登录
Story from the Research Trenches: Bonnie Varlet on Transforming Research Workflows with Zotero
As part of our blog series, “Stories from the Research Trenches,” we often invite researchers and colleagues to share their personal experiences. For this post, we are delighted to hand the floor to Bonnie Varlet from the KU Leuven Cultural Studies Research Group, who offers a closer look at how she integrates Zotero into her research workflow.
Written by Bonnie Varlet
New technology, like machine learning systems, are being deployed across a wide range of institutional backgrounds. Hospitals use it to flag diagnoses, and archives use it to catalog their collections. Modern machine learning has made great leaps in its capabilities. However, these tools do not exist in a vacuum.
My research looks at what governs the relation between institutions and the technology they use. For example, when a machine learning system is introduced into a new operational environment, it changes workflows, changes what skills are required, and creates new dependencies that did not previously exist. Validation processes that were designed for human-scale output volumes become inadequate when a system can produce ten times more at the same time. Accountability structures built around individual judgment become harder to maintain when outputs are generated algorithmically. At the same time, organizations also change technology. Institutional priorities shape which systems get acquired and how they are used, and informal workarounds created by staff can become de facto operating procedures. The system produced by this process is often meaningfully different from the system initially deployed.
These processes do not happen independently. They influence each other; it is iterative, and it compounds. There is currently no widely adopted methodology for tracking this relationship in a way that is observable in comparable and replicable terms. Most existing research either examines technology deployment in isolation or analyzes governance structures without tracing their operational consequences. This remains largely unmapped, which makes it difficult for organizations, regulators, and researchers to fully understand how technology and institutions interact in practice. My work aims to help develop a way of systematically observing these interactions as they take shape in real operational contexts.
Tackling a project like this, especially my first one done independently during my Fulbright, was also a lesson in how small logistical problems can get you off course. Over the course of the project, I collected tens of papers, books, website links, and other sources. At the start, when it was only a couple of papers, it was manageable, but as the project matured, it became increasingly difficult to stay organized. This became particularly challenging given the breadth of the topic, which required me to move between technical material, governance literature, and case-based examples.
At that point, I was lucky to have resources available through KU Leuven, such as the Artes Research team, where I was introduced to tools that could help manage my workflow. I decided to try Zotero, which was easy to set up and start using immediately. What changed right away was that I stopped getting lost in my sea of sources. All my papers, books, and links were kept in one place dedicated to my project, and I did not have to go back and look up publication details because the browser extension stored that information when I saved a new source.
As I explored Zotero further, I also shifted how I organized my work. Because it makes it easy to tag and sort sources, I grouped them in the manner in which I used them. Sources used for case studies were given a case study tag, while sources that provided foundational knowledge were grouped separately. Since my project is broken up into different parts, this made it easier to see where I was pulling information from and how it influenced my analysis. In a project that tries to trace relationships between governance decisions and technical systems, being able to clearly track how different types of sources contributed to different parts of the argument was particularly useful.
I also began annotating and brainstorming directly within the same program, instead of splitting my workflow across multiple tools. For example, I would highlight a quote in a source, save it, and add a note explaining how or why it was useful. This made it easier to trace my thought process and how I arrived at certain conclusions, both for myself and in the final project.
Looking back, I wish I had reached out for help managing my workflow sooner. Not only with this specific software, but also with the more general question of how to structure a research project of this scope. I spent a significant amount of time at the start recreating approaches to organizing my work, rather than focusing on the project itself.
I would not treat exploring research management tools as a last resort. No matter the field, the people around you have likely encountered similar challenges and found better ways to address them. Trying something new partway through a project should not feel like a disruption. In my case, it was what allowed the second half of the project to go substantially better than the first, and made it easier to carry out research that depends on systematically tracing complex relationships between institutions and technology. It also made it possible to more clearly trace how different sources, ideas, and cases connect—something that is central to my research itself, which focuses on understanding how relationships between institutions and technology take shape over time.
