Digital and Computational Studies Blog

Last Friday, Duke Professor and Bowdoin Alum Chris Bail came to campus to give a talk about his work with political polarization as part of the Duke Polarization Lab. Bail discussed his research on social media’s role in the increasing polarization in the United States. Bail noted that since 2012, the majority of Americans have disliked the opposing party more so than they like their own political party, which Bail suggests is alarming and potentially harmful to the functionality of a democratic system. Bail and other researchers at the Duke Polarization Lab aimed to curb this polarization online. An early idea that the group tested was trying to expose people to views online that were contradictory to their own to see if preventing online echo chambers might help to bridge the growing divide. The team had people participating in the study take initial surveys about their political views, then had participants follow Twitter bots that posted content that was progressively more contradictory to the participant’s. At the end of the study, the participants took the same survey to see how, if at all, their political views had changed. To the researchers’ surprise, participants consistently became even more polarized by the end of the study.

These results initially stumped the researchers, even after considering 109 different variables like age, gender, socioeconomic status, and many more. However, Bail theorized that seeing these opposing viewpoints may have stirred a dormant political ideology in the participants. Bail gave an example of a subject in a follow up study who was a 63-year-old woman in upstate New York. In interviews with the subject, Bail realized that despite regularly voting Democrat, she held a number of personal views that were more conservative. Bail also noted that regardless of her views, the subject was quite disinterested in politics in general. Given this, Bail thought that this subject might go against the pattern of further polarization after exposure to opposing viewpoints. However, even this subject became more polarized after following the Twitter bot, and even engaged in her first ever political debate online during the study. This led Bail to his hypothesis that in the case of subjects like this woman who was rather apolitical, the inclusion of opposing viewpoints in their feeds can inspire them to get involved and take firmer positions in their political ideas.

Though it might be concerning that breaking echo chambers was counterproductive in reversing polarization, Bail finished his talk with a discussion of some methods that have shown promise. For instance, Bail suggests that platforms with anonymity might be effective in having productive conversations with members of opposing parties. In a more recent study, subjects were asked to test a social media platform where users anonymously discuss divisive issues with another person whose political beliefs are opposite of their own. The results of this study were very promising, and exit surveys revealed the subjects from both sides of the political spectrum on average came away with less polarized views. Bail suggests that anonymity works to promote productive conversations because when identity is removed from the equation, users feel they can express their personal opinions without fear of how the views would affect their social standing among their peers. However, Bail acknowledges that anonymity is something of a double-edged sword, as some users might feel comfortable sharing especially problematic views and engage in hate speech. Along with anonymity, Bail pointed to other possible solutions, like reversing the incentive structures so that extreme posts – which generate a lot of interaction due to their controversial nature – are not as prevalent in users’ feeds. Nonetheless, Bail’s talk highlighted some of the serious real-world implications that social media can have, and Bail emphasized that there is still much more to learn. To find out more about Bail’s work and the work of the Duke Polarization Lab, check out their website, the free content available on their summer institute page, or read Bail’s recently published book Breaking the Social Media Prism.

Digital & Computational Studies is all about intersections. It could be between genres, musical phenomena, political opinions, art, travel, dance, anything! Technology’s potential to revolutionize the way that we interact with these fields is ever growing, especially with the arrival of artificial intelligence and the increasingly efficient development of code. This lends itself to a multitude of possibilities for intertwinement between technology and any one of these fields. After all, we use data everyday to make important life decisions. It’s becoming increasingly relevant to investigate the ways in which we insert technology into our daily lives, so as to more clearly grasp its effects and potential drawbacks. 

Late October, acclaimed mathematician Cathy O’Neil spoke at Bowdoin College on exactly this, presenting on the destructive power of algorithms. O’Neil received her PhD from Harvard University, is the founder of O’Neil Risk Consulting & Algorithmic Auditing, and recently testified for a Senate Committee Hearing calling for restrictions to prevent online extremism. Broadly speaking, O’Neil’s lecture focussed on the connections between human life and algorithms: health-wise, socially, psychologically, and beyond. 

For the most part, people trust mathematics and data. It’s numbers. How could it be wrong? O’Neil argued that the perception of algorithms as objective is fundamentally misguided and occurs as a result of the hidden nature of algorithms and programming. O’Neil’s analysis of AI marketing explained that algorithms are laden with the opinion of its creator. She articulated numerous examples of how these algorithms are pervasive, hidden, and deeply ingrained in institutionalized structures: from standardized tests that evaluate teachers to predictive algorithms that shape policing in America. Since algorithms are complicated, hidden, and largely inaccessible to those of us without programming experience, the danger is that algorithms are blindly trusted. Despite this, O’Neil encourages us all to be skeptical and question the programs that shape the world around us. 

Below, Math and Economics major Lucas Sheridan shares his takeaways from the talk.

By Doevy Estimphile & Lilo Bean

Bowdoin faculty voted to approve the promotion of DCS from an initiative to an academic department offering both a Coordinate Major and a Minor. This result is the culmination of ten years of discussions amongst the institution and DCS professors including: Eric Chown, Crystal Hall, Mohammad Irfan, and Fernando Nascimento.

DCS falls perfectly in line with the liberal arts education as it spans across multiple disciplines and employs new computational methodologies in order to prepare students change the world. They will be able to bring skills learned in DCS into their other areas of study, preparing them for deeper levels of discovery. They will be fluent in digital literacy spanning from coding to data analysis to design. Further, students will be able to critically examine the implications of technology in relation to society, and understand the ethical consequences of the technology which they develop in the classroom.