Thank you, Madam Chair.
Thank you very much for inviting me to speak to this committee.
Your meeting comes at a crucially important time in the evolution of public support for science. My hope is to contribute insights from some 40 years of studying federal science policy in the U.S., informed by my discipline of science and technology studies, STS.
To begin, in democratic societies, the public has a right to ask that policies for science and innovation are being made for the benefit of the public as a whole. Therefore, in principle, we must have procedures to ensure the accountability of federal science and institutions. The question is not whether such a function should exist, but how it should be institutionalized and implemented.
This is not a new issue. How to hold publicly funded research accountable has been a topic of recurrent debate in the U.S. ever since the 1946 Administrative Procedure Act asked federal agencies to be answerable to the public for their uses of expert evidence.
In this century, the U.S. debate has often centred on the efficacy and integrity of federal science policy in areas of high public concern, such as health, environment and security. Is the government funding research that is reliable, useful, innovative and productive of public benefit? Are the funding policies and the researchers themselves properly answerable to the public?
Three points stand out from these debates that this committee may wish to consider.
First, accountability comes at a cost. A significant fraction of the costs of oversight are passed on to investigators, and they consume time, money and resources that could otherwise be dedicated to research. In my own experience, reporting burdens are continually ratcheted up and seldom relaxed. While the goals are often laudable, one-size-fits-all measures, such as the U.S. National Science Foundation's broader impacts criterion, can be onerous for early career researchers, inconsistently administered and too routinized to serve their original purpose.
Second, transparency can be a double-edged sword. In principle, the public has a right to know what public money is being used for. As Justice Brandeis famously put it, in government, sunshine is the best disinfectant, yet the principle of transparency has always been subject to limits when it comes to scientific data. Most obviously, data that could put vulnerable subjects at risk or violate their privacy should not be in the public domain. Even a presumption of full disclosure can prove detrimental in some contexts, such as public health epidemiology, because it shifts the burden of justifying non-disclosure to the researcher. As U.S. experience illustrates, transparency demands can be politically leveraged to discourage research that might otherwise be of great public benefit.
Transparency also presupposes that those to whom data is being disclosed are qualified to assess what is being shown. This is a risky presumption in the case of highly technical information, for which some level of baseline familiarity is needed to make sense of flaws in the research design or research results. In general, transparency provisions are most effective when the process of disclosure attends to how and by whom the research will be evaluated. Mandated data dumps rarely produce public benefit and may discourage important research from being undertaken. By contrast, well-structured review processes that include a diversity of viewpoints, such as the reviews conducted by the U.S. National Research Council, tend to yield higher-quality results.
Lastly, the committee has been asked to consider the establishment of an independent body or function to ensure the accountability of federal science policies. In my estimation, such centralization would be a mistake. It is reminiscent of the demand in the early years of the Reagan administration to centralize the capacity for risk assessment in the U.S. federal government. An influential National Research Council report concluded that such centralization would impose a conceptual straitjacket on research in diverse fields designed to meet widely varying public purposes.
While the analogy is not perfect, some arguments from that example remain highly relevant today. Science, despite flaws and occasional wrong turns, remains one of the most effective self-correcting human institutions, partly because, as the sociologist Robert Merton pointed out in the 1940s, scientists depend on the reliability of each other's work.
Further, science is not unitary, and criteria for responsible research vary widely across fields. As recent U.S. experience shows, centralized agencies, such as our Office of Management and Budget, are more readily captured by bias and ideology than the decentralized institutions of science, and hence can lose their independence.
In sum, well-designed, open and responsible review processes tailored to specific institutional contexts can do more to ensure—
