Blog

Mathieu Weill, CEO of AFNIC, which runs the French ccTLD .FR, said in an interview with David Goldstein that a new law will allow — perhaps compel — AFNIC to offer registrations for at least the following ccTLDs:

• .GF (French Guyana)
• .MQ (Martinique)
• .GP (Guadaloupe)
• .RE (Reunion) was also mentioned, but it is already registering names.

These TLDs have been silent for years, after a brief floresence in the late 1990s was extinguished when the French (I’m still unclear whether it was AFNIC or the French government) repossessed them and shut them down. I have been critical of AFNIC in the past, so it’s very encouraging to see these TLDs activated.

The French control several other inactive ccTLDs:

• .PF (French Polynesia, has a tiny amount of registrations, not open to the public)
• .PM (St. Pierre and Miquelon, used to accept registrations, was taken over by AFNIC under still-unexplained circumstances, closed to registrations)
• .TF (French Southern Territories, used to accept registrations, redelegated by ICANN, now held directly by AFNIC, closed to registrations)
• .WF (Wallis and Futuna, never activated, held directly by AFNIC)
• .YT (Mayotte, never activated, held directly by AFNIC)

M. Weill hinted that these too would be open for registration soon (emphasis is mine):

David Goldstein: Recently AFNIC announced there will be a new legal framework for all French ccTLDs, not just .FR but also .RE, .GP, .MQ, and .GF. What impact will this have and why did it come about?

Mathieu Weill: The new legal framework will be in force on July 1st. The main consequences will be that AFNIC has to review all its current disputes resolution policies. The new law also impacts the other ccTLDs for French territories. We are pleased that this law finally grants us authorisation to launch registrations under these ccTLDs before the end of 2011. We are looking forward to it.

He also said that .FR and other ccTLDs will benefit from the introduction of new gTLDs:

David Goldstein: The new gTLDs, whenever they come about, threaten to change the domain name landscape significantly – how do you think this will affect .FR?

Mathieu Weill: Well, there are three ways to consider this evolution. First, all the domain name market could benefit from it. For instance, we saw .FR registration increase significantly thanks to .EU opening. That’s because people are getting more and more aware of the importance of domain names.

Secondly, there is a risk that hundreds of new TLDs could also create some confusion in users minds and incite them to rally to the TLDs they know best: the main gTLDs and their national ccTLDs. .FR would also benefit from such a phenomenon.

Thirdly, we do not buy the argument telling that new TLDs will replace the current ones. They are new communication channels or opportunities, but users would better think about naming strategies combining several TLDs than relying on just one TLD.

Overall, we hope that new TLD applicants will bring innovative ideas to our field, and that the outcome of the process will be additional added value services to the end user.

Congratulations to AFNIC for this step forward, and thanks to David Goldstein for a good interview. I welcome any corrections or clarifications.

In his eloquent dissent against approving .XXX, ICANN Board member George Sadowsky talked about blocking and filtering top-level domains. It’s a concise statement of a concern that has been identified by various people, including members of the GAC, as an impediment to the new gTLD program. It’s a thorough defense of a common point of view about blocking TLDs, but while no-one can disagree about the fact of blocking, what is the actual effect?

George Sadowsky’s comment is worth quoting at some length:

Fourth, and extremely important, I believe that the future of the unified DNS could be at stake [if .xxx were approved].

I submit that the approval of the application for dot xxx could encourage moves to break the cohesiveness and uniqueness of the DNS.

In my judgment, it would undoubtedly lead to filtering the domain, and quite possibly instigate the erosion, degradation, and eventual fragmentation of the unique DNS root.

Now, while we know that filtering already exists, I believe that the creation of dot xxx would mark the first instance of an action by this board that may directly encourage such filtering, posing a risk to the security and stability of the DNS.

In my judgment, the board should not be taking actions that encourage filtering or blocking of a domain at the top level.

Further, I believe that the filtering of so-called offensive material can provide a convenient excuse for political regimes interested in an intent on limiting civic rights and freedom of speech.

Further, I believe that such moves provide an incitement to fracture the root, a concern that we’ve recognized in preparation for
the new gTLD program as a distinct threat to the security and stability of the DNS.

