Weapons of Math Destruction: Chapter 2: Shell Shocked Summary & Analysis

Quants who work at hedge funds zoom in on tiny patterns, then train algorithms to predict recurring errors and price swings, then place bets on their occurrences. The smallest of patterns can make millions for the first investor who recognizes them—and those patterns will keep raking in money until the pattern ends, or the rest of the market catches on.

While working at Shaw, O’Neil loved the “treasure hunt” component of finding market inefficiencies. At Shaw, her smarts were translating into money—lots of money. But out of the 50 quants on O’Neil’s “futures group” team, she was the only woman. She was siloed from many of her other coworkers, so that if someone walked away to another firm, they wouldn’t bring other quants’ trade secrets with them. The work was exhausting and sometimes frightening, like when huge sums of money were needed frantically and immediately. But something deeper began to gnaw at O’Neil. The numbers she was playing with all day weren’t just abstract figures—they represented people’s livelihoods, retirement funds, and mortgages.

In July of 2007, interbank interest rates spiked. Even though lots of people had been able to secure mortgages in the housing boom of the last several years, banks were now realizing that there was some “dangerous junk” in their portfolios. Shaw could see that many companies and world markets would suffer—but as a hedge fund, they didn’t plunge into risky markets. Instead, they stood on the sidelines and bet on them. Hedge funds are less like baseball fans who cheer when their team wins, and more like gamblers who bet on movements associated with the game. So, O’Neil and her team, while nervous about what was to come, felt more or less safe. Even though the market was beginning to grow unstable, Shaw was “on top of the world.”

But even Shaw started to get nervous as the market continued to rumble. Mortgage-backed securities—previously “boring financial instruments” that could actually offset risk through quantity—became bigger liabilities. Since the 1980s, investment bankers had been buying up and packaging mortgages into securities, a kind of bond, by the thousands. The mortgages were like the “little pieces of meat of varying quality” that make up a sausage, and the securities were the spices. No one worried about them because they were essentially marked as safe. But mortgage companies had been lending money to people for homes they couldn’t afford, collecting the fees, and then unloading the securities into the market. The deals that the banks were offering weren’t just unsustainable—they were predatory.

These subprime mortgages weren’t WMDs—they were financial instruments, not models. But when banks turned the mortgages into securities and sold them, they relied on mathematical models—flawed ones—to do so. So, the risk model attached to the mortgage-backed securities was a WMD. None of the companies’ mathematicians were updating their data and continuously balancing the risk. The numbers these companies did have had been given to them by people committing wide-scale fraud. The risk ratings were kept hidden from the public, the risk models created a feedback loop by falsely rating defective products, and the fraud was happening at an enormous scale. So, the securities had all the components of a lethal WMD. 

The algorithms that had created the market and analyzed the risk in the securities turned out to be useless. Disaster hit the economy, and the human suffering it created was finally on display. In the financial sector, everyone—including the quants at O’Neil’s firm—began to wonder what would happen next. But by 2009, it was clear that the industry hadn’t really learned anything or changed—there were just a few more hoops to jump through. 

O’Neil had become disillusioned with the world of finance; people were wielding formulas recklessly and inappropriately. O’Neil left Shaw in 2009, planning to work on fixing WMDs from the inside out by joining a group that provided risk analysis for banks. The longer she worked as a risk analyst, though, the more she got the sense that she and her colleagues were seen as “party poopers” or threats—even given the cataclysmic crash that the country had just been through. 

In 2011, O’Neil switched roles yet again. She joined a web start-up as a data scientist, where she built models to anticipate the behavior of the users who visited travel websites and to try to distinguish casual window shoppers from motivated buyers. As she adjusted to her new job, she found lots of parallels between finance and Big Data—the biggest of which was that in both fields, money and self-worth were inextricably interwoven. People believed that they were successful in finance because they deserved to be, rather than because they were simply lucky. This itself was an example of a feedback loop. “Money,” O’Neil writes, “vindicates all doubts.”

O’Neil continued to grow disillusioned by how her new industry sought to replace people with data trails and turn them into more “effective” voters, workers, and consumers. She could see a new kind of dystopia growing around her, and inequality rising nonstop as data controlled and manipulated more and more people. Eventually, she quit her job to devote more time to investigating how algorithms were destroying lives.