Evolving ICANN Policy

Draft

March 15, 2006

[edit] Evolving ICANN Policy

By David R. Johnson

David Johnson

The staff of ICANN recently released a document summarizing the views of multiple constituencies and individuals on the question of whether and how to create new Top Level Domains. There appears to be general consensus that new domains should be created. But there are many differing views on many different questions regarding how the process should work, what contractual obligations should be imposed on new registries, and how these decisions relate to ICANN’s funding and policy formulation process.

The questions are many, but they ought to be simple to resolve. ICANN was created to increase competition, to preserve the technical stability of the internet and to establish limited non-technical rules in the event that there is broad consensus in support of such generally applicable requirements. With these guidelines in mind, the answers to most of the questions posed by the new TLD policy development process become evident. If a registry proposes to open a new TLD, possesses minimum technical qualifications, provides for escrow to protect against failure, pays a fee that covers the costs of the process, and agrees to abide by consensus policies, then the only remaining question is what basis ICANN might have for saying “no”. Individual constituencies or parties might favor imposing additional rules (such as requirements applicable to choices of strings, or protection of trademark owners), but such non-technical policies couldn’t legitimately be imposed by ICANN in the absence of consensus support (and, even then, only if they fit within the narrow confines of ICANN’s policy responsibility, as defined in ICANN’s existing registry agreements). The choice of allocation process and fee levels should favor minimizing the costs to would-be registries and avoiding the creation of a windfall for ICANN that the community hasn’t approved. The pace of allocation should be as fast as feasible in light of technical constraints, if any, so as to favor competition. The market will ultimately determine which new TLDs are needed.

Any policy choices other than this simplest set, which best fit core ICANN mandates, would represent an abuse of ICANN’s institutional levers to serve limited interest groups – a form of the “capture” that ICANN was deliberately (not necessarily successfully) designed to avoid. Some may argue that ICANN should strive to create the “best” design for the name space and protect consumers against confusion. But ICANN is not a consumer protection agency and no one can know in advance what the best design will turn out to be. Some will argue that expansion of the name space will tax IANA resources. But there is no support for claims regarding technical impact (after all, the .com zone can handle millions of subdomains and it operates the same way as the root zone). And any staff burdens would be minor compared to the benefits if new TLDs took hold. Some will claim that there is no demonstrated market demand for new TLDs. But that’s often true with new ventures and we should rely on the decentralized judgments of entrepreneurs to decide which experiments are worth trying, just as we do for every other technical advance.

Unfortunately, this multi-faceted problem may not be so easily solved. Anyone who does not agree with the minimalist view of ICANN, for whatever reason, or who sees ICANN as a regulator that must strive to make the “best” choices on behalf of the “internet community” whenever an issue swims by, will reject the solution outlined above. And many different groups and individuals who don’t share the minimalist view will propose just about every possible different flavor of policy choice. There does not appear to be a principled or legitimate basis (other than the minimalist vision) on which to choose a particular combination of policies. ICANN’s Board is not elected, so there is no way to justify their idiosyncratic judgments as the irrational but politically “legitimate” choices of representatives. There is no rational means by which to weigh the preferences of one group against another. And no particular answer is “right” as a purely technical matter because all the variations embody differing political values or contested economic theories. There is no law requiring one set of choices over another (aside from the “law of ICANN” that arguably points towards the minimalist policy set).

Just to be clear on how complex the issues are, once one strays from the minimalist choice, consider the questions presented for simultaneous resolution. These include:

