The Dead Internet Theory

Source: The Dead Internet Theory, Aperture, 13:01, uploaded 2024-06-26, playlist index 970.

Aperture opens with a blunt claim: the internet is dead and its users helped kill it. Bots fill social platforms with misinformation, AI generates pages that people read, and the material produced by machines returns to the systems that produce more of it. The result feels empty because the person scrolling can no longer tell how much of the network comes from another person.

The video asks how the internet reached this state and whether people can still bring back the part that made it worth having. Its answer moves through the internet’s military origins, the concentration of online power, the market for artificial engagement, the rise of generative AI and the loss of human authorship. The dead internet theory supplies the most extreme version of the story. Aperture’s own conclusion is narrower: the web may remain alive, yet its systems are making genuine human contact harder to find.

From military network to public promise

The internet begins in the video’s history as a Cold War technology. American researchers needed a way to share data and programs across a dispersed network during the competition between the United States and the Soviet Union. Aperture places the earliest work in the early 1960s, when two universities exchanged simple messages. The network spread through phone lines during the 1970s. In 1983, 500 computer hosts were connected. Five years later, the source says, that number had reached 100,000 across more countries.

The World Wide Web changed the scale of the idea. A browser could open a network of hypertext documents, which moved the system beyond scientists and government researchers and into public life by the middle of the 1990s. The internet arrived with a political promise attached to it. Information could move freely, people could publish without a media company, and national borders could lose some of their force. Anyone could become a publisher.

The source treats this promise as a real part of the internet’s history, even though it soon met the same pressure that had shaped other important technologies. Greed brought consolidation. Internet service providers merged into a small group, and smaller firms were bought or pushed out. Aperture compares the process to radio, where thousands of companies had been absorbed by a smaller number of owners.

The video gives the concentration in page views as a measure of the change. The ten largest websites accounted for 31 percent of United States page views in 2001, 40 percent in 2006 and 75 percent in 2010. Smartphones then moved much of the web into semi-closed applications. In 2010, applications accounted for less than a quarter of internet traffic. Aperture says that figure has since reached 90 percent.

The applications made advertising easier to target because they kept more of the user’s activity inside one company. The five companies that survived the dot-com crash, which the video names as Google, Amazon, Apple, Facebook and Microsoft, became difficult to avoid through advertising, online shopping, cloud hosting and the platforms themselves. The open web became a set of privately managed rooms, each one recording what its visitors did inside it.

Attention as a market

The next turn comes through attention. Aperture says that the internet has more than five billion users and that the average user spends six hours online each day. The caption gives the first figure as “nearly 52 billion”, which cannot be correct and needs checking against the source. The larger point is clear in the video’s account: online attention has become a market, and views, likes and reviews can be bought.

The source describes a tiered marketplace in which a few clicks and a small payment can produce thousands of Spotify streams, Instagram likes or YouTube views. Companies sell this service through bots because paying people to click on posts by hand would cost too much. A bot targets a post, creates artificial engagement, raises its apparent importance and draws more attention to it.

Platforms try to remove these accounts. Aperture says Meta deactivated more than 27 billion fake accounts over the previous six years and still could not remove every bot. The number is presented without a source, yet it serves the video’s point about scale. The system depends on signals that can be manufactured, then treats those signals as evidence that something deserves attention.

The dead internet theory first appeared in 2019, according to the video. Its original poster believed that the web had changed around 2016, when automated content had become common enough to make human interaction feel rare. The image is a dystopian ghost town populated by simulated actors. A person drives alone through it and passes automated vehicles on the road.

The theory also describes a change in texture. Images, conversations and memes begin to resemble one another. Forms that once felt strange or exciting become sanitised. The theory’s more conspiratorial version says that this sameness is carefully managed to control culture and political power. Aperture keeps that part at a distance. The feeling of emptiness is easier to recognise than the claim that one hidden group controls every part of it.

ChatGPT and the plausibility of the theory

The theory remained marginal until the public release of ChatGPT on 30 November 2022. Aperture describes the system as a large language model that processes huge datasets and forms relationships between ideas. Computers had exceeded people at narrow tasks such as chess and mathematical calculation for decades. The old question was whether a machine could reach general intelligence.

The video turns to Alan Turing’s 1950 question about machine thought and the Turing test. A machine would pass the test by persuading a person that they were speaking with another person. Aperture presents GPT-4 as one of the first systems to pass it. From this point in the video, the dead internet theory becomes a likely future rather than a fringe conspiracy, since the distinction between a human post and an AI-generated one may become difficult to make.

