New data shows that the polarization of political discourse online has remained largely unchanged since the end of 2020. That’s probably not surprising if you’ve looked at the internet at all in the past year. But the data also shows an underlying pattern in which individual topics—like abortion and immigration—took turns driving divisiveness. While people were consistently mad online about political issues, the issues that inflamed conversations shifted dramatically throughout the year.
The data, which comes from a joint project between Zignal Labs, a social media intelligence platform, and the University of Southern California, helps explain why political discourse may have seemed in 2021 like a never-ending carousel of outrage.
The data
Zignal and USC teamed up to create the Polarization Index, which measures engagement with polarized content on Twitter and calculates a polarization score. Since the index started tracking conversations last year, major political events like the January 6 insurrection, the transition from the Trump to the Biden administration, and the majority of the covid-19 vaccine rollout have occurred. All the while, the PI score has barely moved.
While Twitter is far from a perfect proxy for broader divisiveness, online platforms play a hugely important role in shaping political discourse. Social media platforms such as Meta (formerly Facebook) have again been under the microscope this year, leading to new doubts around these platforms’ ethics and how they can address misinformation, extremism, and hate speech online.
There is a long-standing academic debate about how to measure polarization, and a clear standard has not yet emerged. This index averages polarization scores for 10 political topics—immigration, policing, racial equity, abortion, voting integrity, gun legislation, climate change, minimum wage, covid-19 vaccines, and health-care reform—on a scale of 1 to 100 (100 being absolute polarization).The polarization score is calculated by combining the volume of shares of news links on Twitter with the bias and reliability ratings of the media sources publishing the shared content, assuming that a “low-reliable source on either end of the political bias spectrum is more polarizing than a share from highly-reliable, more centered sources.”
The grouping of media sources according to bias and reliability comes from the Ad Fontes Media Bias Chart, an independent news content rating company that determines political leaning and evaluates reliability on the basis of original fact reporting.
Why being online felt so bad this year
The Polarization Index started with a score of 85.5 at the end of 2020, which the researchers dubbed a “critical” level. The score dropped just 3 points at the beginning of 2021 and has remained consistent ever since.
Currently, immigration is the most polarized topic measured by the index, followed by policing policy, racial equity, and gun legislation. At a topic level, changes in polarization were much more common, and degrees of polarization seemed to shift from topic to topic, keeping the overall score high.
Voting integrity, for example, was the second most contentious issue in Q4 of 2020 and then dropped to sixth out of 10, rising back to fifth in the second half of 2021.
Research published alongside the Polarization Index also found that the news articles shared on the most polarized topics were more likely to come from unreliable, right-leaning sources. The report says that “engagement with right-leaning sourceswas more likely to push conversation into an increasingly polarized direction.”
As an example, this was the case for immigration, the most polarized topic: from the end of 2020 though the third quarter of 2021, right-leaning sources of mid and low reliability came to dominate the conversation, and the polarization score increased from 84.8 to 100.3 over the year. The pattern is consistent for the other highly polarized topics.
What’s to come
In line with the results of the research from Zignal, it’s well documented that more extreme content also tends to be more misleading.
Anya Schiffrin, director of the Technology, Media, and Communication program at Columbia University, says, “A lot of the disinformation is top-down. It’s coming from heads of state, it’s coming from politicians.” Schiffrin also pins the problem on a lack of “gatekeepers” to police the flow of content. Instead, algorithmic recommendation systems on social media platforms tend to amplify extreme material, which Schiffrin says leads to a more “extreme internet.”
The extreme digital environment led to dramatic displays of real-world violence this year. Examples of this relationship include Facebook’s role in violence following the coup in Myanmar and the January 6 insurrection in the United States, which was a result of a flurry of disinformation about the election results.
At MIT Technology Review’s request, Zignal conducted an analysis that looked specifically at how people engaged with different media sources over time on the issue of election confidence and voter integrity. Data shows that engagement with less reliable sources on both the left and right was highest closest to the election and around the events of January 6.
At the end of 2020, engagement with less reliable right-leaning sources in particular dominated the online conversation about voter integrity. This was also the time that the voter integrity polarization score was at its highest, reaching 95. According to the report, the high level of discord driven by divisiveness over voter integrity “led to the January 6 events at the Capitol.”
Notably, highly reliable right-leaning sources account for just .017% of total engagement on the topic of voter integrity, whereas highly reliable left-leaning sources account for around 36%.
According to a Pew Research study at the end of November 2020, 79% of Trump voters said that the 2020 presidential election was not run well, compared with 6% of Biden voters.
Another election year is right around the corner, and conversations about the health of American democracy are again coming to the forefront, putting renewed pressure on social media.
Some reason for optimism, however, can be found across the Atlantic. The European Union is looking at two major bills in the early half of 2022, called the Digital Services Act and the Digital Markets Act, led by the French government. The bills seek to clamp down on hate speech and on the underlying advertising model, generally considered to be one of the most fundamental challenges in stopping the spread of misinformation.