I think this is actually a major point that doesn't get discussed enough. Our human impulse to apply a rigid taxonomy to things, to model them as discrete, absolute units instead of complex systems, is obviously very useful and intuitive, but comes up short when we try to study things of sufficient complexity.