Kuhn’s own emphasis on science as a puzzle-solving enterprise would lead one to interpret him in an instrumentalist manner. [Ernan McMullin, "Rationality and Paradigm Change in Science," in Paul Horwich, ed., World Changes: Thomas Kuhn and the Nature of Science (Pittsburg, PA: University of Pittsburg Press, 1993, 2010), 71]
His [Kuhn’s –DHC] argument that theories on either side of a revolution are incommensurable because the meanings of terms change so radically applies also, and perhaps more plausibly, to the names of disciplines (J.L. Heilbron, “A Mathematicians’ Mutiny, with Morals,” in ibid., 107)
Yet, to admit that observation is theory-laden is a long way from denying that there is a theory/ observation distinction. (Nancy Cartwright, “How We Relate Theory to Observation,” in ibid., 259)
The natural-kind terms current in an old science cannot be translated into natural-kind terms in a new science. (Ian Hacking, “Working in a New World,” in ibid., 278)
The world in and with which we work is a world of kinds. The latter changes; the former does not. After a scientific revolution, the scientist works in a world of new kinds. In one sense, the world is exactly the same. A change in the class of sets of individuals that correspond to scientific kinds of things is not a change in the world at all. But in another sense the world in which the scientist works is entirely different, because what we work in is not a world of individuals but of kinds, a world that we must represent using projectible predicates. (Hacking, "Working," in ibid., 306)
One word of describing this difficulty [concerning translation —DHC] is as a case of polysemy: the two individuals are applying the same name to different concepts. But that description, though correct as far as it goes, fails to catch the depth of the difficulty. Polysemy has a standard remedy, widely deployed in analytic philosophy: two names are introduced where there had been only one before. If the polysemous term is ‘water’, the difficulties are to be lifted by replacing each of the concepts that previously shared the name ‘water.’ Though the two new terms differ in meaning, most referents of ‘water1’ are referents of ‘water2’ and vice versa. (Thomas S. Kuhn, “Afterwords,” in ibid., 318)
Thomas Kuhn's philosophy of science can best be termed "instrumentalist," an epistemic antirealist position concerning science and scientific theories. The key thing to take note concerning Kuhn is that he is not a relativist or subjectivist, a label that is a better fit on philosophers like Michael Polanyi. Rather, Kuhn's method is a historical method looking at the history of the scientific enterprise, and wrestling with how previous generations of scientists have thought and worked on their theories.
One element of Kuhn's thought that had previously eluded me was his promotion of incommensurability, a term which seems to lead to a form of relativism. However, upon clearer reading, especially in this book edited by Paul Horwich, what Kuhn means by "incommensurability" has to do with theories and not the propositions of those theories. It is a term applicable at the meta-level, whereby different paradigms are so different that there is no way to translate one paradigm to another. Or, to put it perhaps in a way that may be easier to understand, it is impossible to translate a set of natural-kind terms to anther set of natural-kind terms, without losing some meaning or even most of the meaning in the process. In other words, it is possible to translate one paradigm into a set of propositions, and another paradigm into another set of propositions, but as a whole, the two sets are mutually incomprehensible and thus incommensurable.
There is no such thing as a "neutral" observation in science, something which sounds quite self-evident but is mostly lost to many scientists. Theory dictates what one looks for or what one expects to see, and this framing of the question makes observation in some-sense "subjective," or theory-laden. This is why incommensurability is so much more than merely interpretation of data, and why the word "incommensurability" is fit to describe the differences between two paradigms, instead of merely stating that they are differing interpretations of the natural world. Thus, in a sense, the world changes between paradigms.
Once all these are clear, then it is clear also why Kuhn's antirealism is not relativism, much less can he be used to promote the idea of science as a social construction. The natural world is objectively out there, but Kuhn's theory have to do with our knowledge of the world even to the point of observation. Now, if there is no transcendent external observer, then Kuhn's theory might lead to some form of relativism. But if Kuhn's theory is limited to humans, and thus an expression of human finitude in science, then Kuhn's theory is congruent with a belief in an objective world and absolute truth, insights which do not come and could not come from science.