored the introduction of new gTLDs said new gTLDs would support database development, and that Internet users are capable of adapting to and determining the value of new gTLDs.

7. The public comments on the draft Initial Report added more implicit support for the introduction of new gTLDs, mostly by suggesting particular strings. Kirk Humphreys proposes to introduce domain names following the three-letter city codes for airports, like OKC for Oklahoma City. Fuad Firudinbayli proposes to use .inaz for various services including education establishments. Alex Ospiov advocates the introduction of the .web. Matthias Jungbaur raises the question of whether IDN strings are considered.

8. On a different note, Mike Norton recommends against having company names as gTLDs and suggests, to counter possible confusion, that introducing a directory function as a gTLD named .icann or .w3c would be useful.

D.2 Selection Criteria

1. The public comments address, in different ways, the selection criteria which could be used in any new round of gTLD applications. GA list commentator Danny Younger indicates that the group had to prioritize its input and focused on "objectionable" criteria which should be removed from consideration. The GA List illustrated views that showed "...it's probably safe to say that most list participants favored an approach that limits criteria only to the technical ability to run a TLD (as the overall broad sentiment with a few exceptions seemed to support the 'let-the-market-decide principle')". This view mapped quite closely to arguments in some Constituency Statements.

2. The GA List also contains commentary that disagreed with the Constituencies that propose to introduce only sponsored gTLDs. The GA List touches on alternate root concerns and domain name collisions or the problems of confusingly similar domains.

D.3 Allocation Methods

1. Amongst the commenters who explicitly considered allocation methods, most appeared to oppose the use of auctions as an allocation method.

2. Danny Younger submitted a summary of excerpts from the GA list discussion on allocation methods, following a week-long discussion of this section of the draft Terms of Reference. He found that "none of the allocation methods reviewed by the group garnered any real measure of support, although each had its own advocates".

3. On the whole, participants in the GA list opposed the use of auctions. The arguments against auctions were that auctions — particularly auctions of strings - may tend to concentrate control, that organizations with greater financial resources could outbid organizations which have shown previous interest in a TLD (for example, .web), and that the highest bidder is not necessarily the best organization to run a TLD. The GA list commenters appeared to explicitly favor a free-market approach to allocation, but also felt that auctions would not deliver the most competitive results.

4. Elmar Knipp argued that comparative evaluations should be used in an auction situation where "startups with fresh ideas would have much lower chances [of success]".

5. As mentioned in section D.1 above, George Kirikos proposed an elaborate allocation method called the Ascension Allocation Method which relies upon Coase Theorem. The rather complex series of string and trademark claims and renunciations required by the method appear likely to increase transaction costs overall. This method did not receive broad support from other commenters.

D.4 Contractual Conditions

1. Submissions from the public comment process also address contractual conditions in some detail. Jeff Williams' submission agreed with that of the ALAC and NCUC in arguing that there should not be "...rounds for applying for new TLDs".

2. Like the RyC and NCUC, public comments advocate the use of "thin" contracts and that there could be some improvements made to ICANN's registry agreement compliance program.

3. Some public comments were diametrically opposed to the position set out by the IPC. They rejected intellectual property owners' "priority rights with respect to generic words, and participants further expressed their ongoing dissatisfaction with ICANN's failure to establish a compliance program". In essence, public commentators wanted to facilitate freer market choice in the selection of new gTLDs; wanted limited selection criteria and simplified contractual arrangements.

E. Submissions on Call for Papers

1. In total, 11 submissions were received in response to the Call for Papers. These submissions are summarized below in relation to the issue areas. The full texts of all submissions are available at http://gnso.icann.org/issues/new-gtlds/new-gtld-pdp-input.htm

2. In addition, many of the respondents to the Call for Papers gave presentations at the Washington DC meeting, adding further explanations to the views expressed. The presentations are available at http://gnso.icann.org/issues/new-gtlds/new-gtld-pdp-input.htm.

E.1 Whether to introduce new gTLDs

1. John Levine, Paul Hoffman (et al) advocate an annual release of 50 new gTLDs at once, stating that such timing wouldn't give any applicant an advantage over the others except in the inherent semantics of the chosen string. They also claim that technical qualifications could be handled separately, in line with other proposals calling for a separate registry operator accreditation scheme. Mr Levine made a presentation at the Washington DC meeting which amplified his views.

2. Dirk Krischenowski (.berlin) sees a multitude of reasons to introduce new gTLDs. These include views also found in other contributions, including avoidance of alternate root scenarios; to diversify ICANN's funding base and to promote local development (which maps to some public comments received). In addition, new gTLDs would increase diversity and choice with decreasing speculative pressure as a possible consequence. Finally, Krischenowski states that there is explicit demand from the community, claims that there is no risk of a land rush for new gTLDs and does not see a need for a limit on the total number of gTLDs, unlike the IPC community who foresees exactly the opposite scenario. Mr Krischenowski reiterated his views in his presentation to the Washington DC meetings.

