Decoding Disinformation: Agent-Based Modeling with David Butts

David Butts sitting in front of two computer screens with text that reads "ICER HIGHLIGHT FOLLOW-UP DAVID BUTTS"

Chronic wasting disease in deer and disinformation on social media have something in common: they spread. David Butts has explored both research areas in graduate school by creating computer simulations to model their spreading properties. He used the models to test mitigation strategies and propose ways to stop their progression.

Butts is a graduate student in Computational Mathematics, Science, and Engineering (CMSE) at Michigan State University (MSU). The Institute for Cyber-Enabled Research (ICER) highlighted his work five years ago when he was just getting started. Now, as he prepares for graduation, he chatted with us again about his latest research, the arc of his time in graduate school, and his life outside of school.

Butts can tackle many types of research questions because his expertise is in the broad field of agent-based modeling, which is applicable across various areas of study. Agent-based models consist of three components: agents, the environment, and rules. The agents are the main entities being modeled, and could refer to people spreading disinformation on social media, deer spreading chronic wasting disease in a forest, or numerous other elements based on what is being modeled. The environment is the simulated world the agents exist in. For example, the graph of a social network, or maps encoding the distributions of trees and water. The models are created with computers and based on rulesets that define how agents interact with other agents and the environment.

Unlike typical analytical models, which work from the top down and often make sweeping generalizations about large populations, agent-based models work from the bottom up and account for individual differences and specific interactions at a high level of detail. Although they are less time-efficient to generate, agent-based models can provide more precise results by allowing researchers to recreate specific environments.  

At the time of writing this article, Butts’ most recent work used agent-based modeling to compare strategies for combating the spread of disinformation. This analysis could be used by policymakers to inform new practices aimed at the same goal. In his research, he compared three policies.

The first policy involved content moderation in which people were removed from the network randomly or based on their influence. The second policy used education aimed at increasing an individual’s skepticism and attention. The third policy was counter-campaigns in which a new group of individuals were added to the network who were committed to the accurate information rather than the disinformation.

The results of Butts’ research showed that the usefulness of a given policy could be context-dependent; however, overall, education-based policies that increased skepticism and counter-campaign policies were the most promising. He provided real-world examples of how these policies can be put into practice. Increasing skepticism could stem from media literacy programs aimed at helping individuals identify trustworthy sources and differentiate fact from opinion. Counter-campaigning could entail launching corrective advertising to directly address false narratives.

Crunching the numbers and running the simulations required significant computing resources. To run his experiments on a regular laptop computer, Butts said it would have taken around two years if it was possible at all. Using ICER’s supercomputer, he completed the computational component of his work in one week.

On a personal level, Butts contemplates the possibility of venturing into national laboratories as the next phase of his career journey following the completion of his Ph.D. But as he looks toward the future, he also takes time to reflect on the past. Thinking about the version of himself interviewed by ICER five years ago in the first research highlight, he feels lucky that his time as a graduate student has been positive and without any big regrets.

In hindsight, he would offer a piece of advice to his former self: to immerse more deeply in the unique and vibrant environment that MSU has to offer. In particular, he believes he could have heightened his university experience by capitalizing on opportunities to join student clubs aligned with his passions—such as running, kayaking, and inline skating. He also recognizes untapped opportunities within MSU for leveraging resources dedicated to professional development and research support. This retrospective insight serves as a reminder for him to encourage current students to seize the full spectrum of opportunities available within the university community.