Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 11: Polarization
SOE 11.4: Vortrag
Mittwoch, 11. März 2026, 16:45–17:00, GÖR/0226
Information Saturation in the Political Center Drives the Spread of Extreme Content — •Markus Hofer1,2, Jan Korbel2, and Stefan Thurner1,2,3 — 1Center for Medical Data Science, Institute of the Science of Complex Systems, Medical University of Vienna, Vienna 1090, Austria — 2Complexity Science Hub Vienna, Vienna 1030, Austria — 3Santa Fe Institute, Santa Fe, NM 87501
Although most voters hold moderate opinions, online discourse is increasingly dominated by ideological extremes. To understand this phenomenon, we employ an agent-based model in which post-sharing mediates all opinion dynamics. First, agents filter posts based on ideological distance to update their own opinions. Then they select one of the received posts based on novelty, quantified as the local information-theoretic surprisal. Here we show that empirically consistent opinion distributions with moderate polarization naturally emerge alongside a disproportionate sharing of extreme content. We identify the underlying mechanism as a density-dependent competition for novelty: high post density in the political center leads to information saturation, effectively suppressing surprisal and resharing. In contrast, extreme content maintains high surprisal, increasing its resharing probability. This mechanism explains the heavy-tailed cascade distributions observed on Twitter/X and points toward interventions that reduce the informational novelty advantage of extreme content.
Keywords: agent-based model; information cascades; opinion formation; social media