§ 230 Citation Network

A computational analysis of the federal appellate citation network for 47 U.S.C. § 230 of the Communications Decency Act, 1997-2025.

Overview

This repository contains the data, code, and documentation for a citation network analysis of 70 validated federal circuit court and Supreme Court opinions applying § 230(c). The analysis applies PageRank centrality, Leiden community detection, Jensen-Shannon divergence, and temporal network analysis to characterize the structure and evolution of § 230 doctrine.

Key findings

• Zeran v. AOL (4th Cir. 1997) holds the highest PageRank in every annual snapshot from 1997 through 2025 (PR = 0.202; α = 0.85)
• Five structural clusters identified via Leiden community detection on the undirected citation projection (modularity Q = 0.238; mean NMI = 0.686 across 100 runs), corresponding to recognizable doctrinal traditions confirmed by legal expert review
• Platform win rate under § 230 declined from 83% (2000-2004, n=6) to 41% (2020-2024, n=17); temporal trend is consistent with increasing judicial skepticism toward broad immunity claims
• Mean pairwise Jensen-Shannon divergence between circuits: 0.65 bits (base 2; range 0.43-0.86)
• DAG hierarchical level correlates with PageRank at Spearman ρ = -0.840 among cases with at least one inbound citation (n=41, p < 0.0001); full corpus ρ = -0.965 is inflated by 29 dangling nodes sharing identical PageRank floor values
• False positive rate in automated CourtListener retrieval: 66.8% (141/211 candidates excluded after review)

Repository structure

. ├── README.md ├── requirements.txt ├── research_design.md # Inclusion criterion, exclusion patterns, CBL ├── codebook.md # Field-level documentation for all data files ├── replication_guide.md # Step-by-step reproduction instructions ├── raw_data/ │ └── s230_validated_20260411_030728.json # Validated corpus (70 cases) ├── data/ │ ├── s230_graph.gexf # Citation graph (Gephi/NetworkX) │ ├── s230_graph.graphml # Citation graph (GraphML) │ ├── s230_metrics.csv # Per-node network metrics │ ├── case_outcomes_raw.json # Outcome codings │ └── [supplementary analysis files] └── [01-23]_*.py # Analysis scripts in execution order

Reproducing the analysis

See replication_guide.md for complete instructions. Quick start:

python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt
python3 02_build_graph.py && python3 10_merge_edges.py && python3 03_compute_metrics.py

Data

The validated corpus and graph files are deposited at Zenodo: [DOI to be added]

Citation

[Citation to be added upon publication]

License

Code: MIT License Data: Creative Commons Attribution 4.0 (CC BY 4.0)