Did Decentralized SNS''s ''Quality Management Experiment'' Succeed?
Decentralized social media Bluesky published its "2025 Transparency Report." As of end 2025: approximately 41.41 million users (60% increase YoY); 1.41 billion posts generated (61% created in 2025). With simultaneous explosive user and traffic growth, post-incident-response-centered moderation reached clear limits — Bluesky set 2025 as a year for experimenting with a pre-intervention-centered moderation model rather than reactive response. Biggest change — toxic content response: instead of content deletion or account suspension, Bluesky focused on reducing initial exposure of potentially problematic speech. System applied from October 2025: posts classified as profanity/spam/hostile replies hidden from default feed; users must make additional clicks to view; followed account replies prioritized for visibility, maintaining conversation context while reducing accidental toxic content contact. Result: daily reports related to harassment and antisocial behavior decreased approximately 79%. Why this matters: demonstrates that decentralization and no-moderation are NOT synonyms; Bluesky proved that a decentralized platform can implement systematic safety management through design rather than control; the approach — reducing exposure of potentially harmful content rather than deleting it — preserves more of the content (it''s still viewable with extra effort) while reducing casual encounters with toxic material. The broader implication: Bluesky''s 2025 experiment is the first large-scale empirical test of "can algorithmic design substitute for content moderation?" — the 79% reduction figure suggests yes, at least partially, and at scale. The question for 2026: whether this improvement is stable at continued growth or requires escalating moderation as bad actors adapt to the new system.