• Whether to open new TLDs at all. (The only point on which there might be said to be agreement – though much of that agreement is conditional.)
• What type of new TLDs to allow (generic vs. sponsored; unchartered vs. chartered; ascii vs. IDN; global vs. territorially based)
• At what pace to make slots available.
• Whether to require a demonstration of technical capability.
• Whether to require a demonstration of financial means.
• Whether to require a demonstration of market receptivity.
• Whether to require establishment of protections against failure.
• Whether to require a particular contract document.
• Whether the contract should be disclosed in advance of the process.
• Whether to prohibit certain strings or types of strings.
• How to allocate strings if multiple parties want the same one.
• Whether to require some kind of “differentiation” of strings and how to measure this.
• What fee to charge for going through the process.
• Whether to allocate by auction, lottery, comparative hearing or first come first served – or a mixture of these and perhaps other techniques.
• Whether to proceed promptly on any application or to have staged “rounds”.
• Whether to give some preference to those whose applications were “deferred” in prior rounds.
• Whether to require all new registries to use all and/or only accredited registrars as their market channel.
• Whether to require all new registries to implement the UDRP.
• Whether to require, encourage or discourage the use of sunrise periods (relating to trademark owners).
• Whether to require all new registries to adopt current “whois” policies and practices.
• Whether to charge the same ongoing fees to all registries – or, perhaps, give some preference to startup registries.
• Whether the contracts should provide for automatic renewal in the absence of breach.
• Whether semantic objections by particular parties to selected strings should be taken into account, and if so how.
• Whether all new registries should be required to follow current transfer practices.
• When to begin the process and what conditions must be met before doing so (e.g., resolution of all policy issues relating to internationalized domain names (IDNs)
• Whether to give preference to some kinds of applicants with regard to some kinds of TLDs (e.g., allocating new IDN registries to ccTLD registry operators for countries with a particular relationship to the script in question).
• How to allocate the burden of persuasion on various issues (e.g., should ICANN have to have a good reason to reject an application, or should the applicant have the burden of persuading ICANN in advance that its approval will achieve some specific goals such as adding value to the name space)

If your head doesn’t hurt, you don’t understand the problem. What can be done to resolve multi-party multi-issue policy disputes such as this? Once one departs from a minimalist vision (ICANN should have a very good reason, within a narrow band of permissible policy areas, to say “no” to any application), the value clashes are genuine. They may not be susceptible to political log-rolling, because large factions gain from continuation of the status quo (failure to reach any generally accepted resolution). Neither the assembled “community” nor the Board are demonstrably representative of the large group likely to be affected by the decision. The affected interests that are “represented” by particular spokespersons differ markedly in scale and scope. So even putting the alternatives up for a vote by everyone who shows up in the ICANN process would not convincingly resolve the matter. So we are presented with a deep, complex clash of heterogeneous goals and values and no rational, legal or even legitimate political means to resolve that clash. I guess we could throw dice. But, obviously, few would find that satisfying. We want a “legitimate” decision – one that will satisfy even the losers (everyone will be in part a loser) that there was a shared, collective “good reason” for adopting one set of policies in preference to all the others. What should ICANN do?

The answer may be found by reference to a growing body of science regarding complex adaptive systems. When nature confronts a complex problem, such as finding the “right” set of traits to enhance the fitness of an organism (even when the traits are many and multiply partially conflicting), it doesn’t engage in top down design. It evolves design from the bottom up by means of replication and differential selection. The only “vote” it knows how to take is the environmentally-driven selection of traits that survive in the next generation. Using this method, nature can respond to constantly shifting fitness landscapes. Collections of organisms with differing traits coexist – but species emerge, at any given time, by virtue of their ability to mate (and mutate) in such a way that their particular traits survive.

Easy enough for nature, you might say, because the criteria for selection are naturally given by the world as it is (including by the presence of other competing and cooperating individuals and species). What could this have to do with resolution of a complex policy debate? Isn’t the very problem that multiple parties disagree about what values and goals should be used to “select” one set of policies as opposed to another? How could we test the “fitness” of a particular policy or policy set without giving undue preference to one constituency’s values and goals?

The good news is that there is a way to do just that: by building a model of the world that most parties engaged in the debate can agree upon. This would only be a model, of course. But it would be able to serve as the selection filter for the (computer based) evolution of complex policy sets. And, if we can get agreement on the model (not easy, but easier than getting agreement on a final policy set), then we should be able to get agreement (or at least grudging acceptance) of the policy set(s) that survive an evolutionary process that uses it as the fitness screen.

The even better news is that most disagreements about these complex policy matters concern which policies should be chosen, not the underlying questions of (1) what effects any given policy is likely to have; (2) which parties will suffer or benefit from particular effects, and to what degree; (3) which policy choices are incompatible with or reinforcing of other policy choices; or even (4) what weight should be given to the interests of various parties in light of the number of such parties and the nature of their interests. It may actually be possible to build a general model that most ICANN participants would agree reflects the likely consequences of various decisions – and to use that model and an evolutionary algorithm systematically to search the solution space for optimal choices.

