I should have made my comment more specific: what is the problem with single namespace without ccTLDs and without per-country exceptions?
Thank you for asking. Harald Alvestrand and I had just this conversation during the IETF IDN WG lifetime, about the point where the Chinese (CN, TW, MO, SG), the Koreans (SK), and to a lesser extent, the Japanese (JP) in the Chinese Domain Name Consoritum (CDNC) and/or the Joint Engineering Taskforce (JET) found IETF consensus process inalterably for an ASCII encoding of a naive transformation of a glyph repetoire (Unicode). The CDNC et al were unable to get an intermediate mapping of the code point repitoire, and proposed an alternative, scoped semantics for code point equivalency classes. That was your question right, what use is there for scoped semantics? Perhaps none, but the CDNC/JET technical people I knew, and the policy people I knew at the time were quite willing to accept all the flagday issues for domain names characters in infrastructure that were outside of the current repitoire. I think everyone here knows what those issues are, and how great a cost their resolution represents.
Per-country exceptions just creates more Balkanization of the Internet, which hardly seems beneficial.
That was Harald's arguement, and as IETF Chair, it carried much more weight than that of any other person I've ever known, in China or outside of China. The "principle of least surprise" ment that a zone file operator (in China) could not create an equivalency class a user (in Norway) would be unlikely to predict. I suppose I should mention that in mainland China, a simplified (modern) form of Han characters are used, in the province of Taiwan, traditional Han characters are used, in Korean some archaic Han characters are used, and in Japan, in the Kanji writing system, some (other) archaic Han characters are used, and in Vietnamese, still another set of Han characters are used -- and there are scads of semantic equivalencies between these different glyphs, all of which are in Unicode, without an equivalency class mechanism. And so we (or rather "they" since this is a North American list) do not have domain names composed of end-user recognizable characters. Oh. While in hospital in Beijing I asked all the medical staff (nurses, doctors, etc.) if they were "OK with ASCII". Not one English speaker was. Limited sample set, your milage may vary, season for taste, etc. It is fellicitous, but the ICANN Registrar's Constituency list just a day ago carried a request, nominally from ICANN President Paul Twomey, for a Registrar with some interest and experience in the problem area to join a President's mumble. I wrote he and Vint to see what they had in mind, and I may as well use this note to prod them again. They may simply mean that RACE needs to be re-euphamized and a few more printer glyphs in Unicode need to be made less accessible to phishers. Eric