It's about time that ICANN finish what it started in 2000 and introduce real competition.
We focus too much on participating in the tacky battles over TLDs, and not enough on rescuing as much as possible of the Internet's value from the effects of those battles. The independent actions taken by a number of subnets to counteract Verisign's show that a consensus view can prevail on the Internet against central mismanagement. Now, we need to apply that consensus in a strategic, rather than reactive way.
There is an unlimited supply of domain names, just not at the top level. We should transition to a system with unlimited top level names. This approach preserves a coherent global name space, starts every top-level domain system out on an equal competitive basis, and allows the most attractive subdomains to capture the most attention.
The cost of a level playing field at the top is that top-level domains must be meaningless tokens, such as numbers, instead of mnemonic and meaningful names. This decouples the permanent handle service from the name lookup service, so that battles about names do not derive from monopoly control of handles, and do not degrade the value of handles, as they do today. Numbers not much longer than international telephone numbers could give everyone a TLD.
No, I don't expect people to remember these numbers (although we do remember telephone numbers to a remarkable degree). The useful/popular numbers will get bookmarked. Users will be able to choose which ones to bookmark, or to choose software partly on the basis of its support for their favorites. But a neglected domain will never face a barrier much higher than the barrier of advertising a telephone number, which is done quite successfully already.
Even better, using public-key signatures and secure hashing, top-level handles can be self-assigned, and verifiable without resort to a trusted authority.
But, how to get this started? How to make the incredibly difficult arguments to ICANN? All unnecessary. The handle system does not need to start out at the top level. The top level has the advantage that everybody remembers it, but a lower-level domain is just as good as the root if people use it. We just need one sponsor with the ability to register and defend its own domain (say, nicesponsor.org), the willingness to establish, (say, handles.nicesponsor.org), and the technical oomph to run a popular DNS zone server.
If nobody uses handles.nicesponsor.org, very little is lost. If it becomes popular, it can be promoted to the root if and when the world wants it. Or, it can just become so popular with its initial name that it doesn't matter that it isn't the root.
Bob Frankston wrote this all up pretty nicely, and called it dotDNS [circleid.com]. Let's do it. Who wants to be the sponsor who saves the Internet from WIPO and Verisign?