CSEG: Modeling the Zero-Sum War for User Attention in Saturated Social Networks

TL;DR

In the "developed phase" of the internet, social networks no longer fight for new users—they fight for the minutes spent on their apps. This paper introduces the Competitive Social Evolutionary Game (CSEG), a framework that models SNS competition using evolutionary game theory. By treating user attention as a finite vector, the authors reveal how platform rules and network structures (like Small-World vs. Regular) dictate which service survives the attention war.

Context: The Shift to the "Developed Phase"

The era of explosive user growth is over. We have reached a point where the number of SNS users nearly equals the total number of netizens. In this "Developed Phase," an increase in time spent on TikTok inherently means a decrease in time spent on Instagram. Existing Social Evolutionary Game (SEG) models were insufficient because they focused on internal dynamics (cooperation vs. defection) within a single silo. The authors argue that we must view SNS platforms as competing organisms feeding on a finite pool of human attention.

Methodology: The Mechanics of CSEG

The core innovation lies in the User Attention Matrix ($A_{mn}$), where each element $a_{ij}$ represents the percentage of time user $j$ spends on platform $i$.

1. The Game Logic

Users engage in a Prisoner's Dilemma Game (PDG) on multiple platforms simultaneously:

• Cooperation (C): Reposting or interacting with a neighbor's status.
• Defection (D): Consuming content without reciprocation.
• Utility ($U$): The short-term gain from these interactions.
• Reputation ($R$): The long-term trust score based on past behavior.

2. The Coevolutionary Loop

The model cycles through three phases:

1. Strategy Updating: Users imitate successful neighbors (Fermi update rule).
2. Partnership Updating: Users sever ties with low-reputation "defectors" and link with high-reputation "cooperators."
3. Attention Adjustment: This is the "Competitive" bridge. If a user sees higher cooperation rates or better utility on SNS A vs. SNS B, they shift their attention (time) toward SNS A.

Figure 1: The coevolutionary process where players interact across multiple SNs, adjusting strategies and ties.

Simulation Insights: Small Worlds and Reputation

The authors simulated competition between a Regular Network and a Small-World Network.

The "Temptation" vs. "Reputation" Balance

A critical parameter in their PDG is $b$ (the temptation to defect). The simulations (visualized via heatmaps of "ending time steps") show that:

• If two platforms have similar temptation values, the "war" lasts a long time (high stability).
• If one platform lowers the temptation to defect (by incentivizing cooperation), it drains attention from the other much faster.
• Reputation Awareness ($p_r$): Higher $p_r$ acts as a "buffer." Even if a network has high temptation to defect, if users are highly sensitive to reputation, the platform remains stable because the "social cost" of defection is too high.

Figure 2: Heatmaps showing the evolution time. Red areas indicate longer, more stable competition, while blue areas show rapid dominance by one platform.

Critical Analysis & Conclusion

The CSEG model provides a rigorous mathematical grounding for what product managers call "retention" and "stickiness." It proves that a platform's survival is not just about its features, but about its social ecosystem's health.

Takeaways for the Future:

• Incentive Engineering: Platform owners can suppress defection (trolling, low-quality spam) by increasing "Reputation Awareness"—essentially making a user's social history more visible and impactful.
• Structure Matters: Small-world networks are more sensitive to initial conditions. A "Small-World" SNS that starts with a poor reputation mechanism is doomed to lose the attention war faster than a "Regular" one.
• Limitations: The model assumes a fixed user pool and fixed total time. Real-world dynamics might involve "attention expansion" (spending more total hours online), which this model does not yet account for.

Ultimately, this work shifts the focus of social network research from growth to sustainability, providing a game-theoretic playbook for the age of attention scarcity.