Basically, it'd mean the (virtually impossible) task of deprecating the non-country TLDs with X years before they stop working, and saying: "Okay, .us is managed by whatever infrastructure the USA sets up, and .uk is managed by whatever the UK sets up, etc."
So the US might choose to allow cocacola.us, while the UK might have on a grouping scheme like cocacola.co.uk, and meanwhile North Korea refuses outright or insists on cocacoala.obeys.dear.leader.nk, whatever.
Countries can break up, dissolve, or form too. I'm not sure that making all domains depend on things that exist in the real world is a good idea. They should depend on abstract ideas that are less liable to change and can be resold easily to different operators.
Not that often, and even when they do it often means very real issues in terms of governance of internet infrastructure.
> I'm not sure that making all domains depend on things that exist in the real world is a good idea.
But domains already depend on things which exist in the real world! Not only are they leased by real-world entities, but even the names of domains already refer to countries, companies, brands, products, activities, clubs, and people.
The real difference with what we have now is that it's a weird global monopoly that has shown it's willing to erode organizational principles in order to make a buck.
Splitting it up by sovereign nations is actually moving back closer to the decentralized principles of the internet, avoiding a single point of (political) failure, and creating subsections which can evolve separately and cover for one another.
Ex: Turkey's emerging dictatorship refuses to let anyone register erdogansucks.net.tr ? Fine, go register erdogansucks.net.de , and advertise with that instead.
So the US might choose to allow cocacola.us, while the UK might have on a grouping scheme like cocacola.co.uk, and meanwhile North Korea refuses outright or insists on cocacoala.obeys.dear.leader.nk, whatever.