Home / Blogs

ICANN: Do Not Allow Closed New gTLDs With Generic Strings

The Board was right in 2014 when it disallowed them.

Within the next year, the ICANN Board may well face a decision that will help determine whether ICANN is capable of serving the global public interest or whether it is degenerating into an industry-controlled self-regulatory association. The issue can be framed quite simply: will ICANN approve a process for the creation of a new wave of new generic top level domains that will include “closed generic” gTLDs?

The issue can be semantically confusing since the word ‘generic’ is used in two different senses. There are at present about 1200 gTLDs, or generic top level domains that are recognized by the DNS. Each of these gTLDs are operated by a registry under contract with ICANN. Most of these gTLDs are ‘open,’ which means that anyone who subscribes to the requirements of the registry and pays the subscription fee can establish a second level domain under the gTLD.

By “Closed Generics” I mean the situation in which the gTLD registry controls registration of all second level domains of the “generic name” of its business or industry. As an example, the new gTLD .book is an open generic gTLD, and anyone should be able to apply for a second level domain name registration under it. Alternatively, the registry may have a stated policy to restrict registration, but must allow domain names to be registered by all “similarly-situated” businesses or industries, e.g., all cloud companies in .cloud, all search industries in .search, and all bloggers in .blog.

A small number of new gTLDs are closed, and the owners of those registries determine how the second and further levels of the DNS within their gTLDs are to be allowed, structured and used. Some brand TLDs are closed, and can be used for the specific purposes of the brand holder; .canon and .ibm are examples of such domains. They have been approved by ICANN and are legitimate TLDs.

There was a real question whether “closed generics” were barred in the 2012 round of the new gTLD process. Some of those who wrote the original gTLD Applicant Guidebook said no, and some said yes. After the evaluation process had begun, the ICANN Board was confronted with a number of applications in which the registries applied for strings that represented generic categories and that were to be closed—specifically closed in that the registry said that ‘it would be owner of all the second level domains of the “closed generic” gTLD.’ Thus, Amazon applied for .book (among others) as “closed generics,” Google for .search (among others) and Dish DBS for .mobile. This would have given these large, senior competitors in these industries and businesses total control of an information space for major project groups and industries—control to shape, or distort, an Internet user’s view of the market for goods and services in these areas.

Asked to review the issue, and after a public comment period that included many new voices from across the world, including the Global South, the ICANN Board decided that closed generic strings (such as .book) would not be allowed in the 2012 round of new gTLDs. The essence of the concern on the part of the Board was that “closed generics” would allow the monopolization of a significant part of the information space and would create a non-level playing field for others within that space. In effect, a registry awarded a closed generic string gTLD would have complete control over which second level domains. Nothing would prevent it from biasing its selection in any manner it wished, whether in the public interest or in any of its own competitive interests.

I was part of the group on the ICANN Board that revised the objections to the pending Closed Generics applications and decided to require these gTLD applications to “become open” before moving forward. My concern was simple. Consider the following analogy: In the mid-1900s, not too long ago, public libraries were a major source of information, and were the equivalent of the Internet of today. Let’s suppose that your local government decided to subcontract its public library to a commercial organization. Suppose further that that the terms of the contract included allowed the contracting organization to have complete control over what books, periodicals, and newspapers they would acquire and make available to patrons. The subcontractor could say, “we don’t believe in racial tolerance, we don’t believe in social class mobility, and we don’t believe in sexual health of women. So let’s make sure that the material that we make available does not focus on any of those things, and let’s only buy nice books that reinforce our readers’ comfort and understanding of the benefits of the status quo. We don’t need to state our beliefs up front; we can embody them in the material that we make available as a result of our acquisition policies, so let’s just let our collection of materials speak for itself.”

What I have just described is analogous behavior to what a registry running a closed generic string could do. In our wildest imagination, would we ever accept a library that acted in a manner described above? In fact, would we ever even consider even accepting such a contractual arrangement? If not, why do we even admit the possibility of consenting to delegate closed generic string new gTLDs?

Closed generics give total control of a very important and intensively used Internet information space about a major product group or industry to that registry, a private organization, often one of the largest competitors in the field. Control of that information space by the registry/competitor includes the power to shape and to distort in any way the registry/competitor wants—an Internet user’s view of that market or industry. Further, the information contained in all of the web pages associated with the domain names of the “closed generic gTLD” are completely under the control of the registry and could route all business to themselves or even contain false information, without the possibility of on-site on-line refutation.

