Ask HN: Is it time for HN to implement a form of captcha?

Hacker News Discussion ↗

Perceived AI/Bot Problem on HN

  • OP asserts “tremendous” bot/AI attempts; several readers say they rarely or never see AI slop, even with “showdead” enabled.
  • Many commenters think most low-quality AI content comes from real users pasting LLM output, not autonomous bots.
  • Some argue this is not yet a problem requiring new mechanisms; others see it as an inevitable worsening trend.

Effectiveness and Costs of CAPTCHAs

  • Strong theme: CAPTCHAs mainly annoy legitimate users while determined bots bypass them (LLMs, cheap human-solving services, headless browsers).
  • Others push back: even imperfect CAPTCHAs can cut a large share of garbage traffic; “pretty good” filtering is still valuable.
  • General agreement that on a high-value target like HN, serious abusers would quickly adapt.
  • Several stress CAPTCHAs are better seen as rate-limiting than as true bot prevention.

Existing HN Measures & UX Concerns

  • HN already uses ReCAPTCHA for account creation and rate-limits new accounts.
  • More challenges could break custom RSS, third‑party apps, alternative browsers, and generally worsen the experience.
  • Some say seeing “find the bike” CAPTCHAs here would symbolize the web’s broader decline.

Alternative Technical Approaches

  • Idea: a tool that records comment-edit history and lets readers replay composition to distinguish pasted slop from real writing; others note bots or scripts could simulate this, and there are privacy/UX issues.
  • Suggestions: behavior-based signals (age, karma, posting patterns), honeypots/hidden content to trap bots, or client apps with built-in spam filters.
  • Several note that robust anonymous “proof of human uniqueness” appears unsolved and may be inherently hard, despite ZKPs and digital ID efforts.

Moderation, Shadowbans, and Voting

  • Debate over shadowbans and silent vote-disabling: some defend current moderation and say bans target repeat bad actors; others report non-transparent filters and want clearer documentation or periodic amnesty.

User-Level Filtering and Block Features

  • Many want per-user, per-domain, or keyword blocking, comparing it to SponsorBlock.
  • Others argue this contradicts HN’s “single global pool” experiment and risks echo chambers and “missing stair” dynamics.
  • Several share practical workarounds using browser extensions and filters.

Identity / “Real Human” Schemes

  • Proposals like work-email verification or credit-score–based trust are widely criticized as dystopian, exclusionary, and fragile.

Overall Leaning

  • The dominant view in the thread: adding CAPTCHAs beyond signup is unnecessary and counterproductive; better to rely on moderation, behavior-based heuristics, community flagging, and optional client-side filters.