The introduction of internationalized domain names (IDNs) offers the world one of the best opportunities it will ever have to introduce more diversity and competition into the domain name registry market. That market is currently dominated by VeriSign, which operates the .com and .net domains (as well as a couple of TLDs it bought, .tv and cc) and hence controls about 80% of the gTLD market. It has proven almost impossible to crack the dominance of .com, but if anything can do it, TLDs in new scripts (Chinese, Cyrillic, etc.) responding to entirely new markets can.

Unfortunately, the politics within ICANN are threatening that opportunity. There is a move afoot to give country code TLD registries a free gift of an IDN top level domain. Did I say “a” free gift? That implies only one. But as I will explain later, it may turn out to be two, or three, or maybe six or even twenty free gifts by the time ICANN's Governmental Advisory Committee and CCNSO gets done with it. There's a real danger that this ccTLD fast-track to new IDN top level domains will reinforce incumbent national monopolies. And it's clear that this option is being considered only because ICANN wants to win political support from governments and the increasingly rich and powerful national registries, many of which are state-owned or state-connected. ccTLDs are starting to look like the old national telephone monopolies of old, and the ICANN process seems more and more like the Internet's version of the International Telecommunication Union, a cozy telco club that until 1997 wrote international rules to protect and advance their interests. Look here for the charter of the specially-created, restricted working group on IDN ccTLDs.

What makes this prospect of a free gift for ccTLDs especially objectionable is that ICANN's Generic Names Supporting Organization has just completed a proposal to allow organizations to apply for new TLDs, including multilingual ones. In that process, applicants will have to pay through the nose just to file an application, and then go through an arcane and treacherous challenge process to win anything. Whatever one thinks of that process, at least it was developed in an open, transparent way and everyone could comment on it. And while it was being developed, we all assumed that the same rules would apply to everyone. But the “ccTLD free gift” proposal short-circuits that process, in order to reinforce the already dominant position of incumbent national registries.

A bit of historical background is unavoidable here. Country code TLDs are a result of one of the worst — and one of the best — decisions that Jon Postel ever made. Postel was the USC computer scientist who single-handedly did what ICANN now does for the first 17 years of the Internet protocol's existence. Postel wanted all top level domains to be generic categories like .com and .org, but some European (whose political mentality apparently had not caught up with the Internet) insisted on having a TLD for his country. This proved to be a terrible idea, as it symbolically linked DNS assignments to national governments. That linkage started to come back and haunt the Internet in 1999 and we are still working out the consequences.

Postel's bad decision was to go along with that request to create country codes. His smart decision was to recognize the dangers of putting IANA in a position to decide who was or was not a country. So he found a standardized list that absolved him and the IANA of that decision. The ISO-3166 list, which has many things that are not national entities (like .IO for Indian Ocean), was at least an objective and neutral way to decide which territorial TLDs should exist.

The point is that there is but one DNS name space, not two. The distinction between ccTLDs and gTLDs is entirely political and arbitrary, not technical or economic. Since there is no IDN equivalent of an ISO-3166 list, the distinction between ccTLDs and gTLDs in the IDN space is even less sharp, one might say downright blurry. Indeed, in the early days of ICANN, cc's were thrown into the same policy making process as commercial registries, trademark owners and the like. ccTLD's broke off into their own supporting organization, due to the fact that ICANN had less contractual power over them and they needed to be persuaded, rather than forced, to join the regime. But as a matter of indisputable fact, ccTLDs are just registry operators and have the same basic economic interests as Afilias and VeriSign. Except that in many cases, e.g. China or Russia, they are closely connected to national governments.

On the whole, ICANN should strive to apply the same principles and standards to all applicants for new TLDs in the name space, whether they are ccTLDs or gTLDs. And so it's not clear why ccTLD's have a “right” to an extra IDN TLD at all. Does anyone believe that .com's owner should be handed a free gift of the “equivalent” of .com in another script? Should the registry for .org get a free gift of the equivalent of the term “organization” in Cyrillic or Chinese? What is the rationale for giving ccTLDs “fast-track” access to self-selected IDN strings and not gTLDs? I do not believe there is any public interest rationale, any technical rationale, any economic policy rationale. There is only a concession to the perceived political power of certain ccTLDs and governments.

But despite ICANN and the US Government's continuous stream of rhetoric about how “private sector led” and “bottom up” ICANN is, governments and incumbent country code registries have more clout than ordinary Internet businesses in ICANN. And a lot more clout than end users. So they have asserted the idea that they have some kind of special claim to new top level domains simply because they already have one that allegedly “represents” a nation. For a well thought out challenge to the link between sovereignty and country code names, see this great paper by Michael Froomkin.

