Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learning from Viral Content

Krishna Dasaratha, Kevin He

Abstract

We study learning on social media with an equilibrium model of users interacting with shared news stories. Rational users arrive sequentially, observe an original story (i.e., a private signal) and a sample of predecessors' stories in a news feed, and then decide which stories to share. The observed sample of stories depends on what predecessors share as well as the sampling algorithm generating news feeds. We focus on how often this algorithm selects more viral (i.e., widely shared) stories. Showing users viral stories can increase information aggregation, but it can also generate steady states where most shared stories are wrong. These misleading steady states self-perpetuate, as users who observe wrong stories develop wrong beliefs, and thus rationally continue to share them. Finally, we describe several consequences for platform design.

Learning from Viral Content

Abstract

We study learning on social media with an equilibrium model of users interacting with shared news stories. Rational users arrive sequentially, observe an original story (i.e., a private signal) and a sample of predecessors' stories in a news feed, and then decide which stories to share. The observed sample of stories depends on what predecessors share as well as the sampling algorithm generating news feeds. We focus on how often this algorithm selects more viral (i.e., widely shared) stories. Showing users viral stories can increase information aggregation, but it can also generate steady states where most shared stories are wrong. These misleading steady states self-perpetuate, as users who observe wrong stories develop wrong beliefs, and thus rationally continue to share them. Finally, we describe several consequences for platform design.
Paper Structure (22 sections, 25 theorems, 121 equations, 4 figures)

This paper contains 22 sections, 25 theorems, 121 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Related Literature
  3. Model
  4. Discussion and Interpretation
  5. Strategy, Symmetric BNE, and Limit Equilibrium
  6. Steady States and Equilibrium Steady States
  7. Definition and Characterization of Steady States
  8. Informative and Misleading Steady States
  9. Equilibrium Steady States
  10. Numerical Illustrations of Equilibrium for $\lambda>\lambda^*$
  11. Comparative Statics
  12. Popularity Distributions
  13. Platform Design
  14. Changing Virality Weight Over Time
  15. Behavioral Interventions
  16. ...and 7 more sections

Key Result

Proposition 1

For any finite $n$ and parameters $q,K,C,\lambda,$ there exists a symmetric BNE. For any parameters $q,K,C,\lambda,$ there exists a limit equilibrium.

Figures (4)

  • Figure 1: The inflow accuracy function for the majority rule with $K=7,$$C=3,$$q=0.55,$$\lambda=1$.
  • Figure 2: The inflow accuracy function for the majority rule with $K=7,$$C=3,$$q=0.55,$$\lambda\approx 0.76$. Here $\phi_{\sigma^{\text{maj}}}$ has two fixed points: the left fixed point is a touchpoint that is only stable from the left side (the red box shows a zoomed-in view). The right fixed point is stable from both sides. Theorem \ref{['thm:ss_half_stable']} implies viral accuracy has a positive probability of converging to each of these two fixed points.
  • Figure 3: The inflow accuracy function for the majority rule with $q=0.55,$$K=7,C=3$, and $\lambda \in \{0.3,0.6,0.9\}$. With $\lambda=0.3$ and $\lambda=0.6,$ there is a single informative steady state. With $\lambda=0.9,$ a misleading steady state appears.
  • Figure 4: Estimated equilibrium mixing probabilities for different population sizes (solid curves) and estimated rational function (dashed curves).

Theorems & Definitions (54)

  • Definition 1
  • Definition 2
  • Proposition 1
  • Definition 3
  • Proposition 2
  • Definition 4
  • Theorem 1
  • Definition 5
  • Lemma 1
  • Definition 6
  • ...and 44 more