Observers and practitioners of artificial intelligence (AI) have conjectured the possibility of catastrophic risks associated with its emergence and development, risks that have led some to propose an FDA-style licensing regime for AI. In this essay I explore the applicability of approval regulation – that is, a model of product intro- duction that combines experimental minima with government licensure conditioned partially or fully upon that experimentation – to the regulation of frontier AI. There are a number of reasons to believe that approval regulation, simplistically applied, would be inapposite for frontier AI risks. Domains of weak fit include the difficulty of defining the regulated “product,” the presence of Knightian uncertainty or deep ambiguity about harms from AI, the potentially transmissible nature of risks, and the potential for massively distributed production of foundation models with min- imal observability of production. I consider four themes for future theoretical and empirical research: (1) the proper mix of approval regulation and other models such as liability or intellectual property regimes; (2) the possibility that deep ambigu- ity or Knightian uncertainty may require a kind of speculative pathology in which conjecturing scenarios is at least as important as placing probabilities upon them, in part because of the Lucretius problem; (3) the likely structure of industry and foundation-model generation, as the feasibility of approval regulation is higher with fewer producers, and much of the future of AI regulation may consist in labs and models monitoring one another; and (4) the possibility of community option value in the incremental development of AI regulation (including approval regulation), as regulatory policies may be more reversible in AI than in other settings, experimen- tation generates important public goods, and regulatory learning by doing is likely to be a property of any portfolio of policies in this arena.

Administrative agencies have undertaken an increasingly substantial role in policymaking. Yet the influence-seeking that targets these agencies remains poorly understood. Reporting exceptions under the Lobbying Disclosure Act allow many of the most powerful advocates to characterize their activity as lawyering, not lobbying, and thereby fly under the radar. Using agency-generated records on lobbying activity, financial reporting, and personnel databases specific to lawyers, as well as LinkedIn, we describe a vast subterranean world of regulatory influence-seeking that the social-science literature has (mostly) ignored. Regulatory lobbying is systematically different from legislative lobbying. It involves different kinds of people and different lobbying firms that bring specific forms of expertise and distinct networks. Our key findings about how regulatory lobbying differs include the following: (1) the regulatory lobbying sector is highly segregated from the reported lobbying sector, with many regulatory advocates failing to consistently register or report earnings commensurate with their activity level, (2) the number of unregistered regulatory advocates working on the implementation of the Dodd-Frank Wall Street Reform Act plausibly exceeds 150% of the registered lobbyists working on that law, (3) the most effective regulatory lobbyists and law firms involved with regulatory lobbying have incomes that dramatically outpace leading reported lobbying firms (which are also mostly law firms), and (4) back-of-the-envelope calculations and more sophisticated decomposition regressions imply that aggregate expenditure on lawyer-lobbying is several multiples of reported lobbying spending. We introduce the case of a particular lawyer-lobbyist and provide a theoretical discussion to situate and contextualize these findings. Collectively, this work opens a window into neglected domains of politics and reveals an important and understudied form of political inequality.

The Covid-19 pandemic highlighted the need to examine public trust in the U.S. Food and Drug Administration (FDA) vaccine approval process and the role of political influence in the FDA’s decisions. Ensuring that the FDA is itself trustworthy is important for justifying public trust in its actions, like vaccine approvals, thereby promoting public health. We propose five conditions of trustworthiness that the FDA should meet when it reviews vaccines, even during emergencies: consistency with rules, proper expert or political decision-makers, proper decision-making and noninterference, connection to public preference, and transparency of both reasons and procedures. The five conditions provide a roadmap of procedural and substantive requirements, which the FDA has variably implemented, focused on ensuring appropriate influence of political interests. While being a trustworthy agency cannot guarantee the public’s trust, implementing these conditions build a groundwork for public trust.

How can historical perspective be brought to the quantitative social sciences? The question has proven immensely popular, but answers that deal squarely with historical context and narrative remain elusive. An important recent book by Gregory Wawro and Ira Katznelson—Time Counts—provides an important new direction and a kit of useable tools. Using Wawro and Katznelson's approach and methods puts social scientists in a better position to appreciate the historicity of their data and to avoid common errors in statistical execution and inference. Time Counts also raises questions every bit as vital as those it answers, especially when it comes to the boundaries between narrative and quantitative work. An important concern is that inference from a particular historical setting (or what I call a “regime”) cannot be reduced to a special case of inference from large-sample statistics. Historical judgment is at least partially incommensurable with the idea of probability, cases are often important precisely because they are not countable, and scientific rigor may demand avoiding quantification for part of the social scientist's approach.

The study of agenda setting has become curiously disconnected from democratic theory and democratization. Following Schattschneider, Dahl, and recent developments in political theory, I call for its reintegration in theoretical and empirical realms. The concept of agenda democracy allows for better understanding of contests over institutions, significant historical-institutional transformations, the study of inequality and its mechanisms of generation and maintenance, and the building and undermining of democracy. Agenda democracy requires a broad understanding of agendas (beyond a mere menu of final policy choices), recognizes that many democratic regimes have institutions that systematically render agendas nondemocratic, and compels us to look at the interstices of institutions and society (party transformation, petition and grievance mechanisms, advocacy campaigns, initiatives to expand what I call the shortlist of the possible) for moments of significant change. Agenda democracy compels the examination of democratizing agenda restrictions, the study of conservative organizations in politics, and the consideration of decomposing the term “movement.”