Building and Breaking Bubbles: Designing Technological Systems that Select and Present News

Communication Shaped by Algorithms: Technological, Economic, Social, and Ethical Challenges (panel)
14 November 2014,
ECREA 2014, Lisbon, Portugal

Of all domains, scholars and the public express most concern about algorithmic selection and personalization affecting people's access to news (Sunstein 2001; Pariser 2011). If social network sites, personalization, and greater choice in news venues lead people to read information that confirms their own opinions and covers only topics they are already aware of, this could have adverse consequences.

This portion of the panel discusses how design choices for the technologies people use to access news can lead to different outcomes for individuals and society. For example, selective exposure theory could inspire designers to build systems that give readers mostly agreeable information. Systems that select content based only on popularity in one's social network could, as a result of homophily, limit the topical and viewpoint breadth of readers' incidental news exposure.

While such systems might do well in getting people to click on articles and feel validated about their opinions and interests, they are likely not in the best longer-term interests of individuals or society. Items that cause the most clicks or shares may not be the ones that best support all of an individual's motivations for reading the news. Societally, exposure to reinforcing views could lead to echo chambers and fragmentation.

Different design choices could lead to systems that are at least as engaging as current tools but better support individual and societal goals. Such systems will balance safe and riskier recommendations and will recommend content with both short-term appeal and long-term value. Their interfaces will make filtering transparent. They will measure and reveal content's value.

In addition to discussing design choices for news aggregators and content feeds, I will discuss how personal informatics tools can help people understand their behavior across the sources they access. I build on results from an ongoing research project, Balancer, which gives users feedback on the political lean of their reading (ICWSM 2013), as well as others' inspiring designs, including Slimformation, which reveals topical diversity in one's online news-reading, and Scoopinion, which reveals a reader's top authors, sources and genres.


The Balance Project

Aggregators such as Digg, Reddit, and Google News rely on ratings and links to select and present subsets of the large quantity of news and opinion items generated each day. This work requires understanding people's preferences for diversity (CHI 2010), as well as developing methods of selecting diverse sets of items (ICWSM 2009), and presentation techniques to make these sets appealing (CHI 2010) or more salient (ICWSM 2013).

Thanks to
  • Paul Resnick
  • Stephanie Lee
  • Daniel Xiaodan Zhou
  • Jeremy Canfield
  • Erica Willar
  • Emily Rosengren
  • Cat Hong Le
  • Brian Ford
  • Peter Andrews
  • NSF award #IIS-0916099
  • Yahoo! Key Technical Challenges grant
  • Intel PhD Fellowship