Beyond Contagion: Decoding Competitive Information Spread via Evolutionary Game Theory

TL;DR

Information dissemination is often likened to a virus, but humans are not passive hosts. This paper moves beyond simple epidemic models to propose a Game-Theoretic Framework where individuals make rational choices between competing products. By modeling "Acceptance" through knowledge, interest, and memory, the study reveals that the speed and reach of information are determined by how we value social influence versus personal interest.

The Human Bottleneck: Why Epidemic Models Fall Short

Classical models like SIR (Susceptible-Infected-Removed) treat information spread as a biological process. However, in the real world—especially on social media—users face "information overload" and "competitive choices." You don't just "catch" a preference for an iPhone; you weigh it against a Samsung based on what you know, who you follow, and how much you care.

The authors argue that the missing link in current dissemination research is the rationality of the individual. Most prior work fails to account for:

• Memory Decay: We forget information over time.
• Interest Levels: Knowledge without interest does not lead to adoption.
• Strategic Interaction: We tend to coordinate our choices with our neighbors (Herd Mentality).

Methodology: The Logic of Utility

The core of this research is a coordination game where the payoff is defined by a multi-dimensional utility function.

1. The Utility Function

A player $i$'s utility $U$ is defined as: $U=\pi A+\sigma P$ Where $A$ (Acceptance) is the product of knowledge and interest, and $P$ (Popularity) is the social influence (degree centrality). The weights $\pi$ and $\sigma$ represent the player's "personality"—are they a self-driven learner or a social follower?

2. The Learning & Forgetting Mechanics

The paper introduces a unique Learning Operator ($\oplus$). When two users interact, they don't just swap information; they combine it based on their Willingness to Learn ($W$). Simultaneously, a Memory Matrix ($M$) based on the Ebbinghaus forgetting curve constantly reduces knowledge and interest at every time step: $k_{i,t+1}=e^{-1}{k}_{i,t}$

Note: The model utilizes a Markov chain to track strategy updates via Noisy Best Response Dynamics, allowing for occasional "irrational" choices (noise).

Case Study: Small-World Dynamics

The researchers tested the model on a small-world network (3,000 nodes). They introduced a "new" superior product into a network where an "old" product was already dominant.

Key Findings:

• Acceptance Weighting: As the importance of personal acceptance ($\pi$) increases, the network reaches consensus on the new product much faster.
• The Power of Knowledge: Increasing the initial knowledge distribution (even slightly) creates a "tipping point."
• Irrationality Matters: The noise factor $\beta$ (representing irrationality) slows down convergence but prevents the system from getting stuck in suboptimal local equilibria.

Figure: The graph shows that higher initial knowledge intervals (moving from left to right) lead to significantly faster and wider dissemination.

Critical Insight: The "Marketing" Takeaway

This paper provides a mathematical backbone for why content depth matters as much as reach. If a product provider only focuses on popularity ($P$) without boosting acceptance ($A$) through education (knowledge), the information will likely decay due to the forgetting curve before it takes root.

Limitations & Future Paths

While the model is robust, it assumes a homogeneous network where everyone follows the same utility rules. Future research should explore "Adversarial Agents" or "Influencers" who have vastly different $\pi$ and $\sigma$ values, and test the model against massive, real-world datasets from platforms like X (Twitter) or Weibo to validate the parameters.

Conclusion

By treating social network users as strategic players rather than passive nodes, this work bridges the gap between economics and network science. It proves that in the war for attention, the winner isn't just the loudest—it's the one who stays in the user's memory the longest.