The fourth question seems least likely to attract general agreement, so let’s discuss that first. Selfish factions would seem to want their own values and goals to be given predominant, if not exclusive, weight. That’s true enough when we are dealing with a political battle or a marketplace decision – where all are tacitly given permission to pursue their own goals even when achieving those goals may harm other interests. But the social world does not consist only of politics and markets. We have another concept – civic virtue – that presupposes that individuals and groups can be asked to think about the greater good. And, when they do so, they inescapably form an implicit view about what weight should be given to the interests of various groups.

When thinking about what weight *should* be given to the votes of individuals (or corporations) in a democratic process, we have gradually coalesced around an agreement that every rational adult citizen should have his or her vote counted equally. When thinking about what weight *should* be given to the interests of a potentially guilty but merely accused criminal defendant, we generally agree that an accused person’s interests should be given relatively great weight in comparison with the public’s dispersed interest in efficient operation of the judicial system. When thinking about transactions in a competitive and otherwise lawful market, we generally agree that interests should be weighed by reference to dollars possessed (or offered for transfer) by the parties. Even when thinking about difficult regulatory questions, such as health care or environmental regulation, we still have general agreement that the interests of patients or of children facing environmental hazards should weigh more heavily than those who are merely pursuing commercial advantage (though, of course, the interests of those building businesses also count).

In short, once we adopt the “civic virtue” stance, we don’t actually disagree very much about how to weight the *importance* of the interests of various affected parties. We may still disagree about policy choices. But we generally agree, when we are not being selfish, on relative weighting of different types of interests. One reason for this is that we all have multiple interests and roles. When we speak as advocates for a company or political faction, we naturally stress the importance of its goals or values. But, off the job, we are all investors and consumers, parents and employees, beneficiaries of sound regulation and adversely affected by unwise rules. Some few may be rich enough to escape the consequences of any given set of policies – but we all agree (when wearing our civic virtue hats) that corporations shouldn’t be allowed to vote. Many may be too poor to protect themselves – but their interests are widely valued even by those who can. So we should actually be able to get a broad consensus, in advance of the selection of any particular policy, on the relative weight to be given to various “interests”.

Let’s test this proposition in the context of the ICANN dilemma. At first glance, it seems clear that the IP constituency will fight to the death for the proposition that trademark protection trumps the interests of those who want to roll out new TLDs, whereas the registrars will contend that enhancing their business model should take priority over the interests of ISPs or registries or general businesses. But these seemingly irreconcilable differences all concern a limited question that we can avoid asking until much later in the process: what particular set of policies should be *adopted* by the ICANN board. We should instead ask a different question: which widely shared interests are entitled to more weight (respect) than others? The civic virtue stance first asks not what the rule should be but, rather, (1) what are the various interests likely to be affected by any particular choice among possible rules? and (2) are any of these interests obviously entitled to more weight than others?

The answers to these questions are much easier to come by. First, we can list all of the various “interests” – really social goals – that might be affected by one or more of the many choices presented in connection with creating (or not) new TLDs. The list would include:

• Registries (as businesses, considering both shareholders and employees)
• Registrars (as businesses, considering both shareholders and employees)
• Internet Services Providers (as businesses, considering both shareholders and employees)
• Trademark owners (large and small) (as businesses, considering both shareholders and employees)
• Aspiring new TLD operators (as businesses, considering both shareholders and employees)
• Internet users (insofar as interested in stability and competition)

The list would, by general agreement, probably not include several other interests that are regularly mentioned (or claimed to be represented by particular spokespeople). The “interests of the internet community” or of the “ICANN community” wouldn’t make the list, because this is an entirely diffuse array of conflicting interests and doesn’t add any clarity to the analysis. The interests of ICANN itself as an organization would be excluded on the ground that ICANN as an organization exists only to serve the broader social interest – not its own corporate goals.

Surprisingly, we would also exclude “registrants” from the list – because registrants are such a diverse group that thinking about their interests also doesn’t add anything meaningful to the analysis. Some registrants favor strong trademark protection, some oppose it. There is no reason to distinguish between present and future registrants, because these will generally either be the same people or have very similar goals. (As a group, registrants might be thought to be concerned about things like the price of registration and transfer policies. But, in the presence of competition, the marketplace can provide these goods. Registries and Registrars as businesses will only prosper by serving the needs of registrants, over the long term. We ALL have an interest in preserving competitive markets and technically stable DNS operations. Once those goals are achieved, and taken as a filter for particular policy choices, there is no separate need to weigh the interests of consuming parties, as such – because the market will do that for us.)

