News content

Customizable political news content is a danger to democracy, study finds

BUFFALO — It’s no secret that political news junkies typically frequent websites that tend to support the party they associate with and personalize their social media feeds so that the headlines they see come from these media. Yet a recent study claims that sites that allow you to choose the content you read based on your inclination are actually hurt American democracy.

Ivan Dylko, an assistant professor of communication at the University at Buffalo and an expert on the effects of communication technologies on politics, has studied the political effects of personalization – a now ubiquitous technology that personalizes the topic of content on a news site. given.

Twitter, Facebook, and Google News all use this technology to sort news and opinion pieces to provide information they deem useful to their users.

(Photo by on Unsplash)

The problem, according to Dylko and his team, is that people rely on these personalization platforms for all their news and information. This creates a space where informed and stimulating debate is increasingly difficult.

The study found that most political websites that automatically personalize content for each user or allow the user to create personalized content profiles increase the tendency of users to consume only information that matches their political ideologies. . The researchers found that this effect was particularly acute among people with moderate political ideologies.

“The increasingly popular personalization tools are likely to lead to a situation where we are surrounded by like-minded information that creates a biased perception of reality, incorrect beliefs, extreme attitudes and under-represented political behavior. optimal,” Dylko explained in a university press release.

Dylko enlisted Igor Dolgov, an associate professor of psychology at New Mexico State University, and his team of graduate and postgraduate students for the study. They gave a select group of study participants a political survey to determine their individual political leanings. Topics were then randomly assigned to one of four different political websites with liberal and conservative content.

Each site had a different method of aggregating and curating personalized content: a user-customizable site, a system-customizable site where researchers manipulated content based on survey responses, a hybrid of first two customizable sites and one non-customizable site. . The researchers recorded the links clicked and the time spent reading each article.

“We found that the presence of personalization technology increased the consumption of pro-attitude information and decreased the consumption of counter-attitude information,” says Dylko.

He added that this phenomenon is known to increase the polarization of political thought. Essentially, the use of customizable news and opinion sites creates bubbles of information that are difficult to disturb and challenge.

“It’s not good for a healthy democracy,” he said. “Living in ideological cocoons prevents the cross-fertilization of political ideas, undermines civil political discourse, and impairs the quality of decision-making in the political context.”

As for what could be done to help alleviate the problem, Dylko says search engines and social media sites should consider ensuring visitors are exposed to news from all corners of the political arena.

“We hope that the decision makers behind websites like Google, Facebook, Twitter and other key gatekeepers of political information will take note of the unintended harm their services could inflict on our society and try to mitigate that harm technologically,” he said.

Of course, information consumers also need to consider how websites can present slanted content and be open to other mediums, adds Dylko.

“We should all be more mindful of how news algorithms might inadvertently affect us negatively,” he continues, “and try to get out of the cozy news bubbles that each of us has created on various news platforms. online news and social media”.

Dylko and his team published their study in August in the journal Computers in human behavior.