Fake news: No ban, No spread -- with Sequestration

TL;DR

Problem: curb fake-news spread without broad prohibition. Approach: extend the Galam opinion dynamics framework to include tie-breaking prejudice and contrarian behavior, analyzing how these interact with fake-news content in a stylized setting; derive update rules such as $p_{4,k,x}$ and identify thresholds $x_c(k)$. Findings: in most parameter regions fake news fail to spread, but when content aligns with prejudices and contrarian share exceeds $x_c$, a tiny initial belief can invade; conversely, other parameter regimes induce sequestration; the critical thresholds vary with $k$. Significance: shows a nonrestrictive route to sequestrate invasive fake news by reshaping social geometry, highlighting the need for interdisciplinary collaboration to implement such measures in real-world social-media interventions.

Abstract

Fake news is today a major threat to free and democratic making of public opinion. To curb their spread, all efforts by institutions and policy makers rely mainly on imposing restriction, prohibition and fact checking sites, which end up to an effective limitation of freedom of speech. This policy of prohibition, supported by a wide consensus, has been recently broken by the controversial policy applied by Elon Musk to regulate the social media X, with a backlash accusing him of promoting hate speech. Here, notwithstanding these two policies, I explore another avenue denoted ``No ban, No spread - with Sequestration", which amounts at the same time preserving full freedom of speech and neutralization of fake news impact. To investigate the feasibility of my proposal, I tackle the issue within the Galam model of opinion dynamics. In addition to the basic ingredients of the model, I explore for the first time the effect on the dynamics of opinion of a simultaneous activation of prejudice tie breaking and contrarian behavior. The results show that indeed most pieces of fake news do not propagate beyond small groups of people and thus pose no global threat. However, I have unveiled some peculiar sets of parameters for which fake news, even if initially shared by only a handful of agents, spreads ``naturally" to invade a whole community with no resistance. Based on these findings, I am able to outline a path to neutralize such invasive fake news by blocking "naturally" its spread, effectively sequestering it in very small social networks of people. The scheme relies on reshaping the social geometry of the landscape in which fake news evolves. No prohibition is required with fake news left free to prosper but being sequestrated. Next challenging step will be designing measures to implement the model's findings into the real world of social media.

Figures (9)

• Figure 1: Evolution of attractors and tipping points as a function of the proportion $x$ of contrarians for $k=\frac{1}{2}$. The lower curve represents the attractor $p_{B, 0.5, x}$ (in blue), the upper curve the attractor $p_{A, 0.5, x}$ (in green and red) and the middle curve the tipping point $p_{t, 0.5, x}=0.5$ (in red). At $x_c=0.055$, $p_{A, 0.5, x_c}=_{B, 0.5, x_c}=_{t, 0.5, x_c}=0.5$. For $x>x_c$ the unique attractor is located at precisely $\frac{1}{2}$.
• Figure 2: Evolution of attractors and tipping points as a function of the proportion $x$ of contrarians for $k=1$. The lower curve represents the attractor $p_{B, 1, x}$ (in red), the upper curve the attractor $p_{A, 1, x}$ (in green and red) and the middle curve the tipping point $p_{t, 1, x}$ (in blue). At $x_c=0.055$, $p_{A, 1, x_c}=0.944$ and $p_{A, 1, 0.20} = 0.778$ and $p_{A, 1, 0.30} = 0.649$.
• Figure 3: Evolution of attractors and tipping point as a function of the proportion $x$ of contrarians for $k=0.60$. The lower curve represents the attractor $p_{B, 0.60, x}$ (in red), the upper curve the attractor $p_{A, 0.60, x}$ (in green and red) and the middle curve the tipping point $p_{t, 0.60, x}$ (in blue). At $x_c=0.114$, $p_{A, 0.60, x_c}=0.843$ and $p_{A, 0.60, 0.20} = 0.648$ and $p_{A, 0.60, 0.30} = 0.537$.
• Figure 4: Evolution of attractors and tipping points as a function of the proportion $x$ of contrarians for $k=0.53$ (left) and $k=0.501$ (right). The lower curves represent the attractor $p_{B, k, x}$ (in red), the upper curves the attractor $p_{A, k, x}$ (in green and red) and the middle curves the tipping point $p_{t, k, x}$ (in blue). On the left part $x_c=0.142$, $p_{A, 0.53, x_c}=0.753$ and $p_{A, 0.53, 0.20} = 0.562$ and $p_{A, 0.53, 0.30} = 0.511$. On the right part $x_c=0.164$, $p_{A, 0.501, x_c}=0.589$ and $p_{A, 0.501, 0.20} = 0.502$ and $p_{A, 0.501, 0.30} = 0.500$.
• Figure 5: Evolution of attractors and tipping points as a function of the proportion $x$ of contrarians for $k=0$. The lower curve represents the attractor $p_{B, 0, x}$ (in blue), the upper curve the attractor $p_{A, 0, x}$ (in red) and the middle curve the tipping point $p_{t, 0, x}$ (in green). At $x_c=0.055$, $p_{B, 0, x_c}=0.056$ and $p_{B, 0, 0.20} = 0.222$ and $p_{B, 0, 0.30} = 0.351$.