We would also exclude “law enforcement”, on similar grounds. The police are supposed to protect us, not their own interests, separately considered. And we ALL share an interest in effective, efficient and fair law enforcement (including protections of civil rights).

The list would not include the “At large constituency” or the “non-commercial users constituency” because, again, no one can credibly claim to represent these large and diverse groups and because, so long as we only consider policies that preserve technical stability and competition in the marketplace, the interests of such groups will be served by the companies that seek their business.

So there you have it. A surprisingly short list of interests that will be directly affected by any given set of policies on new TLDs that ICANN might come up with and that are not so broad or diffuse or conflicted as to cancel out. All but one “interest” on the list represent specific types of businesses whose interests inherently conflict and that would be differentially affected (favored or disfavored) by any given set of rules. The “general internet user” category is explicitly constrained to consider only two of the types of interests hat such a user might have – technical stability and competition -- interests that are not assuredly served by the marketplace in the absence of regulation. I believe that we could get general agreement that this list includes all of the different types of interests likely to be directly affected by ICANN’s choice among possible new TLD policies – and that the excluded interests are either too diverse to be counted as coherent sets of goals and values or would be served by the attainment of the core goals that we all share.

The next question to be addressed is whether any of these interest should be given more “weight” in our fitness model than any others. From a civic virtue (not a partisan) point of view, the answer seems obvious. The first five are just different ways of making money. There may be more or fewer companies in any given category, but even that difference might be offset by consideration of the number of shareholders or employees with a beneficial stake in particular business models. There is really no principled basis on which to conclude, for example, that trademark interests are more important to us (as a society) than ISP interests or registrar or registry interests – especially when one assumes that the generally shared interests in stable operation of the internet and competition in the marketplace will be counted by other means. In contrast, there is clearly a good reason to give the general internet user’s interest in these two goals *much* more weight than that given to the goals of any particular type of business. Even a trademark owner or registry operator or registrar, once off duty, would readily admit as much. There would clearly be many many more people adversely affected by a loss of stable internet operations or the elimination of competition in the relevant markets than would be affected by any kind of adverse impact we can imagine befalling any of these other “interest groups” as a result of any ICANN policy. It may not be easy to get agreement on an exact multiple, but the size of the multiple used in our model may not matter very much. I’ll pick 5. You can rerun the evolution simulator yourself with a different value if you disagree – and I’ll be that you will come to very similar results.

Now that we’re done with the hard part, we just need to decide (and get agreement on the answers to) the remaining issues: (1) what effects any given policy is likely to have; (2) which parties will suffer or benefit from particular effects, and to what degree; (3) which policy choices are incompatible with or reinforcing of other policy choices? To tackle number 1, we just need to list all possible policies (as above) and all possible effects that might be suffered by the interest groups we are considering. The nature of the effects to be considered is pretty obvious – we’ve already shortened the list considerably by eliminating interests that are indirectly served or disallowed (because counting them as distinct interests would conflict with fiduciary duties) or unduly heterogeneous. For the first five groups, we consider whether the policy will make the companies in question richer or poorer. For the two key interests of the internet users, stability and competition, we can readily predict whether a particular policy will have small, medium or large, positive or negative, effects. And it is pretty clear which policies contradict or reinforce each other. The mappings are relatively uncontroversial (and fully reported in the software model). We may argue about which policies should be adopted, but there could be general agreement (by reasonable people adopting the civic virtue stance) as to who would win or lose, and by how much, if particular policies were selected.

