Conclusion
The main purpose of this paper was to cope with some difficulties with which robustness as consilience between different and independent pieces of evidence is confronted, difficulties which have indirect consequences for the problem of scientific realism.
For this purpose, two important meanings of “robustness”—robustness regarded as reproducible stability as against perturbations and variations in parameter values (robustness-as-stability) and robustness as consilience of results from different and independent pieces of evidence (robustness-as-consilience)—was examined.
Following Hacking 1983, who implicitly referred to robustness in his criticism of the no miracle argument, I have argued that, from an experimentalist point of view, the epistemological meaning of robustness-as-stability is not foreign to robustness- as-consilience. More precisely, robustness-as-stability and robustness-as-consilience are to be seen as two cases or specifications of the same thing, that is, of intersub- jective reproducibility as the most valuable clue to reality. As such, though conceptually distinct, they are not only intimately connected, but need one another and supplement one another.
There is no vicious circle involved here, but a technical-practical synergy-or a robustness of higher order-, which is at the heart of the experimental method, and which can help us out of two main difficulties concerning robustness-as-consilience. On the one hand, it is no objection that robustness-as-consilience reasoning cannot be credited with epistemic warrant if one lacks a minimally reliable observational procedure, since what is important in science is not robustness-as-consilience in itself, but only in its synergy with robustness in the sense of technical-experimental reproducibility. On the other hand, it is true that it is impossible to settle once for all that different causal chains, though experimental-operationally established, are also independent.
However, at any particular time, thanks to the mentioned synergy, we have at our disposal a great number of reliable findings that we are not willing, at least provisionally, to put in question. Thus, from a technical-operational point of view, we may say that the truth is something like ‘what shows itself most robust' (in both senses of the world) in the widest context and the longest run.Acknowledgements I presented an earlier (and much briefer) version of this paper at the Congress of the “Academie International de Philosophie des Sciences” at the University of A Coruna, Ferrol Campus, Spain (September 23-26, 2015). Thanks to all those who contributed to the discussion of the paper during and after the conference, among which Thomas Nickles and Alberto Cordero deserve special mention. I am grateful to Mike Stuart, who read a draft of this article and provided extensive and helpful comments. Italian Ministry for Scientific Research (MIUR) provided funds for this research (PRIN 2012).