Recent research is casting doubt on previous findings that Facebook’s news feed algorithm effectively curbed election misinformation. A new study suggests that earlier research may have overlooked significant, temporary changes to the platform’s algorithm, potentially skewing the results and creating a misleading impression of its standard effectiveness.
Debunking Previous Findings
An interdisciplinary team from the University of Massachusetts Amherst has challenged a widely reported 2023 study published in Science. The earlier research, which was partially funded by Meta, concluded that Facebook and Instagram’s algorithms were not major drivers of misinformation and successfully filtered out untrustworthy news during the 2020 US presidential election. However, the UMass Amherst team’s analysis indicates that the original study was conducted during a brief period when Meta implemented a more rigorous, temporary algorithm designed to reduce the visibility of misinformation. This algorithmic shift, which was not fully accounted for in the previous study, reportedly cut user exposure to misinformation by at least 24%. The researchers noted that this more stringent algorithm was reverted in March 2021, returning the platform to its previous practices.
- A key criticism is that the earlier study did not clarify its data captured an exceptional, short-term algorithmic change.
- This oversight may have created a false impression that Facebook’s standard algorithm is effective at combating misinformation.
Algorithmic Influence and Political Bubbles
Further research has explored how Facebook’s algorithms shape users’ experiences and political views. Studies suggest that while algorithms significantly alter the content users see, their impact on political attitudes and offline behaviours, such as voter turnout, may be less pronounced than previously thought. However, these studies also indicate that liberals and conservatives tend to inhabit distinct political news ‘bubbles’ on the platform, with limited overlap in the news they consume. Changes to the algorithm can alter the content users are exposed to, including increasing or decreasing the proportion of political content and content from sources flagged as untrustworthy.
The Role of Meta's Funding and Transparency
Concerns have also been raised regarding the influence of Meta’s funding on research and the transparency of algorithmic changes. The UMass Amherst team highlighted that social media companies can modify their algorithms without public notification, potentially allowing them to improve their public image when aware of being studied. While Meta provided funding and co-authors for the earlier study, they reportedly did not have pre-publication approval rights. This situation underscores a broader debate about whether social media companies prioritise mitigating the harms of misinformation in their business decisions.
Shifting Content Trends
Recent observations suggest a shift in the type of news content gaining traction on Facebook. Right-wing news outlets, including Fox News and Breitbart, have reportedly seen increased engagement in recent months. This trend is partly attributed to the platform’s algorithm favouring content that resonates with its user base, which may be increasingly conservative. However, the most popular stories from these outlets often focus on less overtly political topics, such as celebrity news or personal health updates, which historically perform well on the platform.
Unanswered Questions
Despite extensive research, definitive answers regarding the extent to which social media algorithms drive political polarization remain elusive. The short duration of many studies and the complexity of external factors influencing user behaviour mean that attributing societal trends solely to algorithmic design is challenging. Experts caution against drawing sweeping conclusions, emphasising the need for continued, multifaceted research to fully understand the intricate relationship between social media platforms, user behaviour, and democratic processes.
Sources
- Team debunks research showing Facebook’s news-feed algorithm curbs election misinformation, Phys.org.
- Facebook’s news feed takes a right turn, Sherwood News.
- Researchers question claim about Facebook news algorithm curbing election misinformation, Courthouse News.
- 📱 Facebook and Instagram algorithms affect user behavior but not their political views, warpnews.org.
- New study shows just how Facebook’s algorithm shapes politics : NPR, NPR.
