How Science Gets Drawn Into Global Conspiracy Narratives

A few short years ago, mRNA (messenger ribonucleic acid) was the subject of fundamental research, but it is now known as the basis for COVID-19 vaccines. At the same time, the concept has become linked—particularly on social media—to global conspiracy theories attributing nefarious motives to people associated with science. How did this happen? In our work, we use social media data to track evolving narratives empirically. By analyzing the terms that have become associated with mRNA on Twitter since early 2020, we have gained insight into how seemingly innocuous scientific concepts acquire sinister connotations through association. Understanding how this process occurs can be helpful in determining which countermeasures might be effective. Hashtags are key to this analysis. Used to cross-link social media posts, hashtags generally consist of a word preceded by a pound symbol: #mRNA, for example. Hashtags make such concepts easier to find because they can be easily searched. By observing how hashtags co-occur over time, we learn how ideas are linked to each other on social media. This approach is useful for understanding the ways in which disparate concepts become related to evolving narratives. To find out how the term mRNA became connected to farflung conspiracy theories, we collected a sample of 87,000 tweets containing the hashtag #mRNA over the three-year period from early 2020 to the end of 2022. This allowed us to look at how mRNA was juxtaposed with other ideas on social media over that time. Our analysis looks at time continuously, but we’ve found it helpful to take “slices” from the dataset to highlight the way the narrative took shape and then shifted over time. We looked at where #mRNA occurred next to other hashtags, which gives a sense of how the term became connected to other ideas. We presented this data visually, displaying the connections as a network where each node represents a different hashtag and each edge represents the number of times two hashtags co-occur in our dataset. Starting with tweets using the hashtag #mRNA, this method allows us to see how sometimes unexpected semantic networks of associations can develop around ideas. Although co-occurring hashtags should not be taken as representing general discussions about mRNA, their changing patterns over time may offer insights into how issues may be hijacked and misinformation spread. In our first sample, from early 2020, the hashtags cooccurring with #mRNA were largely scientific or financial, reflecting prepandemic views. To make this network graph more readable, we “cleaned” the data, systematically removing all hashtag nodes above and below certain thresholds determined by number of connections. The distance between nodes indicates how often terms are used together in posts, and the size of the text reflects the sum of its connections. Thus, larger text shows terms that are highly interconnected. The color coding reflects which communities are involved, which is discovered through an automated process that examines connections. The first figure depicts a mostly pale green community made up of hashtags corresponding to scientific terms as well as a gray colored community devoted to discussing biotech investment. These figures make it possible to examine how the narrative around mRNA evolved over time. As vaccines went into production, new semantic networks associated with the term quickly began to develop. One cannot simply assume this method represents all the opinions that are “out there” in society, but it can nevertheless be quite helpful in MARC TUTERS, TOM WILLAERT, AND TRISHA MEYER REAL NUMBERS