I was originally sanguine about this process, thinking, “sure, give each ccTLD one new IDN string in a script of their choice and be done with it.” The govts/incumbents get their pound of flesh and that frees the rest of the world to proceed without interference.

But the international politics of names are too subtle and complex to be so easily contained. First, the countries decided that they couldn't possibly be satisfied with some abstract, two-character representation of their country name in a format that would be the same for everyone. Some countries contended that they couldn't possibly properly represent their country name with only two characters. Sounds fine, except that the beauty of the old ASCII country code list is precisely that it is a standardized coding. It is an essentially arbitrary mapping that obtains its correspondence and meaning through use, not through some divinely ordained correspondence to a country's “real” name. In that arbitrariness there was a complete equality and efficiency. Everyone knows that 2-letter TLDs are country codes. All country codes are roughly equal, each country gets one and no one is any better or worse than another.

Once you depart from that rough equality, then it's possible that some of these IDN scripts will be 3 or 4 characters, some will be 2, some may be 6. You are no longer dealing with “country codes” but with words, which means that they are in effect gTLDs. And it is only a short step from that — a step that many GACers and ccTLD managers are already making — to claim that they should have more than one, because they need to have “their” country name in multiple scripts. The European Union, for example, said that since its territory includes countries with Cyrillic and 3 other non-Latin scripts, it is legally barred from choosing only one IDN to accompany .eu. (Funny then, how was it legal to get .eu in the first place?) India, it turns out, has 24 different language groups that might represent its name differently. And of course, if one country gets 24 new IDN TLDs, won't they all want that many?

Given the fact that national governments have already demonstrated their tendency to claim all kinds of names related to their country, there is also the possibility that ccTLDs will lay claim to multiple words that refer to the country (e.g., the equivalent of “America,” USA, “United States” etc.)

At the end of the day, you get two tracks to a new IDN TLD:

On track one, you have incumbent ccTLD monopolies, who already control anywhere from 98% to 60% of the DNS registry market in their country. Those registries will be given, for free, first pick of two, three, or maybe 20 new IDN TLDs, and be able to implement them before anyone else, in a market that is characterized by extremely high switching costs.

On track two, you have nongovernmental competitors, who will have to wait a year or two extra, pay six-figure sums just to file an application, then another $200-300k to prepare their application and lobby for it at ICANN meetings, and then be subjected to months of challenges, review by “expert panels,” and possibly competitive auctions.

The free gift to ccTLD ideas needs to be carefully weighed to consider its implications for competition and diversity. I can't think of a better demonstration of the new international politics of ICANN in the post-WSIS period.

8 thoughts on “Land Grab? ccTLDs and multilingual names

  1. IO is alpha-2 code from ISO 3166-1 list for British Indian Ocean Territory, which is quite different from the whole Indian Ocean you suggest.
    Elisabeth Porteneuve

  2. Milton,
    Good article, good insights.
    India has 22 languages – represented by 11 scripts.
    One of my concerns is that the fast track ccTLD approach does not promise a review mechanism; I fear that any outcomes generated will simply be sent straight to the Board for a swift vote.
    Although it is convenient and simple to continue to use just one script to represent a country, and many countries will pick one if forced to – there should be no reason why countries should not apply for all the constitutionally mandated scripts that represent the name of their country.
    I agree with you reg. equal treatment on policies.
    You and I have been at opposite ends of the “it's just a symbolic representation” debate. Fact is that there is no easy “1 character, 2 character” rule that fits for local language names of geographic territories. And – no one wants to wait for a new ISO internationalized 3166 list either…

  3. The way you remind us of the history of the Internet is quite interesiting:
    Postel wanted all top level domains to be generic categories like .com
    and .org, but some European (whose political mentality apparently had
    not caught up with the Internet) insisted on having a TLD for his
    country. This proved to be a terrible idea, as it symbolically linked
    DNS assignments to national governments. That linkage started to come
    back and haunt the Internet in 1999 and we are still working out the
    Could you please remind us by the way what happened to .gov, .mil and .edu? Are they today really as “generic” .com, .net and .org you mentioned? If some Indian, Japanese, Iranian, Spanish or Cuban citizen (for example) wanted to register their domain name under those TLDs, do you think they would be allowed to?

