I think that's a very grave fallacy. If
w3.org is unique enough for an email domain then
w3.org followed by a short name is unique enough to identify an XML vocabulary. The system works perfectly well for Java package names.
URIs for namespaces are misguided. They use a clumsy syntax that's highly redundant; the syntax rules themselves are complex, unclear, and unstable; equality matching between URIs is ill-defined; the hierarchic relationship of the components switches in mid-stream from least-significant-first to most-significant-first; and they create the wrong impression that they are identifying resources on the web. Need I go on?