It could be argued that ICANN could write contracts with the owners of generic string new gTLD that would mitigate any such harms, possibly using public interest commitments as a safeguard against various kinds of misuse. To do so would generate a difficult exercise in defining the misuse and the result would surely be gamed, probably successfully. To the extent that adherence to a “no-misuse of a closed generic gTLD” policy would be voluntary, it would simply camouflage the uselessness of such PICs. Further, markets work best when the financial interests of the participants work in the same direction as the public interest rather than against it as is the case here.

Finally, given current ICANN policies of purchase and sale of registries as financial assets, any promises of “openness” and competition with a “closed generic” gTLD could amount to nothing should the original gTLD registry transfer or sell the gTLD to another leading industry competitor—or another owner who might repurpose this term of an entire industry or field with apparent impunity.

At this time, the GNSO’s New gTLD Subsequent Procedures Policy Development Process Working Group is finalizing its recommendations for the specifications of the next (and perhaps ongoing) edition of the Applicant Guidebook—rules that will govern the process of applying for and receiving approval for new gTLDs. The application period is estimated to start in the next several years. Among the undecided issues is whether new gTLDs with generic strings will be allowed as “open” or “closed”. The working group appears to be split, with adherents on both sides of the issue. The issue is likely to be resolved one way or the other in the near future.

To the Subsequent Procedures WG, I would like to share that during the evaluation period of the last round of new gTLDs, the committee of the ICANN Board tasked with overseeing the implementation of that round ruled that delegation of closed generics would not be permitted for that round. I was a member of that committee and I voted in favor of the decision to bar closed generic gTLDs. It was the right decision.

Some people now believe that the way in which this Board decision was reported suggested that the Board expected that the issue of Closed Generics would be revisited and potentially revised for later rounds. That was not my understanding, although my memory of the details of the discussion and decision is not precise after six years. Our Board work at the time was comprehensive and our decision was based on global input.

Just the knowledge that closed generics are again being advocated by some members of the community tells me that there is a lack of understanding of the role of ICANN with respect to the global public interest. If the Subsequent Procedures Working Group—incidentally, one of the most obscure and misleading names a working group could ever have—decides to recommend that new gTLDs in future application rounds may be “closed generics” then this decision will be both against previous board policy, and will significantly reinforce my sense that those members of the working group regard ICANN as an industry lobbying group for registries, and are helping to push ICANN in that direction.

Should the Subsequent Procedures WG make such a decision—I would hope that the ICANN Board would intervene and return the New gTLD Program to its 2014 stance—no Closed Generics allowed.

Should Closed Generics still be implemented in the next and future rounds of new gTLDs—unstopped by the WG, the GNSO or the ICANN Board—I believe this would signal a huge loss of trust and show that ICANN and the ICANN community are indifferent to the global public interest.

Such a result would, to me, signal that the ICANN multistakeholder experiment in Internet governance is inconsistent with a commitment to uphold the global public interest, and that it has failed in action. This would be disappointing to many, but would also be evident. It would then be time to start thinking about significant changes to or replacement of some of the Internet’s governing organizations. It’s essential that we have governance mechanisms that can understand and enforce the concept of the public good, open and competitive markets and can differentiate and ensure its independence from rapacious private enterprise. The “closed generics” issue is one test of whether that time has come.

By George Sadowsky, Information Communication Technology Consultant

The author was a member of the ICANN Board of Directors during 2009-2018 and is a member of the Internet Hall of Fame.

Visit Page

Filed Under


Thank you George for this article. As Jeff Neuman  –  Feb 16, 2020 11:26 PM

Thank you George for this article. As one of the Co-chairs of SubPro I appreciate that this is important and I will forward this to the group. However, I would note that there are a few issues with some of your assumptions and conclusions in this article.

First, the board resolution did in fact not set policy (as that is not its role in our multistakeholder model). It rather decided to implement (sort of) the GAC Advice. And The GAC advice did specifically allow closed genetics if a public interest could be shown.  The Board resolution, however, decided not to send this policy issue to the GNSO, but rather decided on its own to ban them for that round and did send the issue to the GNSO for subsequent rounds.

