New Anti-Toxicity Features on Bluesky

Original Article ↗ Hacker News Discussion ↗

Feature Understanding & Intent

  • New tools let original posters:
    • Hide replies on their own thread from themselves and their followers (still visible to the replier’s audience and via API).
    • “Detach” quote-posts so their name/post no longer appears inline in the quoter’s post, though the quote still exists in that user’s repo and can be accessed by other clients.
  • Supporters frame this as adding friction to harassment and dogpiling, not as deleting others’ speech, and note it’s similar in spirit to “hide reply” tools on other platforms.

Harassment, Dogpiling, and Lived Experience

  • Several describe being quote-posted by large accounts, leading to sudden hostile mobs, death threats, and thousands of notifications.
  • They argue quote-posts collapse context and shift power entirely to the quoter; even honest misunderstandings can trigger abuse.
  • For them, tools to sever that audience link are “necessary safety” rather than censorship, especially given Bluesky currently lacks private profiles.

Echo Chambers, Censorship, and Abuse Risk

  • Critics say the same tools let users:
    • Remove visible critical replies while keeping supportive ones.
    • Use “harassment” as a catch-all to hide dissent or fact-checks.
  • They predict stronger echo chambers, empowerment of trolls, propagandists, and scammers who can hide rebuttals, and “affirmation bubbles” marketed as anti-toxicity.
  • Some argue “hiding” is functionally equivalent to deleting in practice, since few will click through to see hidden replies.

Decentralization and Client Behavior

  • Concern: such features seem incompatible with a decentralized network if other servers/clients can ignore them.
  • Response: ATProto is decentralized at the data layer (user repos), while apps are centralized aggregators; “detach” is an intent record that well-behaved clients respect, like robots.txt.
  • Critics worry this still assumes a de facto “official client” and can mislead users about the real limits of protection.

Comparisons, UI & Moderation Philosophy

  • Comparisons to Reddit, HN, Mastodon, Twitter/X:
    • Examples of echo chambers, heavy moderation, ragebait, and quote-post dogpiles.
  • Broader debate over:
    • Whether downvotes alone suffice vs need for strong moderation.
    • Where moderation ends and censorship begins.
  • UI feedback: current “Removed by author” wording is ambiguous about who removed what and may confuse on screenshots.