Is Research Data Management Ruining Science?

Alvin Weinberg in 1961 asked whether Big Science is ruining science, and he gave an answer that is similarly valid for research data management: "[...] one sees evidence of scientists' spending money instead of thought. This is one of the most insidious effects of large-scale support of science. In the past the two commodities, thought and money, have both been hard to come by. Now that money is relatively plentiful but thought is still scarce, there is a natural rush to spend dollars rather than thought." [1] The result is what Richard Feynman termed "Cargo Cult Science": research that appears scientific but has neither scholarly contribution nor impact. [2] Science is about understanding, and data are not insight, but play the role of critical arguments at best.[3] Hence we need to remember what science and scholarship is all about, and learn to apply the scientific method to our research. Only one part in this is managing our research data, and rather than reinventing the wheel over and over again, we better learn from the expertise of other disciplines, such as engineering, software engineering, or library sciences – and other attempts similar to the NFDI. [4] If we were really to advance science, we better understand that research can be without scholarly contribution, and that data, let alone sharing data, is a highly non-trivial concept resting on many implicit assumptions often not fulfilled. [5] Here, I present a series of (digital) tools developed for "little science" in a spectroscopy laboratory predating the NFDI by several years that have become solutions for my daily work, span the entire data life cycle, and help with enhancing the overall quality of the research: Data provenance during data acquisition [6], an electronic lab notebook (ELN) [7], a framework for fully reproducible data analysis providing a gap-less and complete protocol of each step and relieving the user from actually programming [8],[9] and packages based on it [10]–[15], a repository for "warm" research data [16], lab management and a knowlede base [17]. Furthermore, two lecture series on scientific software development [18] and research data management [19] have emerged from 15+ years of reflecting on the essence of science. We need to teach the students early on what science is all about and why properly handling research data is a prerequisite for scholarly contribution, rather than assuming that they've "caught on by osmosis". [2] "At stake is the future of scholarship." [5]