3. Angela Stanton supports introducing new gTLDs and proposes to redesign the gTLD structure in line with the original taxonomic purpose as a directory. This would call for introducing constraints in registrations for currently unrestricted gTLDs but would substantially reduce the need for defensive registrations. It may enable using the same domain name strings for different registrants in different gTLDs. Ms Stanton made a presentation to the Washington DC meeting to expand on her views.

4. Rahul Goel and Ashutosh Meta support a measured introduction of new gTLDs with increased user choice as their main rationale. To further increase choice for users in developing countries, they advocate that a company with an existing domain in one gTLD should not be allowed to register in any other gTLD, an approach somewhat similar to Stanton's above. Messrs Goel and Meta made a presentation to the Washington DC meeting and explained to the Committee the importance of domain name affordability.

5. Hedd Gwynfor (DotCym) supports the introduction of new gTLDs and advocates that priority be given to sponsored TLDs for cultural purposes. DotCym is an organization interested in establishing a gTLD for Welsh cultural and language interests. Gwynfor refers to the .cat gTLD as "a significant step toward the allocation of sponsored TLDs for single language communities...and is a precedent to which other minority or stateless language based groups can now aspire...".

6. K Bhonsle argues for a limited introduction of new gTLDs with a particular focus on previously deprived users and applications catering to basic needs in a primarily agricultural environment.

7. Peter Gerrand is in favor of introducing new sponsored gTLDs and finds reasons to revert to the originally intended structuring of the domain name space with distinctive, defined purposes for each gTLD. Dr Tonkin gave an overview of Mr Gerrand's paper at the Washington DC meeting.

8. Danny Younger, who also made a presentation to the Washington DC meeting, makes an analogy between the DNS and the "zoning" of city areas. The ICANN Board is seen as the equivalent to a zoning board of such a city. Younger claims that as the need for new zones inevitably appears, zoning is necessarily done in a measured manner after considerable consultation and experiences should be drawn from successful zoning management in the physical world. He also acknowledges that experimentation in zoning is a necessary aspect when applying this model to the DNS. ICANN's primary responsibility to act in the public interest should be focused towards the public at-large rather than towards separate communities or organizations. Inspiration for future zoning could well be drawn from ICANN's Strategic Planning Issues Paper. Younger also makes reference to a paper by Stewart & Gil-Egui on application of the Public Trust Doctrine to Internet resources, implying the obligation to preserve resources that are crucial for intergenerational equity.

9. CORE supports the introduction of new gTLDs for similar reasons to other submissions including counteracting the proliferation of alternate roots; ensuring that .com doesn't become the de-facto or "virtual" root and promoting innovation and creative new paradigms. The CORE submission also advises using lessons learned from earlier rounds of new gTLD introductions. Mr Staub, on behalf of CORE, made a presentation to the Washington DC meeting.

10. Ross Rader & Elliot Noss (TUCOWS) support the introduction of new gTLDs. They propose a new distinction between chartered and unchartered gTLDs to replace current distinctions. They foresee migration of existing gTLDs to new categories. They provide a detailed proposal for a process introducing a gTLD from application to renewal of a registry agreement. Mr Noss presented his views, with Mr Rader, to the Washington DC meeting.

11. Bret Fausett supports the introduction of new gTLDs as an ongoing process with a taxonomy determined by market forces. The rationale for his stance include the desirability to increase registry-level competition; to cater for the needs of prospective registrants; to enable the evolution of the Internet and to better provide for underserved markets. An array of suggested principles is provided to guide various aspects of the introduction.

E.2 Selection criteria

1. Levine & Hoffman explicitly recommend avoiding string exclusivity and allowing parallel strings that are essentially synonymous, like .tooth and .teeth. They also state that a directory approach is bound to fail, especially as search engines are the preferred way of approaching the Internet for a chosen topic, rather than looking up by TLD. A few more unrestricted gTLDs would be welcome and also some certified gTLD for particular purposes, where certification matters (similar to .edu). Creativity in usefulness should be the objective. This approach can be seen as a mix between the IPC proposals and those of the NCUC.

2. Krischenowski supports the selection criteria used in previous rounds and suggests adding supporting criteria like positive recognition by government, potential to foster economic development, potential to promote technical development and socially desirable effects (for example, SME promotion, education and support of local culture). Support of local culture is also recommended by Gwynfor above.

3. Goel & Meta put an emphasis on selection criteria and other aspects that would support less developed countries; differentiated registration fees, country-wise assignment of IP addresses and geographic redistribution of the root servers. This view is in direct contrast with the RyC claim that there is little evidence to support preferential criteria for new gTLDs without "research that supports their claim including the research methodology supplied". The NCUC's view is that "...the best way to do this [expand the use of domain names] is to make ICANN's selection criteria as simple, predictable and content-neutral as possible. Such a politics-free environment would make it much easier for innovators, from whatever locations, social origins or economic status, to propose and implement new ideas".