  4. Good point. No, of course, mil, gov and edu became de facto “American” (although there are some non-US universities in edu). Establishing other country codes may have been the path of least resistance back then — but it also would have been possible to demand that the generics be opened up globally. The problem was linking TLDs to national jurisdictions.

  5. Most ccTLD registries do not have “the same economic interest as Afilias and Verisign”, they steward a national resource in the interest of the national community (however defined locally). In many cases, money is just a nice side product.
    In general, you completely miss the point by reducing this issue to a mere economical debate. The need for prompt introduction of IDN ccTLD in certain parts of the world is, first of all, social, and consequently political for the due representatives of those societies (that is, their governments). Again, the local Internet communities do not give a damn about preserving global competition, if this means that they are deprived of the chance of having domain names in their script. Competition is nice, but it is not a sacred principle.
    However, you could have a point if you turned the question differently: not “why do we need a fast track” but “why are incumbent ccTLD operators necessarily the right managers for these new TLDs”. If I were ICANN, I would ask the local community to agree on which strings are to be introduced, and on who do they want to run the new TLD. If they disagree, as (per WSIS) ccTLDs are a country's own affair, I would ask the relevant government.

  6. To keep following your reasoning. If I understand correctly, do you mean that .com, .net et .org are today less tied to some specific national juridictions than ccTLDs ?
    For instance, if some German .com registrar had a legal problem with .com registry, wich country's jurisdiction would be folowed/applied, Germany or US?
    You may want to look at this old article (I'm not necessarily endorsing it but it may enlighten this specific aspect):

  7. Milton,
    I read with interest your blog “ccTLDs and multinational names”. I too am alarmed at the fast track pace ICANN seems poised to give to ccTLDs for their IDN equivalent(s). I cannot argue with your logic or arguments, save in one paragraph:
    “Should the registry for .org get a free gift of the equivalent of the term “organization” in Cyrillic or Chinese?”
    I would argue “Yes”!! There are very good reasons for letting all gTLD registries have equivalent names in other scripts. We like other gTLDs will rightfully argue that handing out IDNs to new and un-established registries will neither help security and stability, nor will it ameliorate consumer confusion which will happen when multiple registries run the “same” TLD depending on where you are. BUT please remember we (.org) are hardly a monopoly. We have a little over 6 million domain names globally, hardly the stuff of monopolies when you consider the domain market is over140 million names. Furthermore, we are a Not-For-Profit with correspondingly different aims than a publicly traded corporation such as VeriSign. And lastly, as you yourself mention, it will cost us a lot of money to apply for, defend and then (assuming it all works out) establish and service the IDN TLDs properly. The ratio of cost to revenue to an organization such as .org is vastly greater than say to a corporation such as VeriSign. For these reasons, if for nothing else, .org deserves to be considered as a stand alone.
    Alexa Raad
    CEO .org Registry

  8. Vittorio: You say “Most ccTLD registries do not have “the same economic interest as Afilias and Verisign”, they steward a national resource…”
    I disagree, I think you are naively spouting nice-sounding rhetoric used to rationalize monopoly privileges. This points to a major difference between people who have experience in dealing with the policy issues around the breakup of historical telecom monopolies and those who don't. National telecom monopolies ALL used “national sovereignty,” “national security,” “national this and that” as excuses for their exclusive control. But it was really all about control and money, not about “stewarding a national resource” (whatever that means). And as the virtues and benefits of competition became evident those arguments collapsed.
    I have personally talked to two major non-European ccTLDs who will come right out privately and say they don't want IDN competitors horning in on what they consider to be their markets. They want to get into the market first, because they know that whoever does will (just like VeriSign) control most of the market. If you feed them this “stewardship” line I'm sure they will use it to the hilt, but it won't change what the game is about one bit.
    I am not sure what point you are making when you say, “local Internet communities do not give a damn about preserving global competition, if this means that they are deprived of the chance of having domain names in their script.” First, many local Internet communities do give many damns about competition. Second, this is a false dichotomy. Global competition will provide most if not all internet communities with domain names in their script. The issue is whether the operator is the incumbent monopoly or whether someone else has a chance. And when you are dealing with scripts — any script you can name — the potential market NEVER coincides with a “local community” but is always transnational and trans-local. So it's not clear what “community” should decide.
    But to find common ground, it's true that incumbent ccTLDs don't have to be delegated the new IDN ccTLDs; if ICANN is going to abandon its painfully developed new TLD application process and create a “fast track” it should at least require a competitive bidding process in each country that gets one.

Comments are closed.