This paper discusses the expansion of refereeing at journals in the US after the Second World War, and argues that the impetus for the expansion of refereeing in the US was the dramatic increase of federal government funding for science. The increase in funds available for scientific research led directly to an explosion in the size of US scientific journals and an increasing deluge of submissions to those expanding journals. There was little sense at the mid-century that journal refereeing might improve the quality of the selected papers or bestow respectability on an organization that used it; choosing to use referees was a practical decision aimed at making journal workloads more manageable for the editorial staff. However, embracing refereeing led to its own set of editorial challenges. The length of time papers spent in the refereeing phase of publication became a frequent cause for concern among editors, publishers, and contributors, and scientific authors began to express frustration with papers that received negative referee reports on “trivial” or unfair grounds. The expansion of refereeing led to an expansion of criticisms about it.

AI-assisted peer review is increasingly being discussed as a possible solution to bias in the review of scientific manuscripts (Bauchner & Rivara 2024). Proponents of this idea build on the criticism that has been voiced since the 1940s that peer review lacks objectivity and consistency (Goodrich 1945). AI-assisted peer review ought to help remove subjectivity from the review process. My talk challenges the idea that AI-supported peer review is a superior alternative to conventional peer review. To do this, I examine the origins of the claim that peer review should not be subjective. I show that this assumption does not reflect the intentions of historical peer review and that the contemporary process is not intended to be objective. Instead, I argue that peer review requires subjectivity, and that AI-based peer review does not contribute to making the process more consistent. Lee (2012) already notes that reviewers interpret criteria for publication differently, consistent with Kuhn's idea that values in science are interpretable. By introducing the concept of “professional subjectivity”, I take up this idea and develop it further. “Professional subjectivity” refers to the phenomenon that a person's socialization in a professional context result in a combination of (theoretical and operational) expertise, and individual preferences regarding the professional task. This taste includes personal preferences and biases, although it may be shaped by role models, such as former supervisors. I argue that the subjectivity inherent in peer review is a sign of normal science in the Kuhnian sense, and a desirable feature.

Ever since citations in academic papers began to be used as a means of studying science in the 1960s, various observers -- from scientific practitioners to sociologists, philosophers, and historians -- have attempted to produce a "theory of citation," trying to pin down what footnotes mean, and why authors cite other sources. This has not produced any consensus view, but following the contours of this history is an avenue for exploring shifting conceptions of a social epistemology of science in practice (indeed, one that has proved highly consequential). After briefly surveying certain key moments in this history, this talk will zero in on the late 1970s and early 1980s, when challenges to a general theory of citation gained ground through critiques originating from feminist sociologists and scholars studying science based in the global south.