Algorithms have nothing to do with it, we are passionate about toxic discussions on social media

2024-03-23 07:44:16

AGI – Despite profound changes in platform technologies, social conditions and above all in algorithms, human behavior on social media has remained constant over the last 34 years. This is the result of a new study published in “Nature” by researchers from the Center for Data Science and Complexity for Society at the Department of Computer Science of the Sapienza University of Rome, coordinated by Walter Quattrociocchi.

The research was based on data extrapolated from various platforms – Facebook, Reddit, Gab, Youtube etc. up to the less recent USNET on more than 500 million comments. Through a comparative examination of the data itself, recurring patterns of interactions between users were identified, and it was discovered that they remain constant despite the great changes that have occurred over the years. In particular, it should be noted that it became clear that toxic interactions do not influence the engagement of users who continue to participate in conversations even if they have become highly polarized. Furthermore, it emerged that each of us tends to contribute to toxicity, with a marginal importance of single individuals or groups of individuals.

“Despite different platforms and their different algorithms, as well as different social norms, the dynamics of toxicity are always the same. And this applies to very different online communication topics, confirming the universality of this pattern. Furthermore, what emerges is that there is no such thing as a serial hater and we all poison ourselves more or less in the same way on all issues.” (Walter Quattrociocchi)

“Again – continues Quattrociocchi – what emerges, and which is counterintuitive, is that in the face of toxic communication in which heavy comments begin to appear capable in theory of destroying a conversation, instead the conversations do not stop but continue , demonstrating that the social media ecosystem has a strong resilience to toxicity. All elements which, combined with the persistent online polarization – in turn deeply linked to the toxicity of language, so much so that it is a predictive factor – make us imagine that the repercussions on the electoral results of the coming months around the world could be significant. This is why as the Center for Data Science we have established an observatory to monitor the progress of upcoming elections at a global level – Italy, USA, India etc. – so as to study the patterns of related communications and to understand which of the dynamics that emerged in our study will affect the vote”.

The analysis also points to polarization as a significant underlying factor contributing to this toxicity, highlighting the role of divergent viewpoints in fueling online discord. However, the presence of toxic content does not discourage user participation, indicating a complex relationship between toxicity and engagement in online discourse.