Dead Internet Theory

What dead internet theory is

Dead internet theory is the claim that most of the internet’s content and activity is now produced by bots and artificial intelligence rather than by real people, and, in its stronger form, that this hollowing-out is a deliberate effort by corporations or governments to manipulate and control us. The grounded version notices that the web feels increasingly automated. The conspiracy version insists that someone planned it that way.

The idea took shape in 2021, in a post on the niche forum Agora Road’s Macintosh Cafe titled “Dead Internet Theory: Most of the Internet is Fake”. Its pseudonymous author, building on earlier murmurings from imageboards, argued that the internet had quietly “died” around 2016 or 2017, and that what remained was a wasteland of automated posts and recycled content with the real humans long gone. The Atlantic gave the idea mainstream attention later that year, and it has circulated ever since.

It is, strictly, a conspiracy theory, and its strongest claims are not supported by evidence. But it is worth taking seriously rather than laughing off, because it is a rare case of a conspiracy theory built on top of something real. Pulling the two apart is a useful exercise in its own right.

The grain of truth

A good deal of what dead internet theory describes is happening. A large and rising share of internet traffic is automated rather than human. Estimates vary between sources and methods, but by some counts the automated share passed half of all web traffic in 2024, the first time it had outweighed people. Bots crawl, scrape, post and click around the clock, and a meaningful slice of them exist to commit fraud or game engagement.

The content itself has changed too. Content farms churn out articles built for search engines rather than readers. Fake followers, fake likes and fake reviews are bought and sold. Coordinated accounts push narratives in a practice known as astroturfing. On top of all this sits the rising tide of AI slop, the cheap machine-made text and imagery now flooding feeds, and the deepfakes that make even people and events seem fabricated. Recommendation systems then funnel everyone towards whatever performs, so different corners of the web start to feel eerily alike.

The lived experience the theory points at is real. Many people genuinely notice the same threads, the same images and the same replies recurring across sites, and feel that something human has drained out of the place. Social bots had already begun swarming platforms like Twitter and Facebook in the mid-2010s, amplifying particular messages at scale, which is part of why believers date the supposed death to around 2016 or 2017. The shift to algorithmic feeds in those same years meant that what people saw was increasingly chosen by machines rather than followed by choice. That feeling is not imaginary, and it deserves an explanation.

Where it overreaches

The trouble starts when the theory leaps from “a lot of this is automated” to “all of this was deliberately staged”. The full version holds that a coordinated authority, often imagined as government and big tech working in concert, killed the organic internet on purpose in order to herd and gaslight the population.

This leap fails for the usual reasons a conspiracy theory fails. It is unfalsifiable: any evidence that real people are still active online is reinterpreted as part of the deception, so nothing could ever count against it. It requires a degree of secret, flawless coordination across rival companies and governments that strains belief. And it mistakes a diffuse, scattered pattern for a single guiding hand. The web does feel more automated. That does not mean anyone sat down and planned for it to feel that way.

Unfalsifiability is the common thread in conspiracy thinking. A claim arranged so that nothing could ever count against it has stopped being a claim about the world and become a closed loop, sealed off from the very evidence that might correct it.

You don’t need a conspiracy

Here is the reframe at the heart of this entry. A dead-feeling internet falls out of ordinary, uncoordinated incentives, with no central planner required.

Follow the money. Advertisers pay for engagement, so platforms optimise relentlessly for it. Engagement can be faked cheaply, so bots and content farms multiply to harvest it. Generative AI has made content nearly free to produce, so AI slop rushes in to fill every gap. Algorithms converge on whatever holds attention, so the attention economy quietly flattens the web into a few proven formats. The slow rot this produces in the platforms themselves even has its own name, enshittification. Each of these is a small, self-interested decision. Stacked together, they produce a web that feels automated and hollow, exactly as the theory describes.

This is emergence: a large-scale pattern thrown up by many small forces pulling the same way, rather than by anyone directing it. And there is a quiet sting in the tail. The conspiracy version is, oddly, the more comforting story, because a plot could in principle be exposed and stopped. An outcome that emerges from the basic incentives of the system is far harder to undo, because there is no villain to remove. You would have to change the incentives themselves.

Why the theory is sticky

If the emergent explanation is the better one, why does the conspiracy version travel so well? Partly because it offers a tidy villain for a real but shapeless unease. We are pattern-seekers, and a single hidden hand is far more satisfying than a dull list of overlapping incentives. A plot gives the feeling a face.

It is also self-reinforcing. The more slop, bot replies and uncanny content you notice, the more the theory feels confirmed, and the more confirmed it feels, the more you go looking. That loop runs on confirmation bias, our habit of collecting what fits a belief while skipping past what does not. A theory that feels truer every time you open your phone is a theory that is hard to put down.

What dead internet theory is not

It helps to be clear about the limits. Dead internet theory is not established fact. The strong claim, that a coordinated conspiracy deliberately killed the internet, is unsupported and built so that no evidence could ever disprove it. The internet is not literally dead, and real people are still very much present on it.

Nor is it a reason to give up. The most damaging response is to conclude that nothing online is real and to retreat into a blanket cynicism where no source, no account and no recording can be trusted. That is the same exhausted state the liar’s dividend feeds on, and it is its own kind of error, every bit as much as naively believing everything you see.

The grounded version is better treated as a media-literacy prompt than a doom prophecy. Assume that a fair amount of what reaches you is automated. Treat raw engagement numbers, likes and follower counts with suspicion. Seek out the human signal, the specific, the odd, the personally accountable, and weight it more heavily than the smooth and the viral. The internet has not died. But a growing share of it is machinery, and learning to tell the living parts from the automated ones is the practical skill the theory, despite itself, points towards.