Suppose you did not have any knowledge of socionics, but you did have the vague intuitive idea that people differ in terms of personality in fundamental ways to the point where dividing them into sepparate groups is meaningful. How would you then go about discovering exactly how many groups there are, what the commonalities of the people in the groups are, and how they should be named?

Another way to formulate the question: if socionics turns out to be bull-shit, but we stick to the idea that people can be meaningfully classified; how could a theory of personality be formulated and tested in a way that precludes us from making the mistakes that were made during the previous attempts?