Link recommendation algorithms and polarization dynamics in online social networks
Polarization is increasing as political debates shift to online social platforms. In such contexts, algorithms are used to recommend new connections to users, via link recommendation algorithms. Users are often recommended based on structural similarity (for example, nodes sharing many neighbors are similar). We show that the preferential establishment of links with structurally similar nodes potentiates the polarization of opinions by stimulating network topologies with well-defined communities (even in the absence of opinion-based rewiring). When networks are made up of nodes that react differently to external contacts (converging or polarizing), connecting structurally dissimilar nodes improves moderate opinions. Our study sheds light on the impacts of social network algorithms on opinion dynamics and unveils avenues to guide the polarization of online social networks.
The level of antagonism between political groups has increased in recent years. Supporters of a given party increasingly hate members of the opposite group and avoid intergroup interactions, which leads to homophile social networks. While new offline connections are driven largely by human decisions, new connections on online social platforms are mediated by link recommendation algorithms, for example, ‘People You May Know’ or ‘People You May Know’ suggestions. Who to follow â. The long-term impacts of link recommendation in polarization are unclear, especially since exposure to opposing views has a dual effect: connections with external members can lead to a convergence of opinions and prevent group polarization or other separate opinions. Here, we provide a complex adaptive systems perspective on the effects of link recommendation algorithms. While several models justify polarization by rewiring based on similarity of opinion, we explain it here by rewiring based on structural similarity, defined as similarity based on network properties. We observe that the preferential establishment of links with structurally similar nodes (i.e. sharing many neighbors) results in network topologies which lend themselves to polarization of opinions. Thus, polarization occurs not due to a desire to protect oneself from unpleasant attitudes but, on the contrary, due to the inadvertent creation of echo chambers. When networks are made up of nodes that react differently to external contacts, whether converging or polarizing, we find that the connection of structurally dissimilar nodes moderates opinions. Overall, our study sheds light on the impacts of social network algorithms and unveils avenues to guide the dynamics of radicalization and polarization in online social networks.
Author contributions: research designed by FPS, YL and SAL; FPS, YL, and SAL have researched; Data analyzed by FPS; and FPS, YL and SAL wrote the article.
The authors declare no competing interests.
This article is a direct PNAS submission. CP is a guest editor invited by the Editorial Board.
This article contains additional information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2102141118/-/DCSupplemental.