RM: We clearly have some differences in the “definition” of “definition”.
I suppose it all depends on the sorts of things one wants to define, something we might call the context of application. I am not as much focused on using an ontology as a large online lexicon as I am on the task of acquiring scientific knowledge, so the sorts of things I need to define are complex systems of relationships, formal or mathematical models that we use as intermediate objects to deal with phenomena and the realities behind those phenomena.
Objects like that, intermediate and ultimate, typically have high levels of complexity that we are forced to approach in stages, often beginning with “toy worlds” in the classic AI fashion. Those are the sorts of definitions I am after. We could call them specifications if it helps to use another word.