What's so difficult to understand about what Paul said?
You don't understand how DNS works.
I think he does.
As long as the Internet's cache files on each and every system out there point at "a" in their file, there is the potential to break the namespace.
Heck, I guess that would be true of "F" as well. Shall I break the name space? No? Why not? Is it because coherence has great value? OK, so how shall we determine the synchronization signal for this coherence -- that is, who can vary and who is required to follow? The answer is that the owner of a zone can vary, and the publishers of a zone have to follow.
One rogue server in a confederation will cause serious problems.
Which is why I expect that the current InterNIC contractor (NSI) will do whatever the owners of its published but unowned zones (".", MIL, EDU, GOV) tell them to do.
NSI has defacto control, because getting them out of the cache files is a long and slow process, and until they ARE out their answers will be believed.
If NSI tried to become a DNS pirate, I expect that IANA would publish a new "named.cache" file without NSI in it, and that the world would switch in a week or less. Nobody likes DNS pirates -- or hadn't you noticed, Karl?