So now we have a policy “fitness” model that can be used to evolve an optimum (not necessarily the very best) set of policies! If people (adopting a civic virtue stance) can generally agree that this model doesn’t omit or misstate the conflicting interests that are in play, or the nature and extent of relevant effects, then they should be willing to live with any set of policies that an evolutionary algorithm, using this model as a fitness screen, selects. We can start with a population of randomly generated policy sets (policy “organisms” specified by a list of policy “genes” that are created with virtual dice). We can then score each policy “organism” by simply multiplying the numbers that reflect the effects (adjusted to give more weight to user interests, as noted above, and to eliminate conflicting policies). And we can then rank the population by score, kill some portion of the “less fit” variants, allow the surviving policies to mate (exchanging genes in the process of creating new offspring), and throw in a little random mutation (policy cosmic rays, as it were) for good measure. And then we would do it again. Within a quite finite number of generations we can confidently expect to reach the top of a fairly optimal hill in the policy fitness landscape. We may never find the highest hill. But we can run this virtual evolution process over and over, as many times as we like, until we are satisfied that we have searched a large portion of the possibility space and that the policy sets we have evolved are better – within the meaning of the model that we have constructed by consensus – than any variation that anyone else can come up with. (All the variations will lead downhill in the policy fitness space, so defined, by adversely affecting some of the interests without producing larger countervailing benefits for others.)

I’ve run this model, and here are the results:

• open new TLDs
• open chartered and unchartered TLDs
• open territorial TLDs and gTLDs
• gate at 100 per year
• require demonstration of technical ability
• require demonstration of financial means
• require escrow protections against failure
• disclose contract in advance of process
• prohibit confusing strings (apply UDRP)
• allow current stringholders only existing remedies (like UDRP) for any confusion
• charge a cost based fee
• provide subsidy for would be registries without means
• use first come first served to allocate slots
• proceed promptly whenever proposal is received
• require all new TLDs to adopt current whois policies
• provide for automatic renewal if not in breach
• begin process before all IDN issues resolved
• require ICANN to have a good reason to turn an applicant down

I didn’t pick these answers, the model did, based on the list of choices and effects set forth in Appendix A. You might disagree with my views regarding what the choices are, what effects they will have, or what weight should be given to different interests. (The current run didn’t include every possible policy choice. See Appendix A for a list of those included.) I would encourage you to change the settings (to some set that you think most people ought to be able to agree upon) and see whether you can evolve a more optimal solution. At a minimum, differing results from differing settings will help to clarify where the substantive disagreements lie – helping to move the ICANN process away from mere strategic bargaining based on illegitimate leverage.

The key to persuading ICANN and its constituents to accept the results of this model and genetic algorithm process is (1) to ask whether anyone can come up with a better alternative that demonstrably makes everyone better off and (2) to compare the alternative means that might be used to reach a decision. The selection of new TLD policies might be made by a flip of a coin, or by bowing to the unwillingness of those who benefit from stasis to reach any agreement, or by some supposedly reasoned analysis of what are really irreconcilable conflicts among goals and values, or by voting engaged in by people who have not been elected by anyone and who are not in any sense representative of all affected parties, or by allowing money to talk, or by a judge applying no established law. All of these are manifestly less legitimate (less satisfying) means for reaching a decision than the use of an agreed upon model of the nature of the conflict among relevant interests to engage in a Darwinian selection process. A person who would otherwise lose the battle for their own interests if the various forms of luck or force described above were used *ought* to prefer the results of the model, sight unseen. Even the person who is relatively sure that he or she has the leverage to force a favorable result *ought* to prefer the results of the model once s/he adopts the “civic virtue” perspective – asking what is best for all interests, appropriately considered, in the aggregate.

If this evolutionary process can work in the divisive ICANN context, it might work anywhere. The software model is available upon request. Give it a try.

