DNS is a bad idea.
Simple problem. The idea of DNS is for names to be human-readable and -learnable. But if they are human-readable and -learnable, they have to be meaningful to humans. Which in turn means that some domain names are going to be more valued than others - to the tune of many millions of dollars. Where money leads, lawyers follow. It also requires the use of a centralised body - ICANN - because if some domains are highly valued by virtue of their meaning than someone has to decide who gets which one.
It's a poor system. But it's the only one we've got, and (As the above concludes) any alternative that uses names meaningful to humans would inevitably end up the same way.
ICANN appears to be doing it's best to live up to the stereotype of the evil, greedy capitalist right now. First by introducing many new TLDs when there was arguably little use for them, and then just selling off new TLDs in quantity when there is no significent benefit to anyone but ICANN's coffers.
Remember that when this fiasco is done, only two things will have been achieved. A new pile of money for ICANN and a few registrars, and people being able to access the Nike website by just typing 'nike' rather than 'nike.com'. Assuming, that is, that their browser doesn't interpret it as a search query and that their network admin has not been using ancient gods as a naming scheme.
In theory it wouldn't be difficult to make a completly new, decentralised DNS replacement based on simple public key crypto - if one were willing to accept the critical difficulty of having the addresses be uuencoded public keys, and thus just meaningless gibberish to humans. That defeats one-half of the function of DNS, and even with the rise of QR codes no sane company director is going to expect that lot to attract any customers.