US-Soviet Joint Fisheries Research and the Data Legacy of the Cold War / Adam Kriesberg (Simmons University) and Jacob Kowall (Simmons University)
In September 1959, Donald McKernan, the Director of the Bureau of Commercial Fisheries in the US Department of Commerce wrote to the Assistant Secretary of Fish and Wildlife in the Department of the Interior with concerns about the growing presence of Soviet fishing vessels in the Atlantic waters near the North American coastline. Throughout the 1950s, the USSR had increased its efforts to study and harvest fish in areas close to US territory. The federal government wanted to be sure that these activities were scientific in nature and did not include military espionage activities; they also wanted to ensure that Soviet fleets did not overfish in the North Atlantic and deplete fish stocks. While McKernan’s letter was skeptical of Soviet intentions, the US fisheries management community reached out and engaged with their Soviet counterparts directly. Quickly, the goal of preventing global overfishing became more important than investigating possible spying on Soviet fishing boats. These conversations eventually led to a number of collaborations between US and Soviet scientific agencies to better understand the fisheries of the world’s oceans, and to negotiate terms to allow each country to fish while not depleting stocks too much. The ultimate goal of these efforts was to have a common place to compare fishing techniques and catch data from each country that operated in the areas most susceptible to overfishing. The joint US-USSR research efforts beginning in the 1960s laid the foundation for continued cooperative studies through the 1980s and into the post-soviet era. In 1988, delegates of the US and USSR signed the “Agreement between the United States of America and the Union of Soviet Socialist Republics on Mutual Fisheries Relations,” which sought to provide mutual benefit to fisheries operations and research of both parties. Russian and US fisheries management personnel continue to meet and report catch statistics for comparative analysis. Some of the catch data which began to be shared in the 1960s are hosted on the website of the Northwest Atlantic Fisheries Organization (NAFO), a group established to serve as a neutral space to collect these statistics and discuss policies to regulate commercial fishing. The following research questions motivate this project: How did US and Soviet marine scientists work together to study fisheries? How did these researchers from different political systems collaborate to generate and describe data, while also navigating the transition to digital recordkeeping and data exchange? How has this data been preserved over time and how has it informed decisions on marine ecosystems? Can we understand the differences in US and Soviet scientific metadata and recordkeeping from looking at the datasets generated from these projects that are currently available? This paper uses historical records from the US National Archives and data products currently available on the web to examine the legacy of this Cold War cooperative research program. These materials demonstrate how American and Soviet scientists negotiated recordkeeping and data management activities across radically different governmental structures, and their current home at NAFO reveals ongoing risks to their long-term preservation.
Machine Learning and Archival Practice: A Cybernetics Case Study on Computational Approaches to Digital Materials / Bethany Anderson (University of Illinois at Urbana-Champaign)
For archivists, computational approaches to archives present potentially complementary and efficient ways to appraise and curate digital materials. Despite recent projects to implement and use computational tools and methods, questions remain about where computational archival work fits within archival theory and practice. For example, technologies like machine learning and natural language processing appear promising for enhancing archival access, but how can we ensure that machine-extracted data is applied to description in a way that aligns with archival standards? Likewise, providing access in a “collections as data” vein appears promising for meeting new and emerging research needs, but how do we facilitate such uses within current archival services and access systems? Above all, how do we ensure that these new modes of access do not decontextualize archival materials as they are (re)contextualized for new purposes? The University of Illinois Archives explored computational approaches as part of a project funded by the National Endowment for the Humanities—The Cybernetics Thought Collective: A History of Science and Technology Portal Project. Based at the University of Illinois Library, the Cybernetics Thought Collective (CTC) is a collaborative pilot project between the University of Illinois Archives, the American Philosophical Society, the British Library, and MIT to assess computational approaches for enhancing access to digitized archival materials. Over a two-year period (2017-2019), the CTC project machine-extracted data from 60,539 pages of digitized correspondence from four founding members of the field of cybernetics. This data was used to generate visualizations and to create metadata to not only facilitate access, but to also identify latent connections between materials. Experimenting with computational approaches was thus appealing for the ability to create structured metadata from unstructured data, and to facilitate scholarly analysis of machine-extracted data. The former can potentially enhance and augment archival description, while the latter is a step toward providing a more integrated user experience for archival research and digital scholarship. These methods were particularly apt for enhancing access to archival materials relating to cybernetics—and its transdisciplinary correspondence network that documents one of the most influential but least well-known scientific movements of the twentieth century. In addition to assessing computational approaches for machine-extracting metadata and for shedding light on scientific correspondence networks, this paper will discuss preliminary takeaways on implementing these methods for archives more broadly. The CTC project raises questions about the degree of archival intervention necessary (and needed) in computational workflows—and how feasible employing and implementing such approaches really are at this juncture given the realities of staffing and resources, and the absence of models that can be effectively deployed in an archival context. The project also presents an opportunity to contribute to ongoing dialogues between archives and computing, and how archives can meet emerging research needs for providing access to digital materials in an evolving digital scholarship framework. Ultimately, it underscores that computationally-driven archival projects—like any archival endeavor—must reflect core archival principles while exploring new territory.