Second, if the Subsequent Procedures PDP Working Group recommends moving forward with allowing some form of closed genetics, it will not be because the group has been captured by any particular interest at all. And to state an assumption like that is to show a lack of confidence in the PDP process itself and the PDP Leadership Team. For that matter, the Multi-Stakeholder Model as well. I am not going to presume what the outcome will be as we are still discussing the issue, but all recommendations require a consensus of the Working Group.  Consensus is not measured by quantity of votes in favor of a recommendation, but rather is the result of a qualitative objective judgement based on all of the members of the group as well as all of the comments received and positions of all of the stakeholders. If done correctly, capture is not possible.  And Cheryl and I (the 2 co-chairs) fully intend to measure consensus as it is intended to be measured. So, by definition again, IF the PDP working group comes to a consensus on a recommendation to allow some form of closed genetics, it cannot be because it was captured.  And to pre-state that the Board should strike down any recommendation it gets which is contrary to your view ignores the multistakeholder model and presumes any rationale for a position contrary to your view cannot possibly be provided in good faith.

Third, there are theoretical examples of closed generic applications that could be more in the public interest operated by one entity as opposed to just allowing everyone to register the domains at the second level in an unrestricted manner. One example brought up in the working group Is a .disaster operated by the International Red Cross. If for example they used second level domains only for its organization in connection with disaster relief efforts around the world to let the public know that if they get a solicitation from a .disaster name, it is legitimate (eg., hurricainesandy.disaster), that could very well serve a legitimate public interest.  One could argue that using the top level domain in this manner is more in the public interest then allowing Open registration by anyone of domain names in this space (some of them may have bad intent, if any of the names actually ever get used).

Fourth, the definition of generic is something that has been litigated in the courts around the world for many years, especially with respect to trademarks. In fact, there is a case at the US Supreme Court this term looking at Booking.com and interpreting issues involving domain names and the definition of generic. So, one of the key issues we are looking at is to make sure that we are working with a good definition for what is generic regardless of how this issue turns out.

Finally, from whose point of view do we measure the public interest. Is it the interest of domain name registrants to be free to register domain names, or do we look at the public interest from the position of end-users on the Internet looking at how the domain names are used? You may get two very different answers depending on how you examine that issue.

Thank you George for writing this article. I am honestly not sure how this debate will work out, as we are still in detailed discussions on this topic. However, I am merely just pointing out that there are a lot of complexities around this topic which are not fully addressed in your article. Please feel free to join our conversations if you would like. I know we all would value your input greatly.

Jeff:The applicant has to prove "public interest". Alexander Schubert  –  Feb 17, 2020 5:32 PM

Jeff: The applicant has to prove "public interest". If the ONLY way to convey that is a commitment to allow the public to register domains: then you ought to deliver. Such commitment always involves a clear marker: the Sunrise period. If you commit to a Sunrise period - you need to execute one. If you can prove "public interest" through an alternative business model ('.disaster') - then state so in your application; and do NOT commit to a Sunrise period. You will then be subject to scrutinization of your "public interest commitment" of course; if it holds water: you are fine.

Jeff:Regarding the issue of the definition of Alexander Schubert  –  Feb 17, 2020 5:50 PM

Jeff: Regarding the issue of the definition of "generic keywords": I think we might have to categorize. Likely most of us would agree that all dictionary entries are "generic keywords". In regard to the expansion of the DNS I think emphasis should be put on "category defining generic keywords": "Car, lawyer, realestate, cloud" for example. What George tries to convey: There is a very high risk that in the 2nd round MASSES of industry defining generic keywords will be "safeguarded" by the relevant industry leaders. Here is how it works: Consultants did in 2011 (and will do so even more in 2021) swarm out and present to Brand Managers as "matter of fact" that "the competition" has nefarious, wicket plans to get hold of and then exploit for their own benefit "their" industry defining keyword as new gTLD. These consultants will suggest that the only protection and remedy is to submit an application pertaining said keyword - just to see who else is applying and/or to simply "bury" it. In combination with extraordinary low application fee floors (anything below US $100k) this will lead to a large scale "privatization" of category defining generic keyword based gTLD land. Question is: Are we aware of that risk? We are! Would it serve ICANN's mission to allow this to happen? In my opinion: No!

The Public Interest? John Poole  –  Feb 17, 2020 1:44 AM

Thank you George. Apparently you are one of the few “good ones” left at ICANN who actually care about the global public interest. Almost everyone else at ICANN, including its so-called “ICANN community” are self-seeking opportunists with vested interests. ICANN’s previous expansion of gTLDs has been a fiasco on so many levels, and I expect the same in the next round(s). Brand gTLDs were a MISTAKE - a trademark is neither a monopoly nor generic - but ICANN org & back end registry operators, trademark lawyers & brand consultants, were greedy.  As you note, “the ICANN multistakeholder experiment in Internet governance is inconsistent with a commitment to uphold the global public interest, and ... it has failed in action.” Again and again.

