Calling DNS a "hack on top of" IP addressing seems a bit disingenuous to me. DNS is not going away any time soon. You may as well compare assembler with a high-level programming language, or perhaps even machine language with assembler. This is splitting hairs for rhetorical purposes, but not fundamentally contradicting the point that control of DNS makes a difference as long as it remains the primary location identifier used by Internet applications, and there is realistically no end in sight. People were calling DNS a flash in the pan years ago, but alternative paradigms like keywords have not taken off. And at such time when DNS is replaced by an alternative (hopefully an improvement), do you expect that ICANN will not take ownership of that as well, and all the issues that derive from DNS today will not continue to be a serious policy issue deriving from the new paradigm at ICANN? Doubtful. This is ultimately about legitimacy of policy processes at ICANN (and the alternative of locating those processes in other political venues where the issues extend beyond purely technical and operational matters), not about the specific paradigm that is currently triggering the debate.