Stories removed from the Hacker News Front Page, updated in real time

Original Article ↗ Hacker News Discussion ↗

A new tool that tracks stories dropping off the Hacker News front page has prompted a broader look at how the site is curated. Commenters dissect the interplay between user flags, automated systems (like flamewar and dupe detectors), and manual moderator intervention, noting that many removals are for repetition, low-quality or inflammatory content, or off‑topic politics rather than overt censorship. The thread surfaces ongoing tension between those who see this as healthy curation that preserves signal and those who worry about opaque or crowd-driven suppression of certain topics, and it renews calls for more transparency without consensus on how far that should go.

Overall reaction to the removal tracker

  • Many commenters like the tool and say it mostly showcases that moderation works well: most fast-dropped stories look like clickbait, politics, PR, or low-quality repeats.
  • Others see some worthwhile posts among the removals and want explanations or “second chances.”
  • Several people ask for better UX on the tracker (reverse-chron sort, tables, direct links to HN items).

How HN ranking and removals actually work

  • The tracker’s core assumption (“30 → >90 in 1 minute = moderator removal”) is called wrong.
  • Stories drop quickly due to:
    • User flags (which also downrank long before “[flagged]” appears).
    • Automated penalties (flamewar/“overheated discussion” detector, anti-abuse, age-based downweighting, duplicate/follow‑up detection).
    • Occasional explicit moderator downweights, especially for dupes or low-substance ragebait.
  • HN is explicitly curated; front page is not purely vote-driven. A single moderator plus software and users shape it.

Flags, downvotes, and user responsibility

  • Many users treat “flag” as a de facto downvote on stories they dislike, not just guideline violations.
  • Others insist they only flag spam/abuse and see broad flagging as misuse.
  • High-karma users can lose flagging power for abuse; some can “vouch” for dead/flagged items.
  • Confusion remains about why there’s no explicit story downvote and what “flag” is really for.

Politics, COVID, Gaza, and other flashpoints

  • Major friction around political and COVID stories:
    • Some say vital topics (ICJ/Gaza, COVID measures) are being suppressed by partisan flagging.
    • Others argue these threads nearly always devolve into low‑signal flamewars and are off‑topic per guidelines.
  • Similar dynamics noted for Musk, crypto, gender, and “culture war” topics.
  • Guiding idea from moderation side: favor “significant new information” and avoid repetitive, high‑heat/low‑curiosity debates.

LLM/AI fatigue and topic repetition

  • Despite huge interest, many users report being tired of constant LLM stories and generic “$celebrity says $thing about AI” items.
  • Repetition is framed as “enemy of curiosity”; follow‑up and copycat posts get downweighted even if individually fine.

Transparency, trust, and gaming

  • Some call for public moderation logs and visible story‑penalty history; others warn this would invite gaming and create a bureaucratic burden.
  • Moderation stance:
    • Full logs wouldn’t convince the distrustful and would be weaponized.
    • Instead, answer specific questions case‑by‑case and rely on community trust.
  • A recurring tension: desire for openness vs preserving a front page that isn’t dominated by sensational or meta drama.