The source cites 49.6 percent of all internet traffic in 2023 as bot traffic. It then points to search engines adding AI answers to their services. The examples are deliberately absurd. Google’s Gemini reportedly told users that rocks provide vitamins, that astronauts who landed on the Moon encountered cats and that a backpack works as a parachute. Aperture uses these failures to show what happens when a system tries to fill gaps in a huge body of compressed information.

The video describes generative AI as a system that identifies important components in its training data and interpolates missing pieces. Creative work therefore comes from compositions and altered versions of material made by people. Hundreds of artists, the source says, signed an open letter organised by the nonprofit Artists Rights Alliance to prevent AI from undermining their work.

The problem extends beyond authorship. The models have no reliable way to judge whether a source deserves trust. A respected academic journal and a meme written by a teenager for Reddit attention can become neighbouring items in the same body of training material. Aperture gives the example of a person who built a fake website to trap crawlers, the automated programs that download web pages for search engines. The site owners discovered that a GPT bot was collecting around three million pages from their sites each day.

That example leads to the video’s most concrete version of the dead internet problem. If AI-generated material already fills parts of the web, later models will train on the output of earlier models. Search results then carry more synthetic material, and trustworthy sources become harder to find. The source calls this a cycle in which each generation contains less of the human work that entered the system at the start.

Algorithms as cultural authority

Aperture then changes the scale of the argument. Digital robots began as tools, like a hammer or a calculator. The word “robot” comes from the Czech word robot, which the video traces to a 1920 science-fiction play and translates as “slave”. Online, the relation between tool and user has blurred. Artificial systems decide which media succeeds, and advertisers pay professionals to guess how their opaque ranking systems work.

The platforms process large datasets to build profiles of users as consumers. A single click can fill a feed with one subject, which pulls the person toward more consumption. The system rewards imitation because copying a successful format gives a creator a better chance of surviving inside the ranking system. Aperture uses MrBeast and Jimmy Donaldson as an example. Once a creator has worked out what the algorithm rewards, other creators have an incentive to copy the result.

The source connects this imitation to Friedrich Nietzsche’s account of the herd. People follow the herd for protection and avoid showing their own view because authenticity brings risk. The result becomes a kind of nihilism. The algorithm circulates content without a shared sense of meaning, while the people using it repeat the patterns that already receive attention.

This leaves an unresolved question about cultural control. Humans still give the systems their instructions and feed them their material. The systems then decide which human work becomes visible and shape the habits that produce more work. Aperture asks whether people still direct the technology or whether technology has become the master that organises the culture around it.

A weaker claim than a conspiracy

The final section brings the argument back to the scale of daily life. The commercial internet, a small group of dominant companies and fast generative tools have created a system that can exploit human weakness. Bots provide a shortcut for business growth. AI offers a shortcut for making content. Ranking systems feed an appetite for short videos because time spent online remains good for the companies that sell attention.

Aperture asks whether the internet has become a constructed reality. It does not claim that the world has reached the complete simulation described by the dead internet theory. It says that the spread of AI has shaken people’s sense of what counts as real online. The warning is practical. People need to leave the herd, admit that the current systems fail and give human life greater weight than profit.

The source’s conclusion rests on a tension it never resolves. The internet remains a place where people can publish, meet and exchange ideas. The same infrastructure also rewards concentrated ownership, artificial engagement, fast imitation and content that keeps people consuming. The theory works as an account of total control only when its evidence is stretched. It works as a description of the conditions that make human work harder to recognise.

Evidence limits

The video names figures about page views, app traffic, bot activity, fake accounts and internet use without giving studies or links. Its claim that GPT-4 passed the Turing test is presented as fact, even though the video supplies no test details. The Gemini examples, the three-million-page crawler story and the Artists Rights Alliance letter also remain source-owned claims here.

The historical account compresses complicated developments into a clean line from Cold War research to platform control. Its account of AI training, hallucinations and future model collapse works as a warning, while the captions give no technical sources for the mechanism. The dead internet theory itself combines an observable change in online incentives with a much larger claim about hidden control. Aperture leaves the hidden-control claim unproved and ends with the smaller demand that people protect human authorship and attention.

Further reading / references

• Alan Turing’s 1950 question about machine thought, used in the video’s account of the Turing test.
• Friedrich Nietzsche’s writing on the herd, invoked as the source’s explanation for imitation and conformity.
• Rossum’s Universal Robots, the 1920 science-fiction play that Aperture uses to explain the word “robot”.
• The Artists Rights Alliance open letter against AI systems undermining artists, mentioned in the video without a link.