The belief that the universe is neatly arranged into hierarchical structures is flawed. This is generally a structure that we have imposed ourselves. It is not my argument that this model does not in some cases fit well (even perfectly on things with fewer features) with the world. It is, however an oversimplification of the facts. Surely there are certain cases within taxonomies where boundaries are somewhat arbitrary. The structure of the biological taxonomy comes from a period in time when there was no such thing as genetic typing. Some of the divisions must be a relic from this time. Within machine learning, in any real world problem of more than a certain level of complexity, it is generally impossible to find any way of creating a classifying function. This is because complex real world things generally don't exist in hierarchical structures. I'm a computational linguist myself, and can tell you straight out, there's no such thing as a noun, verb, whatever. I would argue that the only empirically real set structures are those in very pure science, that acknowledge a set of axioms and then say, "GIVEN this is true, blah, blah, blah". All that said, science has the right to say, "from the perspective of science, there is no whatever" (although I think it needs to specify which science, as they don't always agree). Science at least has a defensible methodology for setting up its taxonomies. The problem here is that people have a category system in their minds which is conflicting with the scientific version. The people in here are saying, "Okay, maybe you're right about the genetics thing, but there's still race, isn't there?" And it certainly does exist, because we as humans think it does. Yes it is a social and cognitive construct, but this does not make it any less real. You yourself, as well as having your scientific taxonomy will have mind categories that enable you to group people according to proximity to a racial stereotype. This ability may have served a Dawkins style genetic purpose at some point, allowing us to recognise tribes that were close enough relatives to share a few genes. Its a great shame that we still have these reductive categories, and a bigger shame that people believe them to have objective truth, and attatch unpleasant connotations to the groups. I don't think we have any disagreement here at all really. I think that I am just advocating a different approach. Don't be dismissive of people's categories. They are as valid as scientific categories in many ways. Make people understand that they are subjective, that they reflect your own mind more than the things they categorise. If anyone has a group entitled 'negro' it merely means they are a racist, not that the inhabitants of that group are negroes. If people choose to attach attributes like intelligence to these groups, again, it says more about their own.