Ivan Dylko, University at Buffalo – Personalized News Feeds

Should we customize our news feeds?

Ivan Dylko, assistant professor in the department of communication at the University at Buffalo, explores whether curating is a good idea or part of the problem.

Dr. Dylko is interested in the nature and political effects of Internet-based information and communication technologies (ICTs). In his research, Dr. Dylko uses Mix-of-Attributes framework to study how ICTs affect various political outcomes, such as political participation, political knowledge, political selective exposure, and political information processing. Currently, Dr. Dylko is leading several projects:

• Examination of how customizability technology (also known as personalization or tailoring) increases exposure to information that supports individuals’ political attitudes and beliefs, and how such selective exposure leads to political attitude polarization.
• Identification and explication of politically important affordances of social and mobile media.
• Algorithm audit of top-trafficked websites with the goal of describing how much political ideology-based personalization is taking place on those websites.
• Development of a general theoretical model that explains how ICTs create individual-level political effects.

Dr. Dylko’s research is inherently multidisciplinary. He has collaborated with researchers from psychology and computer science, and published articles on such topics as media effects, communication technology, public opinion, and political communication in International Journal of Public Opinion Research, New Media and Society, Journal of Broadcasting and Electronic Media, Communication Theory, and others.

Personalized News Feeds

Customizability is a key element of the modern information age present on many top websites like Facebook, Google News and Twitter.

This tool lets visitors select what types of stories a site delivers to them. This user-driven technology also has a system-driven counterpart that relies on software code – operating unobtrusively and sometimes covertly – that in a political context, prioritizes stories that align with the ideological browsing patterns of individual users.

On the surface, this sounds like a useful feature.

If my interests differ from yours we’re not likely to be interested in the same stories.

But customizability technology actually deepens personal biases. Our study experimentally tested the political effects of customization and revealed the dark side of this technology.

We found that customizability increased consumption of information one agrees with and decreased consumption of information one disagrees with on various political topics. This effect was particularly strong among political moderates.

This is called “political selective exposure,” and it has a very strong potential to increase political polarization, as shown by research we and others have done.

Sites initially created these personalization tools to help users cope with information overload. Unfortunately, it turns out that these popular technologies can unintentionally hurt our democracy by surrounding users with like-minded information. This facilitates development of a skewed perception of reality, incorrect beliefs and extreme attitudes.

System-driven customizability, sometimes called a “filter bubble,” is particularly troubling because sites remove substantial content automatically without consulting or even notifying the users.

Users should be alert to how modern information algorithms might negatively affect them and try to break out of the bubbles created by various online news and social media platforms. Information industry giants should also be mindful of the unintentional harm their popular technological products cause, and try to minimize them.