The SIS Competition Model for Conflicting Rumors

This paper proposes an SIS competition model for conflicting rumors that reveals a novel coexistence mechanism where belief in one rumor paradoxically aids the other, leading to counterintuitive outcomes such as lower infection rates enhancing spread and the spontaneous emergence of majority advantage despite the absence of explicit conformity rules.

The Gist

Imagine a bustling town square where two groups of people are shouting different stories to the crowd. One group is spreading a Fake News story (let's call it "The Monster Under the Bed"), and the other is spreading a Correction (let's call it "It's Just a Shadow").

Usually, we think of these stories like viruses: the one that is "stickier" or more exciting wins, and the other dies out. But this paper proposes a surprising new way to look at how these stories fight for attention. It suggests that sometimes, the very people believing the fake news are actually the ones helping the correction spread, and vice versa.

Here is the breakdown of the "SIS Competition Model" using simple analogies:

1. The Two Roles: Believers vs. Shouters

In this model, everyone in the town has two things happening at once:

• What they believe: They either believe the Monster story (Opinion A) or the Shadow story (Opinion B).
• What they are doing: They are either Silent (they believe it but aren't talking about it) or Shouting (they are actively spreading the rumor).

The magic happens when a Silent person hears a Shouter. Two things can occur:

1. The Echo: They agree and start shouting the same story.
2. The Flip: They hear the story, realize it's wrong (or right), change their mind, and immediately start shouting the opposite story.

2. The Paradox: How "Weak" Rumors Win

The most fascinating discovery in the paper is a paradoxical mechanism.

Imagine the "Monster" story is very boring to its own believers (they don't shout much). However, when a believer of the "Shadow" story hears the "Monster" story, they are so shocked that they immediately flip their opinion and start shouting the "Monster" story to warn others.

The Analogy:
Think of the "Monster" story as a weak, slow-moving fire. Normally, a weak fire dies out. But in this model, the "Shadow" believers are like a pile of dry leaves. When the weak fire touches the dry leaves, the leaves don't just burn; they explode into a new fire that spreads the "Monster" story to new people.

The Result:
The "Monster" story survives because it is shocking to the opposition. If the "Monster" story were too strong and convincing to its own believers, they would just shout it among themselves. But because it relies on shocking the other side to spread, it creates a weird, stable balance where both stories exist side-by-side.

3. The "Critical Mass" Rule (The Tipping Point)

The paper also explains why sometimes a "better" story loses.

Imagine you have a very persuasive "Shadow" story (high infection rate), but you start with very few people believing it. The "Monster" story has fewer believers, but they are already a large group.

The Analogy:
Think of a tug-of-war. Even if the "Shadow" team is stronger (better at pulling), if they start with only 3 people against 97 "Monster" believers, they will lose. The "Monster" team has so many people that the "Shadow" team can't get a foothold.

The paper calculates a specific Threshold. If the "Shadow" story doesn't start with enough believers to cross this line, it will be crushed, even if it is technically "better" at spreading. The majority wins simply because they are already there, creating a self-reinforcing loop (even without people trying to follow the crowd).

4. The Four Possible Outcomes

Depending on how "sticky" the stories are and how easily people change their minds, the town ends up in one of four states:

1. Silence: Both stories are too boring. No one shouts. Everyone stays quiet.
2. Monster Wins: The Monster story is so sticky that it takes over the whole town.
3. Shadow Wins: The Shadow story is so sticky that it takes over the whole town.
4. The Truce (Coexistence): This is the unique finding. Both stories survive.
   • How? The "Monster" story is boring to its own fans but shocks the "Shadow" fans into switching.
   • The "Shadow" story is boring to its own fans but shocks the "Monster" fans into switching.
   • They feed off each other's opposition, creating a permanent, chaotic balance where both rumors live on.

Why This Matters

This model helps us understand real-world problems like Fake News vs. Fact-Checking.

• The Lesson for Fact-Checkers: If you want to stop a fake rumor, simply making a correction isn't enough. If the correction is too boring or complex for people to care about on its own, it won't spread. It needs to be "sticky" enough to survive on its own merits, not just rely on shocking the fake-news believers.
• The Lesson for Fake News: Sometimes, a fake story survives because it is so outrageous that it forces people to argue about it, inadvertently spreading it further.

In short: In the battle of rumors, it's not just about who is "right" or who is "louder." It's about how the two sides interact. Sometimes, the enemy is the best friend of the rumor, keeping it alive by constantly trying to kill it.

Technical Summary

Here is a detailed technical summary of the paper "The SIS Competition Model for Conflicting Rumors" by Yu Takiguchi and Koji Nemoto.

1. Problem Statement

The paper addresses the complex dynamics of conflicting rumors spreading simultaneously in a society, such as fake news versus corrective information, or political support versus criticism. While previous studies have utilized epidemic models (e.g., SIR, SIS) to model rumor propagation and separate opinion models (e.g., Voter model) to study belief evolution, there is a lack of integrated frameworks. Specifically, existing models often fail to capture the bidirectional coupling where:

1. Rumors influence opinions: Hearing a rumor can change an individual's underlying belief.
2. Opinions influence rumor transmission: An individual's current belief determines their likelihood of spreading a specific rumor (e.g., supporters of a party spreading positive rumors while ignoring negative ones).

The authors aim to determine the conditions under which conflicting rumors coexist, one eradicates the other, or both die out, and to identify the mechanisms driving these outcomes.

2. Methodology

The authors propose a novel SIS (Susceptible-Infected-Susceptible) Competition Model that integrates rumor propagation with opinion dynamics on a well-mixed population.

