And if your standard fixes a prefix to mean a URL always, why use a URL or XML Namespaces at all? You can do everything you need to do with the text value of the qname: you dont need the PSVI. (In which case, you are only using namespaces in case some other foreign name uses them, not because they give you any value.) You get the complexity of namespaces, just an extra layer of indirection but enough to be a straw (i.e. camel case), without the modularity that namespaces was designed for.
XBRL is a pretty extreme case, as far as building the spec's formalism on the bouncy castle of XSD: I dont know that the result has helped implementors really. (I spent a year on an XBRL project. They used Schematron for testing document roundtripping through XBRL.)
Rick