The Dead Internet Theory

Let me level with you: the Dead Internet Theory isn’t just another Reddit campfire story. It’s a vibe many of us feel search results that look cloned, timelines full of the same recycled sludge, comment sections that read like they were authored by a bored hive mind. The claim is bold: since around 2016–2017, the internet has been flooded by bots, AI-generated content, and algorithmic puppet-strings, leaving genuine human conversation gasping for air. The idea started bubbling up on fringe boards and broke into the mainstream in 2021 The Atlantic even ran a piece about it, calling it “ridiculous, but possibly not that ridiculous.”

What the theory actually says (minus the spooky music)

At its core, DIT says two things:

1. Bots + AI now generate a huge chunk of what you see, and algorithms decide who sees what and when.
2. Some of this is coordinated by corporations, states, or just opportunists nudging what trends and what disappears.

Crucially, bits of this are verifiable, even if the grand conspiracy part is… a stretch. For instance, bot traffic is massive. Imperva’s annual reports show automated traffic hovering around half of the web, with some recent analyses saying it even surpassed human activity in 2025. That doesn’t prove the whole web is “fake,” but it does mean a lot of what hits servers isn’t you or me.

On social, the numbers get wilder. Pew Research found back in 2018 that bots were responsible for about two-thirds of the links to popular sites on Twitter, X—during the period they studied. That’s links, not all posts, but it shows how automated accounts steer attention.

And the web isn’t just being filled it’s disappearing. Pew’s 2024 “digital decay” report: 38% of pages that existed in 2013 were gone by 2023. That’s the famous “link rot” everyone keeps bumping into dead citations, lost context, broken history.

Exhibit A: “Shrimp Jesus” and the era of AI slop

If you wanted a single weird emblem for the 2024-era internet, it’s the flood of AI-generated kitsch: “Shrimp Jesus,” “Crab Jesus,” lettuce babies, impossible cakes stuff engineered not to inform, but to trip engagement wires and farm the gullible. Researchers and journalists traced how these images were blasted at scale on Facebook for cheap reach, ad pennies, and audience-harvesting. It’s absurd on purpose because absurdity spreads.

Exhibit B: “News” sites with nobody home

Then there’s the ghost-news economy. NewsGuard has tracked 1,200+ AI-generated “news” sites cranking out synthetic articles with minimal editorial oversight designed to look legit, monetize clicks, and sometimes launder propaganda. This isn’t a theory; it’s an observable industry.

Where the theory overshoots (and why that matters)

DIT often leaps from “lots of automation exists” to “most of the web is fake.” That’s where you should pump the brakes. A 2025 Nature study analyzing global social chatter found roughly 20% bots vs. 80% humans still a ton of automation, but not total zombification. Also, not all non-human traffic equals deception: crawlers, uptime monitors, and benign automations count as “bots” too. The nuance matters.

Why the internet feels dead anyway

• Algorithmic sameness. When recommendation engines reward click-alike content, humans start posting like bots to get seen. That “sterile, repetitive” feel? It’s not just you.
• SEO spam & content farms. Low-effort pages crowd out signal; AI makes it cheap to flood the zone. (You’ve felt this every time a product search returns 30 near-identical “ultimate guides.”)
• Digital decay. Half-remembered posts vanish, citations rot, and the surviving web tilts toward what’s profitable to keep alive. Over time, that makes the commons feel hollow.

Two quick, concrete examples you can try

1. The “AI slop” test. Search a trending meme + “Facebook” notice the wave of AI composites, uncanny hands, and off-kilter details. Count how many posts lead to pages with generic names and weird, unrelated captions. That pattern is the engagement-farm playbook documented by journalists in 2024.
2. The “vanishing link” test. Grab a five-year-old article and click every outbound link. Track how many 404. Compare to a newer piece. You’ll likely see the decay Pew measured at scale.

So… is the internet “dead”?

Not dead but sick. The measurable stuff (bot traffic, AI-spam, link rot) explains the “empty room” sensation without positing a monolithic puppet-master. The better mental model is this: incentives are optimized for scale and cheapness, so systems reward whatever can be mass-produced bots, slop, and yes, humans behaving like bots. The fix isn’t tinfoil; it’s better incentives, smaller communities, real verification, and curation.