Among the innumerable characteristics of the social sciences which in reality is not much sense to have this our inveterate habit of applying a theoretical discussion elements epistemological: This theory does not serve (or the other way around, is more better) because you do not agree to (or complies) with such basic characteristics of the contemporary philosophy. Systems theory in its version luhmanniana is one of them (but not the only): it is the great alternative for the social sciences today because it is in accordance with the most advanced epistemology.
Now, in reality one should use a different perspective: The theories are no more or less good depending on its characteristics, epistemological. Epistemology can be an interesting field of study, but it does not serve to guide the development of a science (and if one wants, has abundant foundation the epistemological to defend the idea that it should not be taking into account what it says in the epistemology to investigate, Rorty and Feyerabend dixit). What can guide the theoretical development is that which ‘serves’, in other words that which has hits like perspective.
At the end of the day, in the midst of visions about the science required to explain it all according to mechanistic concepts, appeared the theory of gravity newtonian. A theory from a perspective of mechanistic is a scandal (it has action at a distance, something that simply should not occur if one believes that everything is explained by mechanisms). In fact, it was one of the problems in the adoption of the theory: what is the cause of gravity? But, ultimately, the fact that the theory ‘worked’, and had many successes to its credit, had more than any weakness epistemological.
Theories that can be awful, epistemologically, to have millions of problems, and not to be sufficiently consistent and complete of all forms can be good theories: explain any thing.