The Problem

Vigorous skepticism and contestation are critical to science. Much like the values essential to free speech, science works only if critics have opportunities to raise disagreements and dissent.

Yet the U.S.’s design of the expert bureaucracy organizes government science in exactly the opposite way. Rather than aligning government science with professional norms, the production of that science is subject to centralized control in which political officials can shape the agencies’ scientific agendas, methods, interpretations, and even rely on cherry-picked peer reviewers. All of this can occur, moreover, with little transparency or opportunities for staff to voice disagreements.

Specifically, political officials within agencies and the White House regularly influence the staff’s scientific work while simultaneously insulating many of those interventions from public disclosure through confidentiality protections such as the deliberative process privilege. Agency peer review and scientific integrity programs remain ad hoc, opaque, and subject to substantial political control. Nor are these centralized features of government science solely the product of presidential control through the unitary executive. Congress and the courts generally (although not always) reinforce this top-down executive control on agency science in their ad hoc approaches to design and oversight, which rarely benchmark their reviews against scientific norms.

One might wonder how things got this bad. After all, ever since Vannevar Bush’s The Endless Frontier, the United States has treated government support for science as essential to public protection, national security, and economic growth. But in hindsight, it turns out that this conceptualization of the endless frontier lacked a structural blueprint for how to design government to ensure that the research was reliable. As a result, and particularly over time, both the government’s data collection and scientific analyses have become not only fragmented, but increasingly vulnerable to top-down control by the political branches. This centralized structure for the provision of government science, in turn, undermines the integrity of the science itself, as well as the trustworthiness of the resulting policies that rely on it.

The second Trump administration made these structural vulnerabilities noticeably worse. Unprecedented terminations of staff scientists, removal of public datasets, grant cancellations, the installation of industry-aligned political leadership across major public health and environmental agencies, and elevating political control over the awards for scientific grants are among the earmarks of his second term.

But although President Trump’s actions are more dramatic than past presidencies, this problematic centralized political control over government science long predates his administration, extending at least as far back as the Nixon administration. Over the last four decades, journalists, courts, inspectors general, and nonprofits like the Union of Concerned Scientists have documented political interference in agency science and the suppression or marginalization of internal dissent. The problems we are witnessing with Trump, then, are not simply partisan; they are hardwired into the structural design of the expert bureaucracy.

A Dissent-Based Intervention

To begin to mitigate this centralized control over government science, government scientific work must be subject to more rigorous and transparent systems of scientific scrutiny and dissent. The core goal is not to insulate science from politics altogether—likely impossible—but instead to institutionalize opportunities for constructive critique and dissent, including by staff scientists, in the government’s provision of science.

To accomplish this at a general level, agencies should be required to subject significant scientific analyses and related science-intensive decisions to independent and disinterested peer review administered through an institution or agency that is insulated from direct political control. Ideally, this expert scrutiny should extend not only to the review of key features such as the agencies’ methodological choices, data collection processes, the characterizations of uncertainty, or operating assumptions. It should also include scrutiny of how the expert bureaucracy itself is managed. Candidate agencies for conducting this review must have maximum autonomy from political control, such as the National Academies, the congressional bureaucracy such as the Government Accountability Office or the Congressional Research Service, or similarly designed independent review bodies. (We tried to come up with some more specific ideas on this independent oversight organization at pages 766-69 of this unrelated article.) Of course, there is no institutional arrangement that can fully eliminate politics or bias. But dispersing authority to external, respected organizations and creating visible opportunities for meaningful scientific disagreement is a solid first step to limit the risks associated with concentrated political control.

Equally important, agency scientists and staff should be permitted to place their own disagreements and competing scientific interpretations into the administrative record before this external review occurs. Reviewers can then evaluate both the agency’s analysis and the competing critiques. Designing a staff process in this way should protect opportunities for scientific dissent without simply transferring unchecked authority to individual staff scientists to challenge agency findings without undergoing their own independent peer oversight. Enabling staff to publicly disagree will also allow those outside the agencies to trace the internal scientific debates leading up to a final decision. (Note: Internal agency protections will also be essential to prevent political officials from bullying or threatening staff that take advantage of these dissent opportunities.)

Implementation

The proposed intervention could be institutionalized through any of the branches acting individually or simultaneously. For its part, Congress could require independent, disinterested review for major agency scientific analyses through authorizing or appropriations legislation. Agencies could adopt similar procedures internally through rulemaking or scientific integrity policies. Presidents could institutionalize aspects of the process through executive order or the Office of Information and Regulatory Affairs. And courts could treat evidence of disinterested critical peer review as relevant to the weight afforded agency scientific judgments (currently the courts rarely do this, despite lip service otherwise).

The precise design details will require considerable elaboration and will likely include some prescriptive requirements. For example, the specification of reviewer selection, transparency requirements, conflict rules, disclosure obligations, and the framing of review questions would all require careful attention. Nevertheless, some of these design challenges have been addressed in other institutional settings, so there are some examples to follow.

And, at base, if I am correct that reliable government science requires vigorous skepticism institutionalized through processes that align with scientific norms—rather than being centralized through top-down political control—we may have no other choice. Until we infuse our expert bureaucracy with opportunities for dissent and open-minded disagreements, the provision of government science risks becoming yet another mechanism through which political officials can exercise authority without meaningful accountability.

Why It Matters

Although meaningful opportunities for independent peer review and opportunities for dissent are essential to combatting the political centralization of government science, peer review is hardly a magic bullet. Institutionalizing a more rigorous peer review and dissent process will bring its own share of headaches and disappointments. As scientists and journal editors know well, peer review is highly imperfect, susceptible to its own biases and risks of manipulation, and can sometimes dampen scientific creativity. The literature on the limitations of peer review is substantial.

Even with these limitations, however, several overarching benefits offset peer review’s shortcomings.

First, as noted, critical scrutiny is a central feature of scientific inquiry and is hence fundamental to ensure the legitimacy and reliability of the resulting science. Until government processes provide assurances of open-minded skepticism and opportunities for dissent, the resulting science is vulnerable to being compromised and may not appear trustworthy at all.

Second, curating critical review can be used to help distinguish scientific disagreement from political manipulation. Virtually all contemporary governmental uses of science involve a mixture of science and policy. Review charges should specifically require reviewers and agencies to identify where scientific judgment ends and policy values begin. These expert delineations will sharpen the inevitably fuzzy lines for not only the public, but also for those making the policy decisions.

Third, scientists are ethically trained in ways that demand they be able to openly contest methodological choices and scientific characterizations with which they disagree, as well as to solicit independent peer review more generally. If deprived of these fundamental features of their profession, talented scientists might choose not to work in government or else will serve, only to find their contributions undermined by processes that conflict with their professional values.

Finally, the intervention could help rebuild public trust in government science. Public confidence depends not only on whether scientific conclusions are ultimately correct, but also on whether the processes generating those conclusions appear trustworthy and resistant to political manipulation. More than four decades of documented political interference have—according to several public opinion studies—undermined the public’s faith not only in government science but science more generally. Instituting government processes that align with scientific norms and professional standards may not fix the trust problem, but it should move public trust in a more positive direction.