There can be no doubt that .xxx will be blocked by some countries: the government of India has already announced its intention to do so. The .xxx domain exists in order to be filtered — that’s almost the entire point of it. It is premised on segregating content into adult and non-adult categories, so that people can find it easily — or avoid it. So no-one could disagree with George’s assessment of the likelihood of .xxx leading to filtering.

Widespread Blocking of the Internet Exists Today

As George Sadowsky points out, filtering and blocking already exist. Not just at the second level (individual web sites) or the top level (TLDs), but also of the entire Internet. Consider this graphic from a recent presentation by Packet Clearing House showing Internet traffic in Egypt:

This is filtering on a massive scale, done by a regime that didn’t want its citizenry to have any information that conflicted with its message.

Meanwhile, Hillary Clinton presented the U.S. State Department’s annual report on human rights, running to 7,000 pages. The Associated Press writes:

More than 40 governments are now blocking their citizens’ access to the Internet, and the firewalls, regulatory restrictions and technologies are all “designed to repress speech and infringe on the personal privacy of those who use these rapidly evolving technologies.”

Second-level domain names are blocked almost as a matter of course in large parts of the world. The most-blocked site is Facebook, followed by YouTube, Twitter, and a host of other sites that are hugely popular — some of them porn, but many of them not.

Basically, the governments of the world engage in blocking and filtering on a massive scale. They are blocking second-level as well as top-level names, and sometimes they just block the entire Internet. They block based on content: porn, political statements across the ideological spectrum, religious speech of all kinds, and they also block just on the basis that they don’t want people sharing information. This kind of governmental action is not new. Monopoly of information has long been a goal of many governments: until recently, one of the major goals of a coup or a revolution was to capture the TV and radio stations.

Will TLD Blocking Fracture the Internet?

Blocking of Internet content is pervasive, and the creation of new TLDs which are offensive to someone, somewhere, will probably increase it. But will it fracture the Internet? That’s where I think George’s fears may be out of place. The current blocking is so widespread, so thorough, and so invisible to those who don’t have to deal with it that it’s just part of life in much of the world. Why hasn’t blocking already encouraged a fracture?

For one thing, an alternate root by itself is not a fracture. There are already many TLDs on alternate roots out there, from Karl Auerbach’s .ewe to the semi-autonomous Chinese-language TLDs. The threat to the single root doesn’t come from just the fact of setting it up, it comes in the form of a viable alternative that threatens the current Internet by gaining users and adoption at the expense of the current favorite (think MySpace and Facebook). Karl’s .ewe is not getting a lot of takers, and in China you don’t really have a lot of choice — no-one is “choosing” any of the alternatives. (The only alternative use, in this sense, has come from new TLDs/roots in non-Latin scripts, and ICANN’s push to delegate new IDN ccTLDs has done a lot to alleviate that pressure.) So blocking .XXX (or any other new TLD), as long as it doesn’t threaten to create a competing root, is just more of the same old blinkering of its citizens that governments are addicted to and will never stop unless their people insist on it.

Let’s suppose, however, that it was possible for a mandatory alternate root to be set up, enforced by governmental authority. In a state with just a few major ISPs, the government might compel them to point to the new, alternative, government-mandated root. Isn’t that a problem? (Note that this is not currently the case in China, which allows access to the Internet, just not to many of its sites.)

To examine that possibility, let’s turn to television, where this situation is common. In Iran, for instance, there is a limited roster of TV stations and they are all closely censored. What happens there?

One of the biggest hits on Iranian TV is not on Iranian TV. A kind of Persian “Daily Show” called Parazit is broadcast by the Voice of America. Parazit is watched by millions of Iranians through their illegal satellite dishes, which are extremely common in Iran, despite periodic attempts by the morals police to get rid of them (satellite dishes can also be used to access the Internet). Parazit is a hit — it gets 45,000 You Tube visits a week, and 17 million Facebook visits per month.

The net effect of Iran’s censorship is to make its leaders laughable and hated, but it has not threatened the Iranian TV “root,” which goes on broadcasting propaganda. It has not led to a call for an alternate state television either — people simply bypass the restrictions and access the rest of the world.