Model Formulation

• States: Individuals are categorized into four states based on their supported opinion ($r \in \{A, B\}$) and their activity:
  • $S_r$: Supports opinion $r$ but does not spread the rumor.
  • $I_r$: Supports opinion $r$ and actively spreads the rumor.
• Transitions:
  • Propagation: $S_r \to I_r$ at rate $\beta_{r \to r}$ (spreading to a supporter of the same opinion).
  • Opinion Change & Propagation: $S_{r'} \to I_r$ at rate $\beta_{r' \to r}$ (hearing a rumor for opinion $r$ changes the individual's opinion to $r$ and motivates them to spread it).
  • Recovery: $I_r \to S_r$ at rate $\gamma_r$ (losing interest in spreading).
• Key Assumption: A change in opinion ($S_{r'} \to I_r$) is assumed to occur simultaneously with the adoption of the rumor, based on the premise that impactful rumors that alter beliefs also motivate the new believer to spread the message.

Analytical Tools

1. Basic and Effective Reproduction Numbers:
   • Defined $R_0$ values for same-opinion spread ($R_A^0, R_B^0$) and cross-opinion spread ($R_{B \to A}^0, R_{A \to B}^0$).
   • Defined Effective Reproduction Numbers ($R_t^A, R_t^B$) which depend on the current fractions of supporters ($S_A, S_B$).
2. Linear Stability Analysis: The authors derive all steady states (disease-free, single-rumor, and coexistence) and analyze the eigenvalues of the Jacobian matrix to determine local stability.
3. Singular Perturbation Method: To analyze the realistic regime where opinion changes are rare ($\beta_{r' \to r} \ll \beta_{r \to r}, \gamma_r$), the authors treat opinion change rates as small parameters ($\epsilon$). They separate the time scales into fast rumor propagation and slow opinion evolution to derive analytical approximations for the threshold of dominance.

3. Key Contributions

• Integration of Dynamics: The model successfully couples rumor transmission with opinion evolution, revealing that the "infection rate" is not static but depends on the current distribution of opinions.
• Novel Coexistence Mechanism: The paper identifies a counter-intuitive mechanism where conflicting rumors can coexist stably. This occurs when a rumor is trivial to its own supporters (low $R_0$) but shocking to opponents (high cross-infection rate), allowing the "opposing" rumor to recruit new spreaders from the other group.
• Emergent Majority Conformity: Although the model does not explicitly include a "majority conformity" term (where people adopt the majority view), the analysis shows that a majority opinion spontaneously gains a competitive advantage. If an opinion's initial share exceeds a critical threshold, it can eliminate a rumor with a higher intrinsic infection rate.
• Analytical Threshold Derivation: Using singular perturbation theory, the authors derive an explicit analytical formula for the critical threshold ($\rho_{ss}^A$) that determines which opinion survives in the limit of rare opinion changes.

4. Key Results

Phase Diagrams and Steady States

The system exhibits four distinct phases based on the basic reproduction numbers:

1. D Phase (Extinction): Both rumors die out. Occurs when effective reproduction numbers are $<1$.
2. A/B Phase (Single Dominance): Only rumor A or B survives. This happens if one rumor can sustain a pandemic ($R_0 > 1$) and outcompetes the other in a population of its own supporters.
3. C Phase (Coexistence): Both rumors persist.
   • Condition: Coexistence is stable when cross-infection rates are high relative to same-opinion rates ($R_{B \to A}^0 > R_A^0$ and $R_{A \to B}^0 > R_B^0$).
   • Mechanism: Supporters of A spread the rumor to B (changing their opinion), and vice versa. If the "correction" is too weak to eradicate the "fake news" but strong enough to recruit new spreaders, a dynamic equilibrium is reached.

The Paradox of Lower Infection Rates

The model reveals that a rumor with a lower intrinsic infection rate can sometimes spread more effectively than a stronger one if it benefits from the "shock value" to the opposing group. Conversely, a rumor with a high infection rate may fail if it cannot recruit from the opposing group effectively.

The Critical Threshold (Perturbation Analysis)

In the regime of rare opinion changes, the system relaxes quickly to a quasi-steady state where rumor fractions are determined by current opinion shares. The slow evolution of opinion shares ($\rho_A$) is governed by:
$$\dot{\rho}_A \approx \alpha (\rho_A - \rho_{ss}^A)$$

• $\rho_{ss}^A$ is an unstable fixed point (threshold).
• If the initial share $\rho_A(0) > \rho_{ss}^A$, opinion A grows to 100%.
• If $\rho_A(0) < \rho_{ss}^A$, opinion A dies out.
• Significance: Even if rumor A has a higher transmission potential, it will be eliminated if its initial support is below this critical threshold. This explains why correcting misinformation is difficult once a false narrative has gained a foothold.

5. Significance and Implications

• Policy Implications for Fake News: The model suggests that to eradicate fake news (Opinion A) using corrective information (Opinion B), the corrective information must be "interesting" enough to survive on its own ($R_B^0 > 1$). Merely shocking the believers of fake news is insufficient if the correction cannot sustain itself independently.
• Understanding Polarization: The spontaneous emergence of majority dominance explains why entrenched opinions are hard to overturn, even without explicit social pressure or conformity algorithms. The sheer number of existing supporters creates a buffer that prevents the "minority" opinion from gaining the critical mass needed to trigger a cascade.
• Theoretical Advancement: The work bridges the gap between epidemiology and opinion dynamics, providing a mathematically rigorous framework to study the feedback loop between belief and information spread. It challenges the assumption that higher transmission rates always lead to dominance, highlighting the role of initial conditions and cross-group interactions.

In conclusion, Takiguchi and Nemoto provide a robust analytical framework showing that the fate of conflicting rumors is determined not just by their intrinsic virulence, but by the interplay between their ability to recruit from opposing groups and the initial distribution of public opinion.