Narrator: In 2016, the world watched as nearly every political poll and expert prediction about the U.S. presidential election proved spectacularly wrong. Donald Trump’s victory seemed to come from nowhere, defying all conventional wisdom. But what if the clues were there all along, not in what people told pollsters, but in what they typed into a simple, private, white search box? What if the anxieties, resentments, and hidden biases that swung the election were being confessed, by the millions, to Google?

This is the provocative world explored in Seth Stephens-Davidowitz's book, Everybody Lies: Big Data, New Data, and What the Internet Can Tell Us About Who We Really Are. The book argues that the digital trails we leave behind—our searches, our clicks, our online behavior—have created the most important dataset ever collected on the human psyche. It’s a digital truth serum, revealing what we truly think, fear, and desire when we believe no one is watching.

Narrator: The foundational premise of the book is that people lie. They lie to friends, to doctors, to surveys, and even to themselves. When a survey asks how often people use condoms, men and women give wildly inconsistent answers, and both report numbers that would mean over a billion condoms are used annually in the U.S. Yet, sales data from Nielsen shows that fewer than 600 million are actually sold. The numbers don't add up because people aren't telling the truth.

But in the privacy of a Google search, this social desirability bias vanishes. Stephens-Davidowitz reveals that searches for "sexless marriage" are 3.5 times more common than searches for "unhappy marriage," suggesting a specific, hidden anxiety that surveys fail to capture. People confess their deepest insecurities and prejudices to the search bar. The author found that in the hours after the San Bernardino terrorist attack was linked to Muslim perpetrators, the top search in California containing the word "Muslims" was "kill Muslims." This raw, unfiltered data provides a stark, and often uncomfortable, look at the human psyche, a level of honesty that traditional methods could never achieve.

Narrator: The revolution isn't just about analyzing what people type; it's about reimagining what data can be. Stephens-Davidowitz tells the incredible story of Jeff Seder, a data-driven horse evaluator. For decades, the horse racing industry relied on the gut instincts of trainers who looked at a horse's pedigree and physical build. Seder took a different approach. He began collecting unconventional data, using a portable ultrasound to measure the internal organs of young horses.

He discovered that one variable was a massive predictor of a horse's success: the size of its left ventricle. A larger heart meant more oxygenated blood could be pumped to the muscles. In 2013, at a prestigious auction, Seder’s firm, EQB, was hired by owner Ahmed Zayat. After analyzing all the horses, Seder’s team made a desperate plea: Zayat could not sell horse No. 85. Its left ventricle was exceptionally large, a once-in-a-generation heart. Zayat, trusting the data, bought back his own horse—an almost unheard-of move. That horse was later named American Pharoah, and it went on to win the Triple Crown, the first horse to do so in 37 years. Seder found game-changing data where no one else was looking.

Narrator: Traditional surveys struggle to analyze small, specific groups of people. With massive datasets, however, researchers can "zoom in" on tiny segments of the population to uncover profound truths. For instance, how do we form lifelong allegiances to sports teams or political parties?

By analyzing the Facebook "likes" of millions of baseball fans, Stephens-Davidowitz found a stunning pattern. For men, the single most important factor in determining their favorite team is which team won a championship when they were around eight years old. A World Series win when a boy is eight is eight times more impactful than a win when he is twenty. Similarly, a study of sixty years of voting data found that our political identities are largely forged between the ages of 14 and 24. The popularity of the president during these formative years creates a lifelong political leaning. Someone who turned 18 during Ronald Reagan's popular presidency is far more likely to remain a Republican for life. These critical windows, invisible in small datasets, become crystal clear when we can zoom in.

Narrator: For centuries, science has struggled to prove causation. Does an ad cause you to buy a product, or do companies just advertise products people already want? The digital world makes answering these questions easier than ever through A/B testing—a simple, randomized controlled experiment.

During Barack Obama’s 2008 presidential campaign, his digital team wanted to optimize their website's homepage to get more email sign-ups and donations. They tested 24 different combinations of photos and button text. Would a picture of Obama's family perform better than a solo shot? Should the button say "Join Us," "Sign Up," or "Learn More"? They randomly showed different versions to website visitors and measured the results. The winning combination—a family photo with a "Learn More" button—increased sign-ups by a staggering 40 percent. This simple experiment was estimated to have netted the campaign an extra $60 million in donations. A/B testing turns the internet into a massive, ongoing laboratory, allowing us to replace guesswork with hard evidence.

Narrator: For all its power, Big Data has serious limitations. Stephens-Davidowitz recounts his own attempt, with former Treasury Secretary Larry Summers, to use Google search data to predict the stock market. They theorized that searches for "unemployment benefits" might predict economic downturns or that searches for "buy gold" might signal market anxiety. Despite their efforts, they failed. The market was too complex, and any simple patterns were likely already being exploited by hedge funds with far more resources.

This illustrates a key danger in data science: the "curse of dimensionality." When you test thousands of variables against a dataset, some will appear to be correlated purely by chance. A study once claimed to predict the stock market based on how "calm" Twitter felt six days earlier. A hedge fund was started based on this finding, and it failed within a month. The "calmness" signal was the data equivalent of a random lucky coin. Without rigorous methods and a healthy dose of skepticism, Big Data can easily lead us astray, showing us patterns that aren't really there.

Narrator: The single most important takeaway from Everybody Lies is that we are at the dawn of a new scientific revolution. For the first time, the study of humanity is becoming a true, data-driven science. The digital crumbs we leave behind have given us a "cerebroscope"—a window into the human mind—that allows us to understand ourselves with unprecedented honesty and scale. We can now test theories, uncover hidden biases, and measure the world in ways that were previously unimaginable.

But this new power comes with a profound responsibility. As we learn to predict human behavior with greater accuracy, from loan defaults to gambling habits, we must ask ourselves what we should do with this knowledge. How do we use this digital truth serum not just to understand the world, but to make it better, fairer, and more humane? The data is there, waiting. The most important questions are now up to us.