4. Gwynfor claims that the ISO 639-2 three letter codes for languages should be reserved for future gTLDs for the respective languages.

5. Gerrand recommends that more emphasis be put on the integrity with which the sponsored gTLDs enforce their eligibility condition and lists suitable selection criteria. This comment is equally valid for section E.4 contractual conditions. As to the gTLD strings, he suggests reviewing the policy on ISO 3166-A3 three letter country codes which are currently reserved. He also recommends enabling the use of ISO 639-3 three letter language codes for future gTLDs.

6. Younger sees a development where new gTLD applications are viewed on a case-by-case basis, only being deemed acceptable when the rationale for each proposed gTLD becomes self-evident.

7. CORE illustrates its reasoning with four hypothetical gTLDs for specific applications and with clearly defined gTLD communities. The applicant would need to be representative of this community as an obvious criterion. As the examples go, the presence of a sizeable community is a criterion that demands case-by-case analysis, price per registration is irrelevant as a criterion while productive use of the DNS at the top-level is most relevant. Inherent security requirements in a particular gTLD application may modify the desired score for other plausible selection criteria. In short, the examples are sponsored gTLDs calling for case-by-case judgments.

8. Rader & Noss introduce a distinction between "Delegant", coordinating the activities of a gTLD and "Registry Service Provider", taking care of the technical operation of the gTLD. This idea would mean that the Delegant applies for the gTLD and, if approved, would have it operated by an accredited Registry Service Provider. This would formalize an established practice and would call for different selection criteria for the two separate businesses. In a similar way to the established accreditation of Registrars, the Registry Service Provider should be subject to ICANN accreditation, by applying the current technical selection criteria with a few amendments. This is consistent with other proposals about separate registry accreditation. The Delegant should be requested to abide by chosen aspects of RFC 1591, without restrictions or preferences imposed concerning business plans. These parties would both have agreements between themselves as well as with ICANN. In addition, gTLD string restrictions would apply with no digits allowed and with no confusing similarities to existing strings. Noss and Rader recommend special rules for trademarks as gTLDs which accord with the IPC's ideas.

9. Fausett sees a well-defined market, be it large or small, as a criterion while adding that an application from an able and willing provider is a main indication that such a market exists. Free market entry should be a guiding principle to the greatest extent possible, while recognizing that user confusion and defensive registration are undesirable.

E.3 Allocation methods

1. Levine & Hoffman see both auctions and lotteries as viable allocation methods, while indicating that ICANN should not profit by windfall gains from allocating new gTLDs. They acknowledge that there will be secondary trading in allocated gTLDs. An auction with the [N] highest bids getting their [N] favorite domain strings is a possibility, with the proceeds going to a worthy cause. This may be combined with trademark limits so that only IBM can get .ibm while still needing a winning bid to do so. This can be combined in a tiered approach with a lottery for five or ten names with only non-profits eligible.

2. Krischenowski prefers revolving application windows, preferably 2-4 each year, and a predictable timeframe for the approval process, not surpassing 6 months. He sees no need to structure gTLDs in different classes and contends there should only be an "open" class of gTLDs. Allocation should rely on comparative evaluation and FCFS once the evaluation is positive. Auctions and lotteries are to be avoided and he includes a reference to the effects of the UMTS auctions.

3. Goel & Meta state that comparative evaluation is the most appropriate allocation method, with short-listing of all that satisfy the minimum criteria followed by prioritization based on stability and price.

4. CORE, basing its statement on four examples of sponsored gTLDs, is clearly in favor of comparative evaluations and sees little or no scope for other allocation methods. The allocation process should be recurring, with at least two application windows per year, linked to ICANN meetings. CORE suggests that a standing gTLD applications task force is established to avoid bottlenecks. It ways that preliminary applications should be received for publishing, without review, as a step to enable possible consolidation of similar applications.

5. Rader & Noss advocate an ongoing, self-financed allocation process. They suggest that applications would be received at any time. FCFS should be used as the main allocation method with sealed bid auctions to resolve string contention. Fee adjustments could be used to manage an applications flood and a development fund, built from fee surpluses, could provide financial assistance to deserving applicants.

6. Fausett does not address allocation methods explicitly but it is inherent in the submission that FCFS is foreseen in an ongoing allocation process, for the applications that meet the selection criteria.

E.4. Contractual conditions

1. Krischenowski supports the idea of a separate process for accreditation of registry operators. This is consistent with ideas suggested by others.

2. Both Bhonsle and Goel & Meta propose that sun-rise periods be compulsory for all new gTLDs in order to safeguard IP holders' interests.

