A longitudinal analysis of misinformation, polarization and toxicity on Bluesky after its public launch

TL;DR

This study delivers the first large-scale, longitudinal analysis of Bluesky’s transition to public availability, focusing on engagement, language use, political leaning, misinformation credibility, toxicity, and moderation. It leverages 56 days of Firehose data, NewsGuard/MBFC labels, and Louvain-based community detection to map a rapid growth phase, language shifts (notably English and Japanese), and a left-leaning user base with predominantly high-credibility sharing. The findings show Bluesky’s community structure evolves with substantial original content, manageable toxicity, and active moderation, though simultaneous signals of mass-following behaviors and suspicious actors surface after launch. The work highlights regional adoption patterns, especially the influx of Japanese users, and informs ongoing discussions about moderation in decentralized networks, while acknowledging limitations such as the short observation window and partial labeling coverage. The results have practical implications for understanding user behavior in emergent decentralized platforms and for shaping moderation strategies in nascent social networks.

Abstract

Bluesky is a decentralized, Twitter-like social media platform that has rapidly gained popularity. Following an invite-only phase, it officially opened to the public on February 6th, 2024, leading to a significant expansion of its user base. In this paper, we present a longitudinal analysis of user activity in the two months surrounding its public launch, examining how the platform evolved due to this rapid growth. Our analysis reveals that Bluesky exhibits an activity distribution comparable to more established social platforms, yet it features a higher volume of original content relative to reshared posts and maintains low toxicity levels. We further investigate the political leanings of its user base, misinformation dynamics, and engagement in harmful conversations. Our findings indicate that Bluesky users predominantly lean left politically and tend to share high-credibility sources. After the platform's public launch, an influx of new users, particularly those posting in English and Japanese, contributed to a surge in activity. Among them, several accounts displayed suspicious behaviors, such as mass-following users and sharing content from low-credibility news sources. Some of these accounts have already been flagged as spam or suspended, suggesting that Bluesky's moderation efforts have been effective.

Figures (11)

• Figure 1: Bar chart summarizing key dataset statistics over the whole observation period: total posts, posts containing links, active users, shared messages (with links), follow actions, and block actions. Values are presented on a logarithmic scale due to different orders of magnitude.
• Figure 2: Online activity on Bluesky before and after the public opening (Feb. 6th), indicated by the dashed line.
• Figure 3: (A) Trend of 5 top languages on Bluesky during the observation period. (B) Top 10 languages in Bluesky. Bars of the same color sum to 100%.
• Figure 4: Complementary cumulative distributions of node (A) out-degree and (B) in-degree in the follower network.
• Figure 5: Distribution of the average political leaning score of users that shared at least 5 links to rated domains.