Related Experiment Video
Updated: Jan 26, 2026

Integrating Computerized Linguistic and Social Network Analyses to Capture Addiction Recovery Capital in an Online Community
Published on: May 31, 2019
Information flow reveals prediction limits in online social activity
James P Bagrow1,2, Xipei Liu3,4, Lewis Mitchell5,6,7
1Department of Mathematics & Statistics, University of Vermont, Burlington, VT, USA. james.bagrow@uvm.edu.
Predicting user behavior on social networks is possible using only their contacts' data. An individual's online activities can be accurately predicted using just 8-9 social ties, highlighting significant privacy concerns.
Area of Science:
- Social Network Analysis
- Information Theory
- Computational Social Science
Background:
- Online social networks generate vast user behavioral data.
- Predicting individual activities from social ties has remained a challenge.
Discussion:
- Information theoretic tools estimated predictive information in Twitter user writings.
- An upper bound for predictive information was established for any machine learning method.
Key Insights:
- 95% of potential predictive accuracy is achievable using only an individual's social ties.
- As few as 8-9 contacts can provide predictability comparable to the individual's own data.
- Distinct temporal and social effects in information flow along social ties were identified.
Outlook:
- Results have significant privacy implications, enabling individual profiling from social ties alone.
- Understanding information flow dynamics can advance online activity studies.
More Related Videos
Related Concept Videos
Physiological Pharmacokinetic Models: Blood Flow-Limited Versus Diffusion-Limited Models
The Sense of Self: Reflected Self-Appraisal and Social Comparison
Social Traps
Social Proof
Social Scripts
Limiting Reactant