[edit] Exhibit A

Settings Used in the Model

• open new TLDs LEADS TO increased competition
• open new TLDs LEADS TO positive effect on would be TLD operators
• don't open any new TLDs LEADS TO decreased competition
• don't open any new TLDs LEADS TO negative effect on would be TLD operators
• open only sTLDs LEADS TO decreased competition
• open only sTLDs LEADS TO negative effect on would be TLD operators
• open chartered and unchartered TLDs LEADS TO increased competition
• open chartered and unchartered TLDs LEADS TO positive effect on would be TLD operators
• give preference to IDN TLDs LEADS TO negative effect on would be TLD operators
• open territorial TLDs and gTLDs LEADS TO increased competition
• open territorial TLDs and gTLDs LEADS TO positive effect on would be TLD operators
• open only gTLDs LEADS TO negative effect on would be TLD operators
• gate at 10 per year LEADS TO negative effect on would be TLD operators
• gate at 100 per year LEADS TO increased competition
• gate at 1000 per year LEADS TO increased competition
• gate at 1000 per year LEADS TO negative effect on ISPs
• gate at 1000 per year LEADS TO negative effect on would be TLD operators
• require demonstration of technical ability LEADS TO increased stability
• require demonstration of financial means LEADS TO increased stability
• require demonstration of market receptivity LEADS TO decreased competition
• require demonstration of market receptivity LEADS TO negative effect on would be TLD operators
• require escrow protections against failure LEADS TO increased stability
• disclose contract in advance of process LEADS TO positive effect on would be TLD operators
• pre-approve strings LEADS TO negative effect on would be TLD operators
• prohibit confusing strings LEADS TO positive effect on current registry business model
• prohibit confusing strings LEADS TO positive effect on ISPs
• prohibit offensive strings LEADS TO decreased competition
• allocate strings by auction if conflicting claims LEADS TO negative effect on would be TLD operators
• require clear differentiation of strings in advance LEADS TO positive effect on current registry business model
• require clear differentiation of strings in advance LEADS TO negative effect on would be TLD operators
• allow current stringholders only existing remedies for any confusion LEADS TO positive effect on would be TLD operators
• charge a cost based fee LEADS TO positive effect on would be TLD operators
• auction slots LEADS TO negative effect on would be TLD operators
• provide subsidy for would be registries without means LEADS TO increased competition
• conduct comparative hearing on which proposals best serve dns LEADS TO decreased competition
• conduct comparative hearing on which proposals best serve dns LEADS TO negative effect on would be TLD operators
• use first come first served to allocate slots LEADS TO positive effect on would be TLD operators
• use lottery to allocate slots LEADS TO negative effect on would be TLD operators
• proceed promptly whenever proposal is received LEADS TO positive effect on would be TLD operators
• proceed in staged rounds LEADS TO negative effect on would be TLD operators
• give preference to applicants deferred in previous rounds LEADS TO negative effect on would be TLD operators
• require use of all and only accredited registrars LEADS TO positive effect on registrar business model
• require use of all and only accredited registrars LEADS TO negative effect on would be TLD operators
• require sunset period LEADS TO positive effect on TM owners
• require sunset period LEADS TO negative effect on would be TLD operators
• encourage sunset period LEADS TO positive effect on TM owners
• encourage sunset period LEADS TO negative effect on would be TLD operators
• discourage sunset period LEADS TO negative effect on TM owners
• prohibit sunset period LEADS TO negative effect on TM owners
• require all new TLDs to adopt current whois policies LEADS TO positive effect on TM owners
• provide for automatic renewal if not in breach LEADS TO positive effect on current registry business model
• provide for automatic renewal if not in breach LEADS TO positive effect on would be TLD operators
• periodically recompete all TLD slots LEADS TO negative effect on current registry business model
• periodically recompete all TLD slots LEADS TO negative effect on would be TLD operators
• require all new TLDs to follow current transfer policies LEADS TO negative effect on would be TLD operators
• begin process before all IDN issues resolved LEADS TO positive effect on would be TLD operators
• wait to begin process until all IDN issues are resolved LEADS TO negative effect on would be TLD operators
• allocate new IDN TLDs only to ccTLDs LEADS TO negative effect on would be TLD operators
• require applicant to persuade ICANN that it will add value LEADS TO negative effect on would be TLD operators
• require ICANN to have a good reason to turn an applicant down LEADS TO positive effect on would be TLD operators

• increased competition HAS IMPACT 5 on internet users -- competition
• decreased competition HAS IMPACT -5 on internet users -- competition
• increased stability HAS IMPACT 5 on internet users -- competition
• decreased stability HAS IMPACT -5 on internet users -- competition
• positive effect on current registry business model HAS IMPACT 1 on registries
• negative effect on current registry business model HAS IMPACT -1 on registries
• positive effect on registrar business model HAS IMPACT 1 on registrars
• negative effect on registrar business model HAS IMPACT -1 on registrars
• positive effect on ISPs HAS IMPACT 1 on ISPs
• negative effect on ISPs HAS IMPACT -1 on ISPs
• positive effect on TM owners HAS IMPACT 1 on trademark owners
• negative effect on TM owners HAS IMPACT -1 on trademark owners
• positive effect on would be TLD operators HAS IMPACT 1 on would be TLD operators
• negative effect on would be TLD operators HAS IMPACT -1 would be TLD operators

