Self-moderation in the decentralized era: decoding blocking behavior on Bluesky

TL;DR

Self-moderation on Bluesky investigates blocking dynamics in a decentralized network using a three-month public Firehose dataset to build 86-feature profiles of user activity, content, and network interactions. The study applies clustering, binary classification, and regression to predict being blocked and to identify informative predictors, finding that high activity correlates with more blocks and that graph-centrality features (e.g., Degree, Coreness) alongside engagement signals strongly predict blocking, achieving ROC AUC up to $0.88$ at higher blocking thresholds. SHAP analyses reveal that a small subset of features can match performance, suggesting potential for lightweight, privacy-preserving moderation aids, while AutoML-based regression yields $R^2$ around $0.48$ for estimating block counts. Overall, the work demonstrates the feasibility and value of data transparency for studying moderation in decentralized social networks and informs the design of safety tools that respect user autonomy.

Abstract

Moderation and blocking behavior, both closely related to the mitigation of abuse and misinformation on social platforms, are fundamental mechanisms for maintaining healthy online communities. However, while centralized platforms typically employ top-down moderation, decentralized networks rely on users to self-regulate through mechanisms like blocking actions to safeguard their online experience. Given the novelty of the decentralized paradigm, addressing self-moderation is critical for understanding how community safety and user autonomy can be effectively balanced. This study examines user blocking on Bluesky, a decentralized social networking platform, providing a comprehensive analysis of over three months of user activity through the lens of blocking behaviour. We define profiles based on 86 features that describe user activity, content characteristics, and network interactions, addressing two primary questions: (1) Is the likelihood of a user being blocked inferable from their online behavior? and (2) What behavioral features are associated with an increased likelihood of being blocked? Our findings offer valuable insights and contribute with a robust analytical framework to advance research in moderation on decentralized social networks.

Figures (12)

• Figure 1: Overall methodology for the analysis.
• Figure 2: (left) Time series of the number of daily actions performed on Bluesky during the observation period. Daily observations and a 7-day moving average are reported. The scale of the y-axis is logarithmic. (right) Distributions of the number of daily actions performed on Bluesky during the observation period. The scale of the y-axis is logarithmic.
• Figure 3: Empirical cumulative distributions of the number of actions (log scale) performed by Bluesky users. Median values are: blocks = 2, follows = 2, likes = 8, posts = 6, reposts = 4.
• Figure 4: Empirical cumulative distribution of the number of blocks received by a user (log scale), considering users that shared at least 10 posts over the observation period.
• Figure 5: Correlation between different user features -- number of actions, average credibility of news shared and average toxicity -- and the number of blocks received. We show a linear fit to guide the eye and 100 bins with 95% confidence intervals. The analysis is performed on a sample of 100k users due to computational constraints.