The topic of realism has come up frequently here -- causal realism, critical realism, scientific realism. Each of these realisms comes out of somewhat different fields of questions and assumptions. Within mainstream philosophy of science there is another realism that has been debated in the past twenty years, referred to as structural realism. The view has been developed by philosopher John Worrall, and his 1989 article "Structural Realism: The Best of Both Worlds?", sets the stage (link). So what is this view, and does it have any relevance to the social sciences?
First, what is the view? It is a refinement to the theory of scientific realism advocated by philosophers like Hilary Putnam and Dick Boyd -- the view that we have reason to believe that the world has approximately the features attributed to it by the best available scientific theories. As Boyd put the view quite a few years ago, what else could explain the success of those theories if not their approximate truth and successful reference to the entities and properties of the world?
The problem that gives rise to structural realism is what Worrall calls the "pessimistic meta-induction" (109): in the history of science, most scientific theories have eventually been proven to be false. So how can scientific realists claim, after all, that there is a rational basis for believing that the world has the characteristics asserted by the current generation of scientific theories? The answer to this question, Worrall argues, comes down to a judgment call about the history of science: "just how radical theory-change has standardly been in science" (105). If successor theories have nothing in common with their antecedents except a broader but overlapping range of empirical consequences, then it is hard to say that there is an approximate truth that is captured by both stages of the theory. "If, on the contrary, the realist is forced to concede that there has been radical change at the theoretical level in the history of even the mature sciences then he surely is in deep trouble" (107). Realism, then, depends on some degree of approximate continuity across successor theories. Here Worrall turns to Richard Boyd:
"The historical progress of the mature sciences is largely a matter of successively more accurate approximations to the truth about both observable and unobservable phenomena. Later theories typically build upon the (observational and theoretical) knowledge embodied in previous theories." (Boyd, 1984, "The Current Status of Scientific Realism" in Leplin, ed., Scientific Realism)
But many philosophers and historians of science have disputed the degree of continuity that Boyd postulates here. They emphasize the discontinuities that often occur across the process of theory change in physics. However, Worrall argues that there is a more abstract way in which physical theories show substantial continuity. This continuity isn't found at the level of entities and causal powers, but rather a set of more abstract characteristics that are attributed to the features of the world under study.
Structural realism gets going, then, if we concede that the history of physics shows radical change at the level of the properties attributed to natural objects but we maintain that it also shows a strong degree of continuity when it comes to the basic structural properties that are postulated by theories of physics.
In application to the series of theories offered to explain the behavior of light, the continuity was abstract:
There was continuity or accumulation in the shift [from Fresnel to Maxwell], but the continuity is one of form or structure, not of content. (117)
Worrall attributes this idea about a specific but abstract kind of continuity in physics to Henri Poincare, and he argues that it lays the basis for a weaker form of realism that might be described as syntactic or structural realism(117).
Roughly speaking, it seems right to say that Fresnel completely misidentified the nature of light, but nonetheless it is no miracle that his theory enjoyed the empirical predictive success that it did; it is no miracle because Fresnel's theory, as science later saw it, attributed to light the right structure.... There is no elastic solid ether. There is, however, from the later point of view, a (disembodied) electromagnetic field. The field in no clear sense approximates the ether, but disturbances in it do obey formally similar laws to those obeyed by elastic disturbances in a mechanical medium. (117-118)
So structural realism when applied to the history of the theory of light says two things: successor theories had radically different and inconsistent hypotheses about the mechanics and substance of light; but they agreed approximately about the mathematical properties of light. And it is the latter that is preserved across the progress of this area of science.
This is a very weak form of realism, as Worrall acknowledges:
[The structural realist] insists that it is a mistake to think that we can ever "understand" the nature of the basic furniture of the universe.... On the structural realist view what Newton really discovered are the relationships between phenomena expressed in the mathematical equations of his theory, the theoretical terms of which should be understood as genuine primitives. (122)
So the commonsensical questions we might want to ask of contemporary physics -- are there electrons, is space curved, is the speed of light constant -- do not have defensible answers, according to structural realism. What the success of modern physics allows us to conclude is something much weaker: whatever the fundamental components of matter, space, time, light, and gravity are, the world conforms to the mathematical transformations that are specified by our best confirmed contemporary physical theories. It is the transformations, equations, and constants that we can be realistic about, not the concrete theories of the mechanics of the things that embody these equations.
My real interest in opening this topic was to consider whether it has any relevance to the social sciences. And the short answer seems to be -- not much. Theories in the social sciences rarely have the mathematical specificity that is crucial to the structural realist argument. So it is difficult to make the argument that Ricardo, Marx, Pareto, and Keynes were describing the same structural reality when they wrote about capitalism. Their substantive assumptions are quite different; but further, the expected "mathematical" behavior of the capitalist market system is also substantially different across the theories. Perhaps a more plausible case is the transition from Marx's classic theory of exploitation, based on the labor theory of value, to John Roemer's theory of exploitation in A General Theory of Exploitation and Class, based on neoclassical and game-theoretic economic assumptions. The two theories arrive at similar "structural" features of a capitalist economy, in spite of the fact that the underlying substantive assumptions are quite different.
(Katherine Brading and Elise Crull offer a very nice treatment of Worrall's interpretation of Poincare in "Epistemic Structural Realism and Poincare's Philosophy of Science (link).)