Given that human categorizations of anything are necessarily constructs, I don't imagine the demarcation will ever be a fixed thing. It's the nature of constructs, yes? It seems all the vocabulary humans use to describe the world is ultimately some sort of construct. Even things that are taken for granted, like, say, labeling colors. It was an interesting thing taking a color theory class in college and seeing arguments erupt about where the color "green" ends and "yellow" begins. It's subjective, and the lines are only there because humans decide to put them there. And, since humans never entirely agree with one another, our maps of the territory are ever fluid.
There are more objective ways to approach things. Rather than argue about where "green" ends and "yellow" beings you can talk about the hue, saturation, and value in terms of a number without any other labels. How would an approach like that work for sciences? Not sure it could be done, given the sciences are not something that can be simplified into easy mathematical quantities. Oh well.