Researchers examine hatred and malice on Twitter

What would Elon Musk think of these research results? (Image sources: bloomicon via Adobe Stock, max kukurudziak via Unsplash)

According to a recent study by computer scientists at Cornell University and University of California, Berkeley, hatred and malice on Twitter have increased since Elon Musk took over (Evening Standard via Cornell University). In summary, four conclusions can be drawn from the study.

However, the authors of the study had no internal access to the Twitter platform. Also, the study is titled Twitter’s Algorithm: Reinforcement of Anger, Hostility, and Affective Polarization no long-term study.

Partly because of this, such studies provide only limited insights – which nonetheless provide interesting insights.

What did the researchers find out?

Insight 1: More anger. The researchers divided emotionality into the four dimensions of anger, happiness, sadness and fear. During the study, the authors found that that the algorithm amplifies tweets that are emotionally charged – especially anger. The subjects of the study were more emotional after reading such tweets, but also especially angry.

Insight 2: Othering. Another finding from the study is that when confronted with political tweets, the behavior of the subjects often changed. In particular, the consumption of political tweets led to a psychological process that the scientists called Othering describe. Othering causes people to develop negative ideas about groups with opinions that differ from their own.

The study says about the Twitter algorithm behind it: Political tweets are characterized by A) partisanship and B) hostility towards groups that think differently.

In addition, political content that is foreign to the group is more likely to be displayed. In other words: a left-wing person is shown content from the right-wing spectrum a little more frequently than left-wing content. So you can’t accuse the algorithm of creating a filter bubble or echo chamber for certain opinions – rather, opposites and animosity are strengthened.

Insight 3: Affective polarization. Exposed to the content played out by the algorithm, this leads to the fact that the test persons they perceive their own political group more positively, while other political groups rather negatively. According to the authors of the study, this dynamic reinforces an affective polarization of the subjects.

In other words, the test subjects felt they belonged more to their own group and were averse to other groups. This feeling was not based on a critical examination of a party program, but arose from a gut feeling.

Insight 4: More of the same. The more the test persons reward content of a certain emotional color with a like, the more likely it is that similarly emotionally charged content will be played out. So if a test person rewards a lot of feel-good content with a heart, the same test person is very likely to see more happy tweets.

Always enjoy study results critically!

The study authors point out that they had no internal access to the Twitter platform during their study, but only to the Twitter profiles of their subjects.

The study authors are nevertheless convinced that they can draw conclusions about long-term effects based on their short-term study.

In its article, the Evening Standard cites an opinion from Marvin Winkelmann, managing director at marketing firm AFK Digital. Winkelmann doubts whether hatred and malice have really increased on Twitter since Elon Musk took over.

Marvin Winkelmann also says:

Do the research results match your experiences on Twitter – or do you see it differently? Do you like to hang out on Twitter? What content will you flush into the timeline? Write us in the comments!