You can't trust the internet anymore

Shift in trust and failure modes

• Commenters argue the big change isn’t that the internet suddenly became untrustworthy, but that the failure mode has shifted.
• Before: you’d occasionally hit bad info or an intentional hoax.
• Now: you’re overwhelmed by low-effort regurgitation and AI “slop,” which is then scraped and fed back into models, degrading everything downstream.
• Many describe the web as “burned” or “dead internet” becoming reality; the ratio of humans to bots (readers and writers) feels near zero.

Causes: AI slop, SEO, and incentives

• AI makes it trivial to mass-generate plausible-looking content for ultra-niche topics (obscure games, old CPUs, new game mechanics) purely for ads or affiliate links.
• This continues older SEO spam practices, but with much lower cost and far greater scale.
• Google is blamed for tolerating and effectively partnering with SEO content farms instead of suppressing them, creating a race-to-the-bottom prisoners’ dilemma.
• Some see this as poisoning future AI training data, making current models a relatively “clean” historical snapshot.

Debates on who/what is to blame

• Disagreement over how much to blame specific countries vs global incentives: some point to spam/scam hubs; others counter that this is long-standing and overemphasized.
• Several insist the web was never high-trust; what changed is (1) everyone now uses it as primary information source, and (2) the cost of bullshit collapsed.
• There’s a side debate over whether prediction markets materially incentivize misinformation; some think volumes are too small, others see potential for subtle manipulation.

Language, “enshittification,” and platform decay

• Extended argument over the term “enshittification”: originally a specific pattern of platform decay vs its now-popular generic use (“things got worse”).
• Some want to preserve the precise meaning; others accept rapid semantic drift as inevitable, especially online.

Proposed solutions: trust mechanisms and enclaves

• Ideas include human-certified, digitally signed authorship; web-of-trust systems; invite-only communities; “vetted webrings”; local mesh networks; alternative protocols like Gemini; and selective archiving of consented sites.
• Skeptics note these can be infiltrated, may simply rebuild the same problems at smaller scale, or solve issues at the wrong layer.
• There’s nostalgia for smaller, reputation-based communities (BBSes, local forums) and current analogues (private forums, small Fediverse instances).

Outlook: inevitability vs opportunity

• Some are fatalistic: infinite AI content will soon be indistinguishable, and maybe people will stop caring.
• Others see a business and social opportunity in tools or spaces that reliably surface “real,” human-authored, non-astroturfed content.