3. CORE finds the current contractual framework for sponsored gTLDs is an appropriate model. It highlights the need for a stable system of checks and balances for the sponsored gTLD's delegated policy-making authority.

4. Rader & Noss propose omit price controls for new gTLDs and to relax, and eventually eliminate, price controls for existing gTLDs in reverse chronological order from their initial delegation. Presumptive renewal should be a standard provision, with remaining cancellation powers for ICANN in case of breach of the contractual terms. A sanctions program should be developed for handling contract violations regarding all agreements. Sunrise periods are discouraged. Presumptive renewal of agreements should be the rule.

5. Fausett suggests regular use of escrowing registration data. Back-up registries should be selected through market mechanisms and published migration plans are useful provisions that may lessen the impact of registry failure.

F. Recommendations

1. Given that this Initial Report is an evolving draft, this section is intended to enable open interaction as further drafting work progresses through the Wellington meetings. Current findings and unresolved issues are outlined. These will be further developed through work and consultations within the GNSO itself, in co-operation with other ICANN Supporting Organizations and Advisory Committees and in consultation with the wide Internet community through scheduled public comment periods.

2. It appears that there is support for the introduction of new gTLDs. No submission argued that there should be no additional gTLDs, even if views on the addition rate and conditions for adding new gTLDs vary widely. Most submissions recognize that new additions to the root are within the scope of ICANN's technical mandate, are necessary if ICANN is to meet its core mission and values (particularly with respect to competition and usability) and are part of ICANN's normal operations

3. .

4. There is disagreement about how many new gTLDs should be introduced and at what pace. There is also disagreement about whether gTLDs should be sponsored or unsponsored and how new strings should be allocated. While the constituencies use these distinctions, it is equally important to address whether new gTLDs should be restricted or unrestricted.

5. There is general agreement that standardized contractual conditions for registry operations should be published prior to any agreement being signed. There ought to be an improved compliance regime and there should be minimal interference with consensus policy positions.

6. Further analysis and discussions could be pursued along two slightly different scenarios. The first is a very limited introduction which is restricted to one kind of gTLDs. The other would feature a broader process which could accommodate a more diverse range of applications.

7. Secondly, further analysis is required about the operational impact on ICANN of introducing new gTLDs. This would enable a better understanding of the full costs of introducing new TLDs including, for example, legal counsel, operational and policy inputs and Board consideration time.

8. Thirdly, fact-based market analysis would be useful to inform decisions about the desirability of introducing new gTLDs from an end user perspective. Such analysis could provide better understanding of unmet demand patterns and potential effects on competition.

9. Fourthly, specific ideas and input that have not been addressed by others should probably be tested early on for potential consensus. The proposal from Rader & Noss to reclassify gTLDs into "chartered" and "unchartered" could be such an example.

10. A particular aspect introduced by the IPC is to bring a subset of the WIPO-2 recommendations, notably protection of IGO names and abbreviations, into modified UDRP provisions. This relates to contractual conditions and could be allocated to a dedicated work group to finalize as a separate track.

11. Further work still needs to be done on some elements of the Terms of Reference as not all questions were answered in the submissions. It would be helpful if the Council could identify areas where further work is necessary and advise about how it would like those areas addressed.

G. Analysis of Impact on Each Constituency

1. This area of work needs particular examination after the next version of the Initial Report is finalized.

2. The Constituencies need to formalize their impact statements and add these to consideration in the next round of PDP input.

H. Relationship to the PDP on IDNs

1. As a parallel IDN Policy Development Process is in progress, references to IDNs in this document are scarce. In deciding to divide the work on new TLDs into two parallel PDPs, the GNSO Council clearly stated the strong interdependencies and the necessity to merge the outcomes. This approach calls for addressing the issue of IDNs at the top-level in the further discussions of this draft Initial Report.

2. It is worth noting here that the BC has recommended that no new gTLDs be introduced unless there are IDN-enabled TLDs and are conducted under a sponsorship regime.

I. Relationship to PDP FEB 06

1. The new PDP on Policies for Contractual Conditions for Existing TLDs is now underway. There are direct co-relations between the work of that PDP and the work being conducted here on contractual conditions for new gTLDs.

2. To ensure consistency, a deliberate plan to put the work together needs to be formulated as is the case in the IDN PDP

J. Next steps

1. The next steps for this PDP are the proposed working sessions on Saturday 25 and Sunday 26 March in Wellington. It is proposed to release the final version of this report on 2 May 2006 which will include input from the Wellington meetings.

2. On 2 May 2006 a twenty day public comment period could commence ending on 24 May 2006. It is also expected that the final version of the Initial Report will be sent to other Supporting Organizations, the Governmental Advisory Committee and the ALAC for their input.

3. On 2 June 2006 the draft Final Report could released in time for consideration at ICANN's Marrakesh meeting.