So what is “science,” then? Personally I want to view it as an extension of the activities in which we all take part in everyday life. What we all want is success in what we do, and I cannot see why science should have any other goal. Thus, science should be practically fruitful to merit its license.
This is what, in the context of philosophy of science, is usually called instrumentalism – a view which is often subject to very little discussion (if discussed at all) in the literature, probably because it has never been very popular. The most famous proponent of instrumentalism is probably the economist Milton Friedman, who claimed that unrealistic assumptions in economics (and, one would assume, also in other disciplines) is not a problem as long as we can, with the help of those assumptions, develop hypotheses that can be tested by correct predictions.
Now, it may happen that Friedman’s own theories was not that strong when it came to providing correct predictions; but I still think there is something in instrumentalism as a philosophy of science. What separates instrumentalism from “regular” science, i.e., science directed at finding causal links? A significant affinity seem to exist, but it is probably true to say that instrumentalism is more “tolerant” than, for instance, logical positivism. It does not deem “metaphysical” or other unclear statements to be illegitimate as long as they are fruitful when it comes to producing empirically verifiable hypotheses. And according to instrumentalism it is not as important to lay bare causal mechanisms as clearly in order to talk about a “legitimate” scientific hypothesis. If we can find strong correlations between variables, then that might very well help us to handle reality better, even if we fail to see the “underlying” mechanism absolutely clearly.
One could also claim that instrumentalism is more tolerant than Popper’s principle of falsification. Making predictions in the social sciences (and sometimes also in the natural sciences) is not about finding “laws” which are always valid. The hypotheses we work with pertain to probabilities, which means that one can always find cases that do not behave as the hypothesis predicts. One might, perhaps, ask whether such probabilistic predictions are of any use. Some philosophers of science claim that instrumentalist science cannot really contribute much, since it cannot improve a lot upon the “folk psychology” and intuitive understanding of the world which we all possess. Personally, However, I think that it is possible (and it happens all the time) to conceive of (and verify) probabilistic hypotheses that go counter to what we intuitively think is true and that improve our folk psychology. Nevertheless, there is some affinity between instrumentalism and falsificationism, if one by the latter mean that repeated falsifications must take place in order to reject a hypothesis (if one, in other words, can show that the probabilistic predictions made do not seem at all to be in line with the observed frequency).
The clearest difference between instrumentalism and other philosophies of science can be found when it is compared to the sort of social (and humanistic) science which is not at all interested in formulating verifiable hypotheses. To “interpret” or “understand” the world cannot be viewed as science, according to the instrumentalist, unless we by, for instance, understanding mean the understanding of observable consequences with the help of predictive hypotheses, or the like. By the same token, studies which only aim to describe the world cannot be counted as fully scientific, even if descriptive studies are usually quite valuable (to say the least) preparations for the scientific activity of formulating hypotheses. It is, however, important that descriptive studies are made in a way that is fruitful for continued scientific use.
The hardest things to defend as “scientific” in the court of instrumentalism is probably very abstract theorizing, or different kinds of case studies, which experience has shown to be pragmatically unfruitful. This does not mean that such studies cannot be valuable for other reasons (they might, for example, be pleasurable to conduct or to read), but in theses cases, the label “science” may not be appropriate to apply.
Lastly, however, we should add the following point: if science is only supposed to serve our pragmatic goals, who chooses which goals are relevant? My thinking is that this is ultimately decided by those who pay for the research, and in this case we are hopefully talking about democratic institutions which have enough sense to understand that science thrives when there are many approaches and perspectives that work side by side. But “science” that decade after decade produces results without any pragmatic relevance at all should not be paid for by the public purse. (If there are private actors prepared to continue the research one can, of course, turn to them.)