• open new TLDs IS INCOMPATIBLE WITH don't open any new TLDs
• don't open any new TLDs IS INCOMPATIBLE WITH open new TLDs
• open only sTLDs IS INCOMPATIBLE WITH open chartered and unchartered TLDs
• open only sTLDs IS INCOMPATIBLE WITH open only gTLDs
• open chartered and unchartered TLDs IS INCOMPATIBLE WITH open only sTLDs
• open territorial TLDs and gTLDs IS INCOMPATIBLE WITH open only gTLDs
• open only gTLDs IS INCOMPATIBLE WITH open only sTLDs
• open only gTLDs IS INCOMPATIBLE WITH open territorial TLDs and gTLDs
• gate at 10 per year IS INCOMPATIBLE WITH gate at 100 per year
• gate at 10 per year IS INCOMPATIBLE WITH gate at 1000 per year
• gate at 100 per year IS INCOMPATIBLE WITH gate at 10 per year
• gate at 100 per year IS INCOMPATIBLE WITH gate at 1000 per year
• gate at 1000 per year IS INCOMPATIBLE WITH gate at 10 per year
• gate at 1000 per year IS INCOMPATIBLE WITH gate at 100 per year
• allocate strings by chance if conflicting claims IS INCOMPATIBLE WITH allocate strings by auction if conflicting claims
• allocate strings by auction if conflicting claims IS INCOMPATIBLE WITH allocate strings by chance if conflicting claims
• require clear differentiation of strings in advance IS INCOMPATIBLE WITH allow current stringholders only existing remedies for any confusion
• allow current stringholders only existing remedies for any confusion IS INCOMPATIBLE WITH require clear differentiation of strings in advance
• charge a cost based fee IS INCOMPATIBLE WITH auction slots
• auction slots IS INCOMPATIBLE WITH charge a cost based fee
• conduct comparative hearing on which proposals best serve dns IS INCOMPATIBLE WITH use first come first served to allocate slots
• conduct comparative hearing on which proposals best serve dns IS INCOMPATIBLE WITH use lottery to allocate slots
• use first come first served to allocate slots IS INCOMPATIBLE WITH conduct comparative hearing on which proposals best serve dns
• use first come first served to allocate slots IS INCOMPATIBLE WITH use lottery to allocate slots
• use lottery to allocate slots IS INCOMPATIBLE WITH conduct comparative hearing on which proposals best serve dns
• use lottery to allocate slots IS INCOMPATIBLE WITH use first come first served to allocate slots
• proceed promptly whenever proposal is received IS INCOMPATIBLE WITH proceed in staged rounds
• proceed in staged rounds IS INCOMPATIBLE WITH proceed promptly whenever proposal is received
• require sunset period IS INCOMPATIBLE WITH encourage sunset period
• require sunset period IS INCOMPATIBLE WITH discourage sunset period
• require sunset period IS INCOMPATIBLE WITH prohibit sunset period
• encourage sunset period IS INCOMPATIBLE WITH require sunset period
• encourage sunset period IS INCOMPATIBLE WITH discourage sunset period
• encourage sunset period IS INCOMPATIBLE WITH prohibit sunset period
• discourage sunset period IS INCOMPATIBLE WITH require sunset period
• discourage sunset period IS INCOMPATIBLE WITH encourage sunset period
• discourage sunset period IS INCOMPATIBLE WITH prohibit sunset period
• prohibit sunset period IS INCOMPATIBLE WITH require sunset period
• prohibit sunset period IS INCOMPATIBLE WITH encourage sunset period
• prohibit sunset period IS INCOMPATIBLE WITH discourage sunset period
• provide for automatic renewal if not in breach IS INCOMPATIBLE WITH periodically recompete all TLD slots
• periodically recompete all TLD slots IS INCOMPATIBLE WITH provide for automatic renewal if not in breach
• begin process before all IDN issues resolved IS INCOMPATIBLE WITH wait to begin process until all IDN issues are resolved
• wait to begin process until all IDN issues are resolved IS INCOMPATIBLE WITH begin process before all IDN issues resolved
• require applicant to persuade ICANN that it will add value IS INCOMPATIBLE WITH require ICANN to have a good reason to turn an applicant down
• require ICANN to have a good reason to turn an applicant down IS INCOMPATIBLE WITH require applicant to persuade ICANN that it will add value