Wikipedia Deep Dive

Content farm

8 min read

In 2016, a small town in North Macedonia called Veles became ground zero for one of the strangest cottage industries of the internet age. Teenagers and young adults, many unemployed or underemployed, discovered they could make thousands of dollars a month by writing fake news articles about the American presidential election. They didn't care who won. They just knew that American Facebook users were worth four times more in advertising revenue than the global average, and that sensational headlines about Hillary Clinton and Donald Trump got shared like wildfire.

Over 140 websites from this single town pumped out fabricated stories designed to look like legitimate American news sources. Some of these articles reached hundreds of thousands of people.

This is what a content farm looks like at its most brazen. But the phenomenon runs much deeper and wider than election interference—it's become one of the defining features of how information flows online, and it's only accelerating.

The Economics of Attention

To understand content farms, you first need to understand how most websites make money. The dominant business model of the internet is advertising. When you visit a webpage, you see ads alongside the content. The website owner gets paid based on how many people view those ads or click on them. More visitors means more money.

This creates a powerful incentive: get as many eyeballs on your pages as possible, by any means necessary.

A content farm is an organization built entirely around this logic. The goal isn't to inform, entertain, or create anything of lasting value. The goal is to generate massive quantities of content that will rank well in search engines and attract clicks. Think of it as the digital equivalent of a factory farm—industrial-scale production optimized purely for output and cost efficiency, with little regard for quality.

The term "content mill" captures this perfectly. These operations churn out material like a flour mill grinding wheat, except the product is articles, videos, and social media posts designed to capture fleeting attention.

How the Sausage Gets Made

The mechanics are straightforward. Content farms hire large numbers of freelance writers, often paying them almost nothing. Three dollars and fifty cents per article is a common rate. At that price point, nobody is doing deep research or careful fact-checking. Writers crank out dozens of pieces per day on topics they know little about, racing to meet quotas.

To maximize profit margins, many content farms have historically outsourced this work to writers in developing countries where such wages, meager as they are, stretch further. The quality suffers, but the economics work—at scale, even pennies per click add up.

The numbers are staggering. In 2009, a company called Demand Media, which owned the website eHow, was publishing one million pieces of content every month. To put that in perspective, that's the equivalent of four complete English-language Wikipedias every single year, except Wikipedia is built by volunteers striving for accuracy, while eHow was built by underpaid contractors optimizing for search rankings.

Another major player was Associated Content, which Yahoo acquired in 2010 for ninety million dollars. They rebranded it as Yahoo Voices before shutting it down entirely in 2014, an outcome that hints at the unsustainability of the model—or at least its eventual collision with quality expectations.

The Rise of the Machines

Then came artificial intelligence.

The release of large language models—think ChatGPT and its competitors—transformed the content farm industry almost overnight. Why pay humans three dollars and fifty cents per article when you can generate hundreds of articles per day using AI, with virtually no marginal cost?

By 2023, a media watchdog organization called NewsGuard had identified over 140 major brands unknowingly advertising on AI-driven content farms. These sites produce articles with minimal human oversight, creating an assembly line of text that often contains errors, fabrications, and what researchers call "hallucinations"—AI's tendency to confidently state things that are completely made up.

The scale is breathtaking. A single operator can now run what used to require a building full of underpaid writers.

The Problem of Polluted Information

Critics have compared content farms to fast food restaurants. Just as fast food prioritizes cheap calories over nutrition, "fast content" prioritizes cheap clicks over accuracy or value. And just as a diet of fast food damages your health, a steady diet of fast content damages your understanding of the world.

The issues go beyond mere low quality. Content farms have become vectors for misinformation—conspiracy theories, fake product reviews, and outright fabrications dressed up to look like legitimate journalism. The Macedonian fake news factories weren't anomalies. They were a logical extension of an economic system that rewards attention without caring how that attention is obtained.

The situation gets more dystopian when you consider a phenomenon researchers call "AI cannibalism." Large language models are trained on text from the internet. As AI-generated content floods the web, these models increasingly train on their own output. It's like making photocopies of photocopies—each generation drifts further from the original, accumulating errors and distortions. A content farm using AI that's been trained on AI-generated content creates a feedback loop of degradation.

When the Fake Spills Into the Real

Perhaps most alarming is how content farm tactics have leaked into high-stakes environments where accuracy genuinely matters.

Lawyers have been caught submitting legal briefs containing citations to cases that don't exist—hallucinations generated by AI tools they used carelessly. Court proceedings have been disrupted. In one remarkable case, a New York man attempted to use an AI avatar to represent himself in court, raising profound questions about authenticity, bias, and the reliability of machine-generated arguments in settings where the stakes couldn't be higher.

Even seemingly trivial examples create ripples of distrust. When The Chicago Sun-Times published a summer reading list written by AI, the backlash wasn't really about book recommendations. It was about the erosion of trust in institutions that we rely on to filter and validate information. If you can't trust a newspaper to actually read books before recommending them, what can you trust?

The Bot Economy

Content farms don't just produce fake articles. They produce fake engagement.

Bot armies create inauthentic reviews of products, pumping up ratings and drowning out legitimate customer feedback. They generate manufactured website traffic that makes worthless advertising space appear valuable. Companies using automated bidding systems end up paying premium prices for ads that no human will ever see.

The estimated cost of this fraud is thirteen billion dollars annually. That's money siphoned out of the legitimate economy and into the pockets of operators running elaborate digital shell games.

Fighting Back

The major platforms haven't been entirely passive. In 2011, Google rolled out what it called the Panda update—an algorithm change designed to push low-quality websites down in search rankings. The name was an internal codename, but it stuck. Panda was explicitly aimed at content farms, and it did cause real damage to some of the worst offenders.

But the arms race continues. Content farms adapt their tactics, and new ones emerge to replace those that fail. In 2024, the search engine DuckDuckGo implemented measures specifically targeting AI-driven content farm sites, suggesting the problem has evolved rather than disappeared.

Advertising exchanges—the platforms that match advertisers with websites—generally have policies against content farms. But NewsGuard's research found that enforcement is rare. Google, the dominant player in online advertising, was found to be "overwhelmingly more likely" to serve ads from content farms compared to other platforms. When the incentives all point toward volume over quality, good intentions tend to collapse.

The Bigger Picture

Content farms are symptoms of a deeper pathology in how we've built the internet's economy. When attention is the only currency that matters, and when algorithms determine what gets seen, the inevitable result is a race to the bottom. The content that spreads fastest isn't the most true, most useful, or most beautiful—it's the most provocative, the most emotionally triggering, the most optimized for machine distribution.

The term "enshittification," coined by writer Cory Doctorow, captures this trajectory. Platforms start out providing genuine value to attract users, then gradually degrade that value to extract maximum profit. Content farms are both cause and consequence of this dynamic—they exploit the system's weaknesses while simultaneously making those weaknesses worse.

Related concepts have proliferated to describe various aspects of this degradation. "AI slop" refers to the flood of low-quality AI-generated content. "Brain rot" describes the cognitive effects of consuming too much of it. "Spamdexing" and "SEO spam" are technical terms for the tactics used to game search rankings. Each term tries to name a piece of the elephant, but they're all touching the same beast.

Understanding content farms means understanding that the information environment we swim in every day isn't neutral. It's shaped by economic incentives that often have nothing to do with truth, clarity, or human flourishing. The phantom writers—whether underpaid humans in distant countries or language models hallucinating confident nonsense—are just the workforce. The real authors are the business models that brought them into being.