Algorithms: Shaping News, Dividing Views
Algorithmic bias in media feeds shapes what news users see. This often narrows perspectives, creating filter bubbles and risking societal division.
Topics: media, technology, politics
Example
In September 2021, a report from NYU Stern's Center for Business and Human Rights concluded that Facebook, Twitter and YouTube are not the original or main cause of rising US political polarisation, but that their engagement-driven feeds intensify it. In 2023, studies run with Meta and published in Science and Nature found that switching some US Facebook and Instagram users to a simple time-ordered feed for three months did not measurably reduce their polarisation. The evidence on algorithmic 'filter bubbles' is real but mixed.
Key facts
- A September 2021 NYU Stern report concluded that major platforms are not the original or main cause of US polarisation but intensify it.
Limitations
- Both studies are about the United States, and the 2023 experiments ran for only three months around one election, so they cannot rule out slower or wider effects elsewhere.
Evaluations
media evaluation
Support
The NYU report finds that feeds ranked by engagement push content that provokes outrage, so the news people meet on social media is skewed towards the divisive.
Counterargument
The 2023 Meta experiments found that removing the ranking algorithm for three months did not change users' political attitudes, which weakens the simple 'algorithm causes division' story.
Rebuttal
The two findings fit together: platforms can intensify division at the margins while deeper causes, such as partisan cable news and local politics, drive most of it.
Additional support
The chronological feed in the experiments showed users more untrustworthy content, a reminder that removing algorithms is not automatically better curation.
technology evaluation
Support
The case shows how a design choice, ranking by likes and shares, can have social effects its engineers never set out to produce.
Counterargument
Users also choose whom to follow and what to click, so blaming the code alone ignores human preference for agreeable views.
Rebuttal
Design still matters because it sets defaults that billions of people accept; the question is how much weight to give it, not whether it counts.
Additional support
Meta's agreement to let outside researchers test its feeds shows that independent access to platform data is possible, and is needed to settle such claims.
politics evaluation
Support
The NYU report links polarisation to falling trust in elections and political violence, showing why governments take feed design seriously.
Counterargument
Polarisation in the US began rising before social media, according to the NYU report itself, so regulating platforms would not by itself reverse it.
Rebuttal
Regulation can still reduce harm at the edges, for example by requiring transparency about ranking, without claiming to fix every cause.
Additional support
The EU's Digital Services Act now requires large platforms to offer a feed option not based on profiling, one political response to these concerns.
Sources
- NYU Stern Center for Business and Human Rights (2021-09-13): Fueling the Fire: How Social Media Intensifies U.S. Political Polarization
- NBC News (2023-07-27): Facebook opened its doors to researchers. What they found paints a complicated picture of social media and polarization
More GP guides for this example
Topic guides
- Society and Culture GP Examples
- Politics and Governance GP Examples
- Science and Technology GP Examples