I think that your point is not so off-topic as Orias suggested. We have names for racial groups. They are enshrined in law, so the government has tried to define them. However, racial groups suffer from the same semantic vagueness problem that word meanings have. That is, we have prototypical concepts of a "white person" or "black person", but the prototypes are based on associations of properties. The moment you try to get serious about defining racial groups in absolute terms, you quickly find that there are in-between areas where the criteria fail. It turns out that there are no objective genetic correlates to race, just a lot of different traits that act in combination to trigger the intuition of a coherent match against a prototype. Semantic vagueness is a real problem for lexicographers.
The same can be said for the meanings of words like "atheist" or "god". We have good examples of people and beings that fit those categories. Richard Dawkins is pretty typically an "atheist" by anyone's judgment, and the Christian God is a "god" by anyone's judgment. But what happens when you move away from the prototypical center? When do you stop calling someone an "atheist" and switch to another name? Or do you just try to stretch the label into a "one size fits all" category?
Making up a definition that satisfies everyone is problematic, because the prototypes that we develop as we learn and use language turn out to be based on personal experiences. Everyone's prototype is slightly different, and the perception of identical lexical meaning is something of an illusion from the individual's perspective. For some people, a collie is closer to the prototype for a dog. For others, it might be a beagle or a terrier. Human races exist in the same sense that different "species" of dogs exist. In reality, they are all the same species that fall into somewhat arbitrary subgroups.
