Abstract: As social networks are ubiquitous in everyday life, problems such as misinformation, bots and polarisation of opinion gain more and more importance. This paper focuses on the last one as we propose novel methods to reduce the amount of polarisation in a social group. We leverage the voter model in which users hold binary opinions and repeatedly update their beliefs based on others they connect with. Stubborn agents who never change their minds ("zealots") are also disseminated through the network. We are interested in two equilibrium metrics, the diversity of opinions $\sigma$ and the density of active links $\rho$. The latter is a measure of the exposure to adverse opinions. We derive formulas to compute them that are valid in any directed, weighted network. Then we study the problem of finding optimal numbers of zealots in order to maximise these quantities. We account for the presence of a backfire effect, which may lead the group to react negatively and reinforce its level of polarisation in response. We provide exact solutions in the specific context of a complete unweighted graph, and propose a method for $\sigma$ in the general case. Finally we apply these problems to the network of the US House of Representatives. The network exhibits high levels of opinion diversity but lower levels of active links density. We find optimal numbers of zealots that maximise these quantities and show that both can be efficiently increased in some cases.
Introduction
In the study of social dynamics, understanding how disagreement—or "discord"—persists in networks is crucial for addressing issues like polarization and misinformation. Traditional voter models often focus on the time to reach consensus, but real-world social structures rarely achieve total agreement. They are characterized by "zealots" (stubborn agents) and persistent discord. This paper provides a rigorous mathematical framework to calculate exact discord probabilities for any directed, weighted network, moving beyond the limitations of previous mean-field approximations.
Theoretical Framework
The researchers model opinion evolution as a linear system of ordinary differential equations. By analyzing the rate at which agent pairs transition between agreement and disagreement, they derive a universal formula for the equilibrium state.
A key contribution is the "generalized active links density," which accounts for long-range influence through the matrix exponential of the adjacency matrix. This metric reveals that opinions are mathematically coupled if agents share common "ancestors" (sources of information), even without direct interaction.
Empirical Validation
The theoretical model was validated against simulations on four diverse real-world datasets: Zachary’s Karate Club, a college football network, an email communication network, and political blogs (Polblogs).
As shown in the figure above, the theoretical formulas match simulation results with high precision, with relative differences often as low as $10^{-3}$. The results demonstrate that discord ($\rho_{ij}$) is strongly influenced by path strength and ancestry similarity across all tested network types.
Key Insights: The Zealotry Paradox and Community Dynamics
The study reveals several counter-intuitive behaviors in polarized networks. In a Stochastic Block Model with two communities ($C_0$ and $C_1$), the level of discord depends heavily on the relative "zealousness" of the groups.
- The Zealotry Paradox: If one community is significantly more stubborn than the other, increasing connections between them can paradoxically decrease internal discord in the less-zealous group. The "loudness" of the external zealots forces the internal group into a forced consensus with the outsiders.
- Discord vs. Opinion Difference: There is a distinction between the probability of disagreement ($\rho$) and the distance between opinion distributions ($\Delta x$). While increased connectivity often raises the frequency of local disagreements, it can simultaneously cause the underlying opinion distributions to converge.
- Clustering as a Buffer: High clustering (many triangles in the network) acts as a stabilizer for consensus. Breaking these clusters through network rewiring significantly increases discord, especially in homophilic networks.
Conclusion
This work offers a powerful tool for identifying "bridge" nodes and optimizing zealot placement to manage network-wide tension. The authors applied these methods to the US House of Representatives network, finding that while opinion diversity is high, the density of active links is lower, suggesting potential for targeted depolarization. While the current model assumes static weights, it lays the groundwork for future studies on adaptive networks where agents change their connections based on the opinions they encounter.
Summary updated