Brand gTLDs are a fantastic tool Jean Guillon  –  Feb 17, 2020 7:21 PM

dotBrand new gTLDs for Trademarks are a great tool for notoriety and branding. The only problem is their price: $25,000 a year in ICANN fees blocks many from accessing such tool. Let's hope this changes in the future.

Dear Jean,What application fee floor / annual Alexander Schubert  –  Feb 17, 2020 9:51 PM

Dear Jean, What application fee floor / annual fee would you consider appropriate? "Cost-based" could mean as little as US $10k application fee & US $5k annual fee if we had many tens of thousands of brand gTLDs. Would you welcome that? Thanks, Alexander

I welcome thousands of .BRANDs Jean Guillon  –  Feb 18, 2020 5:38 PM

We have the experience of round one on both sides (gTLD applicants and the ICANN) so validating a .BRAND Top-Level Domain (specification 13 or equivalent from the next round) should be faster and more efficient thank to the appropriate Working Groups. The ICANN application fee for .BRANDs could logically be lowered: keeping a $185,000 would shock me. The actual 25K/year "tax" is prohibitive for a Trademark compared to the cost to acquire a traditional domain name. I don't understand why there would be an annual tax for TLDs/Registries with a number of domain names created below the 100 but I guess that a tax could be acceptable since the ICANN here grants access to the root, which is different from using a domain name (some Trademarks have thousands of domains and they don't pay that tax). Even a $1K/year tax increases drastically the price of a single domain name (+Backend Registry fee, +Escrow fee +annual Domain name fee...) Come on...who can afford this and for what use?

GREAT ARTICLE! Alexander Schubert  –  Feb 17, 2020 5:27 PM

First: THANKS dear George - I am member in the new gTLD subsequent round PDP group and am trying to convey this line of thinking since years.

You are right by stating:

“As an example, the new gTLD .book is an open generic gTLD, and anyone should be able to apply for a second level domain name registration under it. Alternatively, the registry may have a stated policy to restrict registration, but must allow domain names to be registered by all ‘similarly-situated’ businesses or industries, e.g., all cloud companies in .cloud, all search industries in .search, and all bloggers in .blog.”

The key is:

“Must allow names to be registered….”

Here what happened in the 2012 round: Some applied for closed “generic category defining keyword based string” gTLDs - then the board forced them to revert to be “open gTLDs” - but they NEVER LAUNCHED! Despite having launch commitments in their applications (sunrise phase).

In a few years their contract renewals come up: And they WILL be renewed. We should not let that happen in the 2nd round:

If you apply for a generic “category defining keyword” based new gTLD; you are forced to contract and enter the root (by penalty of losing the string); and your ONE AND ONLY job is to “allow names to be registered”; if you can’t execute that within a DECADE: why should ICANN allow a renewal?

The trick here is: Some apply for an open gTLD - never launch, hereby revert it to a closed gTLD (they can then still run 100 domains - e.g. your own portals and products).

Contract renewal should be denied to registries that have committed to -but never executed- a sunrise period. That still allows for new business models that do not require a sunrise phase: just do not commit to one and you are fine. If you can ONLY prove “public interest” by promising public registrations of domains - you need to deliver on your commitment. If you can prove “public interest” in another way: put it in your application and let ICANN decide whether it’s sufficient. Just do NOT engage in “bait and switch”: commitment to public availability (via Sunrise) - then never start such launch period. Such behavior should be grounds to the denial of contract renewal.

Comment Title:

  Notify me of follow-up comments

We encourage you to post comments and engage in discussions that advance this post through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can report it using the link at the end of each comment. Views expressed in the comments do not represent those of CircleID. For more information on our comment policy, see Codes of Conduct.

CircleID Newsletter The Weekly Wrap

More and more professionals are choosing to publish critical posts on CircleID from all corners of the Internet industry. If you find it hard to keep up daily, consider subscribing to our weekly digest. We will provide you a convenient summary report once a week sent directly to your inbox. It's a quick and easy read.

I make a point of reading CircleID. There is no getting around the utility of knowing what thoughtful people are thinking and saying about our industry.

Co-designer of the TCP/IP Protocols & the Architecture of the Internet



Domain Names

Sponsored byVerisign

IPv4 Markets

Sponsored byIPv4.Global


Sponsored byDNIB.com


Sponsored byVerisign

Threat Intelligence

Sponsored byWhoisXML API

Brand Protection

Sponsored byCSC

New TLDs

Sponsored byRadix