It’s a simple, straightforward fact that the root is not a TLD. However, the current policy around new gTLDs treats the root like a TLD registry and as anyone who runs a TLD registry knows, they have certain inescapable characteristics that may not be the best for the root.

In almost every TLD, once a domain name has been registered, the registrant can use it commercially with few restrictions:

1. they can be a TLD registry;
2. they can not be a TLD registry and:
   2.1 use it for their own online presence;
   2.2 not use it;
   2.3 lease it to someone else;
   2.4 sell it on.

As far as I can tell, all the same possibilities exists for the registrant of a new gTLD. I expect there will be some argument with me as to whether 2.2-2.4 are possible for a new gTLD but my point is about the difference between 1 and 2, is the new gTLD used as a TLD registry or not. My definition of a TLD registry is pretty straightforward, it means selling (or possibly giving away) domains to otherwise unrelated people. Just to be clear, giving domains to other parts of your organisation is not acting as a TLD registry, it is acting as an in-house registry.

At this point, some new terminology can help to distinguish the two uses. I’ll continue to use TLD or TLD registry as equivalents and introduce the new term Top Level Registration (TLR) for a registration in the root that is not used as a TLD registry but for one of the uses under 2 above.

To think through the implications of allowing TLRs, we need to start with the implications of failure of the TLR and what that means for risk and the cost of application:

• The impact of failure is much less for a TLR than a TLD because only the registrant of the TLR and their business channel are affected. Otherwise unrelated third parties are not affected. This contrasts with a TLD failure where multiple registrants and multiple business channels are affected.
• The registrant of a TLR only has the duty to protect themselves and not the duty to protect others and so the risk they take on in running the TLR is much less than running a TLD. This shows itself in areas such as insurance against legal liability claims or the investment cost of multiple redundant systems.
• The global reputational damage to the process for managing the root is much less when a TLR fails, again because no unrelated third parties are affected.

For all these reasons, the level of assurance required for a TLR applicant and the cost to ICANN to provide that assurance are much less than for a TLD applicant. If we recognise this then we can safely predict the inevitable demand for differential pricing of TLDs and TLRs in further rounds of new gTLD applications. The break will come when ICANN is forced to acknowledge that setting the standards for registering a TLR at the same level as for a TLD, is unnecessary and unjustifiable. For a body like ICANN it will become increasingly difficult to defend, from a competition standpoint, charging much more than the cost of registering TLRs in order to maintain artificial parity with the cost of registering TLDs. The people who will be demanding this split are of course the same people who have bullied ICANN over trademarks well beyond the point of fairness and there is no sign that their approach will change. No matter how much ICANN wants to (or claims to be able to) resist differential pricing, it will give in at some point.

This is where it goes horribly wrong because the moment that differential pricing is introduced, there’s only direction that TLR pricing can go—down. And as everyone with even a basic understanding of economics knows, when the price goes down, demand goes up. So the question becomes, at what price does an organisation decide to only go for a TLR and not register in a TLD—$5,000, $10,000, $50,000? Every process improvement for registering a TLR means a cost reduction, which means greater demand, leading to greater revenue and it spirals. ICANN will never be able to hold out as the one organisation that never refines its processes, never drives down costs and never offers a cheaper price to its customers for the same service.

This is very different from the cost of registering a new TLD registry, which will always be quite different. The impact will always be much higher than for a TLR and may even increase over time as use of the Internet becomes increasingly ubiquitous.

The end result of this spiral is a much flatter namespace. The survivability of DNS and the Internet in turn is predicated on multiple levels of domains under different operators. A flat namespace is probably more of a threat to the DNS than any of the imagined predictions of doom that are regularly promoted.

Not only will the namespace be flatter, it will also be tiered by wealth. Anyone with sufficient funds will be in the root and the rest of us under TLDs. If you want to imagine it getting even worse then all it takes is for the IP bullies to start on the “if a trademark is in the root then the same string should be blocked below” campaign and trademark offices the world over will be selling TLRs alongside trademark registrations.

If we want to truly protect the root then we need to limit the term of any TLRs that are granted in this round and ensure that future rounds only allow TLDs, which means treating the root like the root and not a TLD.