Bluesky has become an alternative to Twitter with a large left-leaning user base. (El editorial/Shutterstock)
In a nutshell
• Research shows Bluesky user growth directly correlates with controversial Twitter/X events, with major sign-up spikes following Musk’s announcements about fees and during platform outages.
• Over 63% of political content on Bluesky comes from left-leaning sources, with only 8% from right-leaning sources, creating a politically homogeneous space unlike Twitter’s more diverse user base.
• Despite innovative offerings like user-created “feeds” (40,000 created but used by only 139,000 of 5 million users), Bluesky follows similar network patterns as other social platforms, with few highly active users creating most content.
ZURICH — When Elon Musk took over Twitter in late 2022, users began jumping ship. Many found their way to Bluesky, a decentralized social media network initially conceived by former Twitter CEO Jack Dorsey. A new study published by the University of Zurich reveals this migration wasn’t random; it was directly triggered by specific actions taken by Twitter’s new management.
The research paper, published in PLOS One, offers the first deep dive into Bluesky’s user base, network structure, and political makeup. Study findings show a platform that looks remarkably similar to other social networks in its basic structure, but with one striking difference: it is largely dominated by users who share left-leaning political views.
Twitter Drama Drives Bluesky Growth
The researchers analyzed data from five million Bluesky users from the platform’s invite-only beta release from February 2023 through May 2024. They looked at all types of interactions: follows, likes, reposts, and replies.
What stood out was the clear link between Bluesky sign-ups and Twitter drama. The researchers spotted four key events that sent users flocking to Bluesky. On September 19, 2023, when Musk announced that Twitter users might face fees, Bluesky saw its biggest influx of new users relative to its existing user base. Other migration waves followed Twitter outages in July and December 2023, and the announcement of a $1 fee for new Twitter users in certain regions in October 2023.
This pattern shows Bluesky didn’t just grow naturally; it specifically attracted users frustrated with Twitter’s new direction under Musk. These reactions suggest that when users become frustrated with X, they view Bluesky as their most suitable replacement option.
A Left-Leaning Digital Haven
By analyzing the links shared on the platform, the researchers discovered that over 63% of political domains shared on Bluesky came from left-leaning sources, mostly “left-center.” Only about 8% came from right-leaning sources, with a tiny 0.16% classified as “extreme right.” The platform contains no content classified as “extreme left.”
This political makeup contrasts sharply with Twitter, which traditionally hosted users across the political spectrum. Bluesky has become a refuge for left-leaning users, much like how Parler and Truth Social became homes for conservative users.
Despite the left-leaning user base, clear divisions emerged on specific issues, particularly the Israel-Palestine conflict. Using a machine learning algorithm to categorize posts as pro-Israel, neutral, or pro-Palestine, the researchers found that before the October 7 attacks on Israel, most posts about Israel-Palestine relations were neutral. After the attacks, users split sharply, with pro-Palestinian voices significantly outnumbering pro-Israeli ones over time.
This finding highlights that even on politically similar platforms, hot-button issues can still create divisions. Political agreement on broad ideologies doesn’t mean users will agree on every specific topic.
Same Structure, New Features
Beyond its political makeup, Bluesky works much like other social networks. A small number of very active users create most of the content while most users have relatively few connections. This pattern seems to be basic to how social networks function, regardless of their technology.
One of Bluesky’s standout features is its “feeds” system. Unlike Twitter and Facebook, which use proprietary algorithms to recommend content, Bluesky lets users choose or even create their own content filtering systems. Users had created nearly 40,000 different feeds, from simple keyword filters to more complex algorithms. Popular feeds included “Discover” (showing trending content), “Science” (curated by scientists on the platform), and topic-specific collections for art and music.
However, despite all these options, only about 139,000 of Bluesky’s five million users had liked at least one feed. While the feature offers greater control, engagement with it has been relatively limited.
Unlike some alternative platforms, Bluesky has very little misinformation. Only 0.14% of posts contained links to domains classified as spreading fake news, conspiracy theories, or propaganda. Additionally, only 3,704 users (out of five million) had ever posted such content.
What Bluesky’s Rise Means for Social Media’s Future
What emerges from this research is a picture of Bluesky as a platform that follows many of the same social patterns as established social networks, despite its innovative technical features. Its user base leans heavily to the political left, likely as a direct response to perceived rightward shifts at Twitter under Musk.
When Twitter users became dissatisfied with Musk’s changes, they sought a new space with a different approach to moderation and community-building. What they created wasn’t just a Twitter clone with better privacy features, but a platform where left-leaning perspectives are more prevalent.
Paper Summary
Methodology
The researchers collected data from nearly five million Bluesky users through the platform’s API. Because of Bluesky’s decentralized nature, they could access public information about user profiles and activities. Starting with a list of 5.28 million user IDs, they gathered data for all active users and checked for any additional users until no new ones were found. To analyze political leanings, they looked at website domains from links shared by users and classified them according to Media Bias Fact Check ratings. For their analysis of the Israel-Palestine conflict, they built a machine learning model to detect the stance of posts related to the conflict, training it on a sample of 2,000 manually labeled posts.
Results
Bluesky’s network structure resembles other social networks, with a few highly active users and most users having limited activity. User sign-ups clearly spiked after specific events at Twitter/X, especially controversial announcements by Elon Musk and platform outages. Bluesky’s user base is overwhelmingly left-leaning, with over 63% of political content coming from left-of-center sources and less than 8% from right-leaning sources. The platform also showed very low levels of misinformation. Despite this political similarity, the researchers found significant division around the Israel-Palestine conflict, with distinct pro-Palestinian and pro-Israeli groups forming in the network.
Limitations
This study, while thorough, has some limitations. The researchers couldn’t share the raw data due to privacy concerns. The study used Media Bias Fact Check for political classifications, which represents just one way to measure political bias. When analyzing the Israel-Palestine conflict, their machine learning model achieved a 65.3% accuracy rate, meaning some posts may have been misclassified. Finally, the research covers a specific time period (February 2023 to May 2024), and user behavior may change as the platform grows.
Funding and Disclosures
The authors received no specific funding for this work.
Publication Information
The research was published in PLOS One in February 2025 by two researchers from the University of Zurich. If you’re curious about reading the full study, you can find it online as an open access article—meaning anyone can read it without a subscription. The researchers did this work without special funding, and they don’t have any financial ties that might influence their findings.
