I saw your HN comment

Cold outreach emails and messages are increasingly being generated by large language models that scrape Hacker News comments, GitHub repos, and blog posts to create personalized-sounding pitches at scale. Many users describe a sharp decline in the quality and trustworthiness of unsolicited contact, saying what used to be a rare source of genuine connection now feels like spam, social engineering, or low-effort grift, with similar effects spilling into workplaces and other platforms. Proposed responses range from technical measures like watermarks and better spam filters to social and architectural changes such as invite-only communities, vouch or reputation systems, disposable emails, and a renewed emphasis on smaller, human-centric online and offline networks.

LLM‑Driven Outreach and Spam

  • Many report a sharp rise in cold emails that reference specific HN comments, GitHub commits, or Show HN posts, then pivot into product pitches, recruiting, “consulting,” or promotion schemes.
  • Much of this is obviously templated or LLM-written; some is sophisticated enough to evade both spam filters and initial human skepticism.
  • Similar behavior is seen on Twitter/X, LinkedIn, email lists, and GitHub: engagement‑bait replies, fake “questions,” and recruiter or “career coach” spam, sometimes shading into phishing or malware attempts.

Emotional Impact and Behavior Changes

  • People describe a loss of joy and trust: where an email about an HN comment used to be a pleasant surprise, it now defaults to suspicion.
  • Several remove emails from profiles, route HN‑related mail to spam, or require specific subject formats to signal humanity.
  • Some still value genuine outreach and accept the extra filtering as a tradeoff; others feel forced to retreat to pseudonyms or closed communities.

Broader Concerns About AI and the Internet

  • Many see LLMs as accelerating an existing trend of “enshittification”: more spam, shallow discourse, and performative expertise at work and online.
  • There is anxiety that a large share of visible content and even workplace communication is now LLM‑mediated, eroding authenticity and making proof of human authorship difficult.
  • Some argue we’re in the “last years” of the internet as it was known, predicting a shift toward in‑person interaction or identity‑verified spaces; others note similar panics have accompanied earlier technologies.

Proposed Mitigations and New Models

  • Technical ideas: text watermarking, better classifiers, captcha/challenge replies, Craigslist‑style email proxies, disposable/aliased addresses, karma‑ or word‑count‑gated contact, and proof‑of‑humanity systems.
  • Community design ideas: invite‑only or vouch‑based networks, web‑of‑trust reputation, small closed forums, or modest paywalls to deter low‑effort spam—tempered by worries about excluding people in poorer regions.
  • Some advocate naming and shaming spammy companies; others doubt shaming will meaningfully deter them.

Divergent Views on Using LLMs

  • A minority sees LLM‑mediated outreach as just another tool; they suggest simply answering and moving on.
  • Many others emphasize effort asymmetry and deception: LLMs let senders fake high effort at near‑zero cost, pushing the filtering burden and emotional cost onto recipients.
  • There is also meta‑critique of people decrying LLM slop while using LLMs themselves (e.g., to analyze their inbox), seen by some as hypocritical and by others as a pragmatic necessity.