Ask HN: Is it time to fork HN into AI/LLM and "Everything else/other?"

Hacker News Discussion ↗

Proposal & Context

  • Thread asks whether HN should split into “AI/LLM” and “everything else” because AI-related posts and comments feel overwhelming and crowd out diverse, serendipitous content.
  • Several participants say this complaint appears repeatedly, and that on some days AI stories take ~1/3 of the front page or more, plus AI shows up in comment threads on unrelated topics.

Arguments Against Forking HN

  • Many oppose any split: HN’s “secret sauce” is its stability and single stream; fragmentation risks ghost-town sub-sites and weaker discussion.
  • AI/LLM is seen as a central, current frontier of tech and startups; HN’s mandate is “whatever good hackers find interesting,” so showing lots of AI is functioning as designed.
  • Others note that previous fads (Erlang, Rails, JS frameworks, crypto, Rust, Web3, etc.) also dominated and then subsided; AI is framed as another long hype cycle, though some say this one is larger and more persistent.

Desire for Filtering, Not Exclusion

  • Many would like to keep one HN but have topic filters or tags:
    • Server-side tags (like lobste.rs) to follow/block topics.
    • User-side keyword filters, uBlock rules, or Greasemonkey/Tampermonkey scripts.
  • Some explicitly dislike the quality of much AI content: repetitive “vibe-coded” Show HNs, shallow business/opinion pieces, or AI being shoehorned into every thread.
  • Others report despair or anxiety: AI hype, job-replacement talk, and focus on ad-tech vs real-world problems make the site emotionally draining.

Tools and LLM-Based Filtering

  • Multiple user-created tools/extensions are discussed:
    • Simple keyword-filter frontends to HN (e.g., hide “ai”, “llm”, “agentic”).
    • More sophisticated reskins that re-rank HN using an LLM and a user “profile”.
    • Suggestions for browser extensions or local LLM “user agents” that rewrite the DOM of any site to match user preferences.
  • Irony is noted: people use AI/LLMs to filter out AI content, and simple keyword filters miss many AI stories or over-block unrelated ones.

Culture, Moderation & Alternatives

  • Some feel HN has become more flamey and echo-chamber-like, with extreme hype vs extreme pessimism in AI threads.
  • Moderation aims to keep “major ongoing topics” high-quality by downweighting low-value AI follow-ups, but can’t fully prevent fatigue.
  • Alternative communities (lobste.rs, custom HN clones, RSS workflows) are raised; others argue starting or joining a different site is preferable to reshaping HN.
  • A recurring meta-point: topic fatigue is inevitable on any popularity-based feed; the practical solution is better personal filtering rather than structurally excluding trending domains like AI.