Having thought about this a bit more, I think we need to separate the two parts:
-
Namespace - a single immutable label that refers to a single immutable concept for ever. This should be like a character set eg ASCII or UTF-8. The code means the character. That’s all. No logic and no ontology at this stage. This means that (apart from probably needing to expand the namespace every so often) that code will always mean that character.
-
Everything else - the hierarchy, ontology, combinatorial logic etc is added as a further layer or set of layers on top of 1). Doesn’t have to be managed by the same people who manage the namespace. There can be a multitude of different hierarchies, ontologies, classifications etc (as indeed there already are). But this would work in the same way as French and English having their own, perfectly internally consistent rules, despite using the same character set. As long as you have a common character set, though, you can build tools that work for both.