Artificial IntelligenceBigTech CompaniesNewswireTechnology

Bluesky Hits 40M Users, Rolls Out ‘Dislikes’ Feature

▼ Summary

– Bluesky has reached 40 million users and will soon test “dislikes” to improve personalization in feeds and reply rankings.
– The platform is introducing updates to foster more fun and respectful exchanges, including better toxic comment detection and reply controls.
– Bluesky emphasizes user tools for experience control, such as moderation lists, content filters, and the ability to detach quote posts.
– A new “social neighborhoods” feature will prioritize replies from users with closer connections to make conversations more relevant.
– Additional changes include downranking toxic replies and modifying the reply button to encourage reading threads before responding.

The social media platform Bluesky has surpassed 40 million users and is preparing to test a new “dislikes” feature aimed at refining personalization within its Discover feed and other areas. This development arrives alongside several conversation management upgrades, including reply adjustments, enhanced toxic comment detection, and methods for prioritizing more relevant discussions for each person.

Once the dislikes beta launches, Bluesky will use this input to better understand what content individuals prefer to avoid. The system will learn from these signals to adjust not only how posts are ranked in feeds but also how replies are ordered. The company emphasized that these changes are intended to foster more enjoyable, authentic, and considerate interactions. This focus comes after recent user complaints regarding moderation choices, even though Bluesky operates as a decentralized network where individuals manage their own moderation.

Rather than imposing platform-wide bans, Bluesky is concentrating on providing users with tools to shape their own experience. Current options include moderation lists for blocking multiple accounts at once, content filter controls, muted words, and subscriptions to external moderation services. Users can also detach quote posts to reduce unwanted attention, addressing the toxic “dunking” culture familiar from other platforms.

Beyond dislikes, Bluesky is experimenting with various ranking updates, design modifications, and additional feedback mechanisms to enhance conversational quality. One notable initiative involves mapping “social neighborhoods”, identifying clusters of users who frequently interact. By prioritizing replies from those within a user’s network, Bluesky aims to make feeds feel more relevant and connected. The dislikes feature may also play a role in fine-tuning these neighborhood-based recommendations.

This approach could resolve a common issue faced by competitors like Meta’s Threads, where users sometimes encounter disjointed conversations from unfamiliar sources. As one observer noted last year, Threads often presented confusing feeds where posts seemed to appear at random, making it difficult to understand context or connections. Bluesky’s neighborhood mapping may offer a clearer, more coherent experience as the platform expands.

Additionally, Bluesky’s latest model improves identification of toxic, spammy, off-topic, or insincere replies, downranking them in threads, search results, and notifications. A change to the Reply button now directs users to the full conversation thread before allowing them to compose a response, encouraging people to read existing comments first. This simple adjustment seeks to minimize redundant replies and “content collapse,” a frequent criticism of other social networks.

The platform is also making reply settings more prominent, ensuring users are aware they can control who responds to their posts. These collective updates reflect Bluesky’s commitment to putting customization and user control at the forefront of its growth strategy.

(Source: TechCrunch)

Topics

user personalization 95% dislike feature 90% moderation tools 85% conversation control 80% toxic comments 80% social neighborhoods 75% content filtering 70% reply rankings 70% feedback tools 70% decentralized moderation 65%