I'm Being Trained and So Are You - Barry's Economics

In 1938, psychologist B.F. Skinner placed a rat in a box with a lever. When pressed, the lever dispensed food. But Skinner tested multiple reward schedules: sometimes food came every press, sometimes every fifth press, sometimes randomly. The result was unequivocal: random rewards produced the most obsessive, compulsive lever-pressing behavior. Skinner called this a «variable ratio reinforcement schedule», and he was reportedly disturbed by its effectiveness.

When the rat never knew which press would deliver food, it kept pressing endlessly, frantically, unable to stop — even when food stopped coming entirely. The unpredictability made the behavior impossible to extinguish. Skinner spent years worrying about the implications for human society, and his concerns were well-founded.

Today, that same variable ratio schedule powers every major social media platform. Instagram, TikTok, and YouTube aren't platforms — they're Skinner boxes with better graphic design and worse privacy policies. Sometimes a post gets 2,000 views, sometimes 400,000, and you never know why. You just keep pressing the lever, keep posting, keep scrolling, searching for that next unpredictable hit of dopamine.

Nicholas Perry was a classically trained violinist who started a YouTube channel in 2014 sharing vegan recipes and gentle music. Nobody watched. Then, as an experiment, he tried mukbang — eating massive quantities of food on camera. That first video got 50,000 views in two weeks, more than everything he'd done before combined. The algorithm had spoken: put down the violin and eat a family bucket of KFC while having a breakdown.

Over eight years, Perry gained over 250 pounds. He developed sleep apnea, fractured his ribs, became disabled, and bought a mobility scooter. He filmed himself crying and eating 20,000-calorie meals three times a day across five separate channels. His direct quote should haunt every creator: «They like when I'm sick. They like when I'm upset. And they like when I'm hyper. So I just give them that.»

He knew. He could describe the mechanism in plain English. And he still couldn't stop. Because knowing you're in a Skinner box doesn't get you out of the Skinner box — the reinforcement schedule works regardless of intelligence, insight, or self-awareness. That's what made Skinner so terrified of his own research.

Most people think dopamine is a pleasure chemical, but its primary function is to signal change — to help you detect novelty and shifts in your environment. That detection system works by contrast: you notice a small dopamine increase against a low baseline, like spotting a candle in a dark room. But when your baseline is already elevated from constant notifications and reward signals, small changes become invisible.

Phil Reed at Swansea University calls this «dopamine overdose». Research shows that artificially elevated dopamine during discrimination tasks actually reduces your ability to learn new information. Too much dopamine makes you worse at noticing that something has changed — including noticing that you are changing. For a creator in the middle of a viral spike, dopamine hits come from every direction: views, subscribers, comments, shares. Your elevated baseline impairs the very system that would normally flag drift in your content, tone, or beliefs.

This means the creators most at risk of being changed by the algorithm are precisely those experiencing the most success. The bigger the spike, the harder it is to see yourself clearly. The algorithm rewards certain content, the creator makes more of it, the dopamine reward blinds them to their own drift, and the audience reinforces the direction because that's what gets served. At no point does anyone feel manipulated — it all feels like authentic choice, which is exactly what a perfectly designed conditioning system would look like from the inside.