Data profiling now predicts behavior across marketing and crime

Understanding Data as a Tool

The episode opens with Ali Ilyas explaining his background: “my specialization in terms of education is data analytics.” He frames data not as inherently good or bad, but as the foundation underlying everything from marketing to politics. Muzamil and Ali begin by dissecting Cambridge Analytica, the political consulting firm that profiled millions of Facebook users to influence the 2016 U.S. election. The key insight: profiling isn’t new, but the scale and precision of data-driven profiling is.

How Cambridge Analytica Profiled Voters

Cambridge Analytica’s technique was simple but powerful. By analyzing what people liked and followed on Facebook, the firm built psychological profiles that predicted political beliefs. Ali explains the mechanics: if someone liked certain pages and followed specific accounts, combined with age and geography data, the algorithm could predict whether they leaned conservative or liberal. Once profiled, Cambridge Analytica showed different versions of divisive political content to different groups. On abortion, for example, conservatives saw one Hillary Clinton statement while liberals saw another, each tailored to deepen existing divisions. “They were pushing both left and right,” Ali observes.

The Origins of Criminal Profiling

Ali then pivots to the history of criminal profiling, which follows the same logic but for crime detection. In the 1950s, a detective sought help from psychiatrist James Brussel to catch a serial bomber placing explosives across New York City. Instead of traditional investigation, Brussel used what he called “reverse psychology”: analyzing the crimes to deduce the criminal’s profile. Brussel predicted the bomber was paranoid schizophrenic, of Eastern European descent, living in Connecticut, and antisocial. Most strikingly, he predicted the bomber would be wearing a double-breasted jacket when arrested. When police finally caught George Metusky in Connecticut, he opened his door and said, “oh, I know why you’re here. You think I’m the mad bomber.” He was wearing exactly the jacket Brussel had predicted. This story illustrates that profiling, whether in marketing or forensics, works because human behavior follows patterns.

Profiling in the Modern Age

Muzamil and Ali discuss how profiling has become inescapable in 2020. Social media platforms now collect data on every page a user likes, every search term entered, every pause in a video. This granular data stream enables profiling far more sophisticated than anything James Brussel had access to. A user searching for baby products on Google will see baby-related ads on Instagram; their digital traces reveal their life stage and interests instantly. Ali notes that “profiling has probably been happening for a very long time,” but the speed and scale have transformed its impact.

The Gap in Pakistan

When Muzamil asks what’s happening in Pakistan, Ali is direct: forensic psychology and criminology are underdeveloped fields. While the New York City Police Department brought James Brussel in to solve the Mad Bomber case, Pakistan’s law enforcement lacks comparable infrastructure. Ali mentions a data scientist, Sishan Osmani, who published a dataset of Pakistani bombings from 2001 to 2015, analyzing blast radius and bomb composition to narrow suspect profiles. But this remains the exception, not the standard practice. Without institutional support and trained professionals, Pakistan cannot match Western crime-solving capabilities.

Profiling as Science

Muzamil reflects on a misunderstanding from his first podcast episode. He described demographic patterns he observed among his audience and faced criticism for “stereotyping.” But profiling is not stereotyping; it’s statistical probability. Muzamil clarifies: “I was just saying an anecdote… I was describing a profile.” If data shows 83 percent of his viewers are aged 18 to 34, that’s a profile. It doesn’t eliminate the outlier, the 50-year-old viewer, but it identifies the most likely user. This distinction matters because profiling enables precision: marketers target their likely customers, law enforcement narrows their suspect list, and platforms recommend content to their core audience.

Governance, Control, and the Future

The conversation closes on governance. Ali and Muzamil discuss China’s social credit system, which rates citizens on financial behavior, social conduct, and civic obedience. In the West, similar profiling exists but is framed differently: credit ratings determine loan eligibility, background checks influence employment, and social media algorithms shape information. Both systems use profiling to control behavior; they differ mainly in transparency and institutional rhetoric. As profiling technology becomes more powerful, the question shifts from whether societies will profile citizens to who controls that profiling and for what ends.