by Elena Koustova

In the long gone Soviet higher (university-based) education system there were no electives. As aspiring biologists (e.g. natural scientists) at the Moscow State University, we were expected to study all kinds of physical and formal sciences, all with equal degree of dedication and rigor. One of the inescapable courses would bring the entire class into our largest school auditorium, weekly, for the full first year. Begrudgingly, we were taking the course “The natural philosophy and the history of biology” taught by the archaically looking Professor Piotr Matyokin. The rumors were that he was one of the last surviving (Russian) Habsburgs, and his careful recitations of Plato and the Sophists and the points about the perpetual tweaking of the scientific method seemed as antiquated as the Austro-Hungarian Empire itself. Nowadays, I astoundingly find myself going back to the essence of that class when I think about the science’s “reproducibility crisis”.

The opinion of the scientific community on the “reproducibility crisis” is not unified. While some deny the existence of this crisis altogether, others sound the alarm of various intensity going as far as to suggest that “a science war is in full swing which has taken science’s reproducibility crisis as a battleground” (Saltelli, 2018).

Historically speaking, the “wars” in science are nothing new. The origin of science as a privileged monopoly to reveal objective facts, truth, reason and reality driving the social progress has been the basis of an enduring Western tradition. This intellectual tradition has been vigorously challenged by similarly resolute schools of postmodernist philosophical thought, culminating in the Science Wars of the 1990th. The Science Wars (in its intellectual part) were believed to be a symptom of the deep unresolved conflict, internal to science, and the reflection of the scientists’ ambivalence about the object of scientific knowledge. The XX century Wars were themselves embedded into the ancient battle over the meaning of the terms “reason”, “truth”, “knowledge” and “reality”. In the Platonian-Aristotelean “Gods” against the “Earth Giants” battle, one had to choose between two positions. The “Gods” hold knowledge as being qualitatively different from beliefs and opinions. Knowledge “is that about which one cannot be wrong”, as it is necessary (must be true), certain, and universal (if it works in Athens, it works in Kansas). Sophists, the “Earth Giants”, were concerned with action, making decisions in real life. For them, knowledge is a species of belief most strongly held in the society at the moment. Knowledge is particular (works in the specific context), contingent (based on assumptions themselves not known to be true), and probable (with some uncertainty always present). The contemporary dominant idea is Platonic, but the “Earth Giants” were never completely extinguished, and the colossal dispute was never resolved. Today, there is no consensus on how knowledge (how do we know what we say we know) in the strict philosophical sense is possible. The enduring “knowledge battle” in philosophy was deeply internalized when the modern science was created, and the problem of “knowledge” muddies the problem of “scientific knowledge”.  What is the object of scientific knowledge: it is reality or experience? When the scientific discovery is claimed and published, what is it that the scientists claim they know? What does knowledge mean within science? It is easy to see the staggering responsibilities of the claims that we, scientists, reveal the truth about nature, as science alone claims to be capable of knowledge of the true causes of experience. Why are the climate change or Darwinian theories more threatening and highly contentious compared with the engineering theories that are used to build bridges? Because scientific theories claim to be true, beyond simply working. This point is vibrantly illustrated in the American sitcom “Big Bang Theory”, when Sheldon (the scientist) persistently puts down Wolowitz (the MIT-educated engineer).

Interestingly, for more than 400 years, science itself remains intrinsically ambivalent about the object and reality of the scientific knowledge. The conflict among competing conceptions of the problem of knowledge (e.g. what do we, as scientists, produce, the description of reality or experience?) is an ongoing process internal to science. Until the beginning of the XX century, natural scientists diligently contributed to the unravelling the “knowledge problem”, but after the 1920th, philosophers and historians took over the knowledge problem within science, and the scientists dropped out. The exception is the debate in physics (see Sokal affair, Bogdanov affair).
Let’s now revisit the reproducibility issue. Interpretations of the issue seem to fall into two categories (NPR and others):

  • This is how science works. Science is inherently uncertain, and contradictions happen all the time. The problem is that we do not know how to manage our expectations of science.
  • This is not how science works. Conflicting studies expose flawed or malfunctioning science. The solution is for science to change its practices.

Category 1 response is prominent among basic (exploratory) scientists, while the scientists working in translation are subscribing to the category 2 response more often. Researchers who are not particularly concerned with the fact that many research findings are not reproducible often invoke the view that science self-corrects. But if we assume that, as new observations accumulate, current theories are to be replaced, is that the true knowledge, in the strict sense, we produce or just the experience-validated educated opinions?

To confront the issue, the current efforts mostly focus on the technical machinery of reproducibility (e.g. raising the threshold for statistical significance, increasing funding for research replication, pushing researchers to share data, and setting standards that require journals to be more transparent about their process of peer review), leaving the broader methodological and philosophical questions of science unaddressed. Is it possible that, while focusing on research “machinery”, we are neglecting the ultimate scientific instrument, the scientific reasoning itself? Recent examples demonstrate that failures of reproducibility occur even in fields in which specific problems with good experimental design and data management do not arise, such as mathematics or computer science (Redish, 2018) and that precisely the same data can yield different results and conclusions when handled by different research teams. The founders of modern science (Galileo, Descartes) were exquisitely trained in formal sciences, such as mathematics and logic, and were obsessed with the applications of the scientific method, which essence has later been the subject of the writings (and disputes) of Karl Popper, Thomas Kuhn, Paul Feyerabend and Imre Lakatos. In the end, there is no and there has never been such thing as a single scientific method. There are principles of the scientific method, as opposed to a definitive series of steps applicable to all scientific enterprises. But what is the right way to reason about anything and where can it be learnt?

Perhaps, present day thinking about the difficulties in science reproducibility could benefit from a philosophical and historical perspective. Scientists decidedly staying away from the conflict over the nature of rationality, reason and truth and, at the same time, upholding that only scientists should speak about science, may miss the point that there is much more to “knowing” than just acquiring information. The research community’s ambivalence about the irreproducibility may be the manifestation of the veiled ambivalence about the object of scientific knowledge itself.  In my present position, I have not met any colleague that was formally educated on the scientific method or history of science, which makes me very grateful for our “last Habsburg”.