In the world today, Internet blocking and filtering of all kinds is widespread and deep, and it has not threatened the single root. Censorship is a favorite habit of some governments, and they are not weaned from it easily. Limiting “controversial” TLDs in order to appease that impulse, in the name of preserving a single root, is illusory (alternate roots already exist), not likely to matter (people will get out to the “real” Internet somehow), and it doesn’t really make sense.

As Milton Mueller put it: “The idea that it is somehow better for the Internet to use centralized, global administrative mechanisms to block domains from existing in order to prevent a few individual countries from using technical means to block them locally is absurd and dangerous.” Or, to quote another ICANN Board member, Suzanne Woolf: “The issue of governments or any other entity blocking or filtering access to a specific TLD is not unique to the issue of the dot xxx sTLD. What we agree is blocking of TLDs is generally undesirable. If some blocking of the XXX sTLD does occur, there is no evidence the result will be different than the blocking that already occurs.”

George Sadowsky is a principled person who clearly loves the Internet and wants to preserve it. He gives a clear voice to a fear that many have. But when we look at what blocking actually is, and what it does, I think the fears are unfounded. People will find a way to see what they want to see, and ignore stuff that they don’t like. Blocking of a TLD by a local government is not going to lead to the fracturing of the Internet. If the case of Egypt is any guide, it’s more likely to lead to the fracturing of the local government.

Some bad ideas at ICANN crop up again and again, stubborn and persistent like a case of herpes. These ideas seem innocent at first, but after they take root they turn out to be difficult to control and are guaranteed to scare people away. Even after treatment, they re-appear with unnerving regularity.

Such is the idea of categorizing and segregating and prioritizing new top-level domain names. In 2001, we had the Board picking names (category: “We like them!”); in 2005, sponsored TLDs (category: “They’re safer!”); now in the current round we have communities (category: “They’re worthy!”) and geographical names (category: “Government property!!”). If some people have their way, we’ll soon be adding any number of other gradations of purity to the selection criteria (category: “Priority for strings I approve of!”). These last are the supposedly “uncontroversial” strings — except that nothing about categories is uncontroversial when you get down to the details, as we shall see.

Over the last ten years, the ICANN mania for categorization has prompted an enormous collective yawn from Internet users, as they issue one boring TLD after another. (The one interesting TLD, .xxx, was introduced only after ICM Registry, the proposer, spent millions of dollars to force ICANN to approve it.) Even the biggest winner among these “picked for success” TLDs, .info, was at one point was reduced to giving away domain names to try to stimulate usage. People have said that some of these TLDs are successful according to their own lights, and perhaps they are, but let’s be serious — every other Internet initiative is judged successful or not according to adoption and usage, and in this sense TLDs in the “chosen” category are failures. The extensive history of the use of categories in domain names — at both the top level and the second level — has shown that they have one overwhelming effect, which is to limit interest and use.

History of a Bad Idea

The idea of TLD categories has been around since before ICANN. One of its more extreme manifestations comes in a 1996 proposal from Simon Higgs, in which he says:

This document covers… the framework necessary to define the function, delegation, and use of new top level domains. Several factors need to be addressed such as why the TLD exists in the first place, who accepts registrations for the TLD, and what special purpose (if any) the TLD serves. These questions can be answered by the recognition of TLD “classes”.

Sounds sensible and innocuous, right? Yes, until you get to the conclusion — that all TLDs should be shoe-horned into 45 categories, including such doozies as:

• .ARTIF – for artificial limbs, eyes, and teeth
• .DERM – for leather goods (and Valtrex?)
• .CUL – for culture (works particularly well in French)
• .ITAR – for guns and ammo, I kid you not
• .WEAR – for clothing

… etc.

Chanel, under this scheme, would be required to register its name as chanel.wear, and Coach to sell its leather luggage under the super-sexy name coach.derm. This is a vision of the Internet, the most powerful commerce engine ever devised, crippled and turned into a branding nightmare. This is a vision of the Internet, the most powerful means of finding information ever invented, categorized into a reductionist caricature of trademark law. (In fairness to Mr. Higgs, the list of names was presented as a draft, with room for improvement. But still….)

We can wonder at this logical but hare-brained scheme from 1996, but domain classification schemes are still in force today, with depressing results. From the Higgs proposal to the seven TLDs proposed by the IAHC, from the “beauty contest” of the initial ICANN round to the “proof of concept” sponsored TLD round, not one categorized TLD has achieved anything resembling widespread approval or adoption. Contrast and compare to the market-driven launches of .RF (Cyrillic) and .CO, which have been instant successes.

What the Research Says

So what does drive adoption of a new TLD? Dr. Matthew Zook of the University of Kentucky was commissioned by APTLD to do a study of the factors that lead to acceptance and adoption of top-level domains, in this case ccTLDs. He identified three essential factors: population size (in other words, the addressable market), income level, and finally registration policies, which turn out to be a key factor. TLDs that restricted delegations based on categories of applicants, for instance allowing only businesses to register, or requiring applicants to submit proof of identity, were found to be very poorly received — not least because each restriction meant instituting a time-consuming and frustrating bureaucracy to verify an applicant’s eligibility. Going beyond Zook’s study, we can also observe some restricted TLDs, for instance .FR and .SE, abandoning their rules and seeing adoption rates go up substantially.

Administrators of some ccTLDs early on put into place categories and distinctions whose effects still ripple through the domain name industry today, resulting in inefficiencies and costs for registrants. I know this well: I built and sold two successful businesses, NetNames USA and NameEngine, that were devoted to helping corporations figure out how to register names in different ccTLDs, and how to “comply” with various restrictions on eligibility. For instance, if the rules required that a registrant needed to have an in-country contact, we would supply one. The net effect of the TLD categories and restrictions was to increase costs to applicants, thereby ensuring that those with resources got what they wanted, while the poor were hampered by the rules. Today, Mark Monitor and others continue to run successful businesses that help corporations decipher and exploit the arcana of domain name eligibility policies.

Categories = Restricted Access

“But,” (some will say) “this is not the fault of categories, but of the restrictions themselves.” That’s a false distinction. The fact is that there never was a domain name category that was not used to restrict access or eligibility in some way. Restrictions are the other face of categories, they are inseparable. Eligibility restrictions are the whole point of categories. And restrictions lead to cost, delays and depressed levels of adoption.

No Such Thing As a Simple Category

Categories are chimerical. They appear simple, commonsensical, real, but has there ever been a category as simple as it seems? One category that ICANN has allowed, in a nod to governments, is that of geographical names. This grouping, which seems uncontroversial — what could be easier than place names? — has proven to be fraught with intractable nuance. What, for instance, is a city? The GAC member from Greece said at the ICANN meeting in San Francisco that a “city” was any place on earth where more than one person had settled. Absurd as that may seem, in a testy twenty-minute exchange on the subject between the ICANN Board and the GAC, that’s as close as anyone got to a definition.

As well as having calamitous practical effects, categories are inherently unfair because they are impossible to define. This always leads to a subjective judgment of who is in, and who is out.

The Guardians of Rectitude

Dividing up applicants or strings into groups sounds easy and simple, but when you get down to the details categories are difficult to define, devilish to administer, and unfair and expensive to applicants. If you’ve worked at a registry or registrar, you know this.

If you haven’t, you may nonetheless notice the unpleasant whiff of grinch emanating from certain quarters of ICANN. The very idea that people could have any domain name they want, just because they want it, is intolerable to the misanthropes who would like to exercise permanent micro-regulation over new gTLD registrations. Categories are their weapon, because with categories come boundaries, and with boundaries come infractions, and with infractions come punishments, and for punishments you must have judges — and here they come, the merry judge-volunteers, ready to rain red tape on unwitting registrants and registries. “How dare you want allow a flip-flop company to register a .shoe domain? Everyone knows that flip-flops are not real shoes! The registration must be revoked!”

Abstinence Is a Virtue

Categories in the domain name world have proven to be pernicious. The registration requirements they engender depress demand, introduce costs and bureaucracies and delays, and because those with money and persistence will always find a way around the rules, they exacerbate inequalities of access. Add to this the near impossibility of coming up with a category that can be defined with any precision. The same dynamic that has turned some ccTLDs into ghost towns will take hold in the new gTLD program if new categories are introduced into the process.

The ICANN Board and staff are reluctant to expand TLD categories. That is very wise; seductive as they are, domain name categories are a disease with symptoms that are painful, embarrassing, and difficult to manage. And also gross and itchy.