What profiling really is

In the 1970s, FBI agents began visiting prisons and speaking with convicted murderers. Those conversations - there were 36 of them - gave rise to the best-known profiling model: the "organized" and "disorganized" offender. The logic is simple: how a person behaves at a crime scene is how they live in general. A neat scene means they plan, hold down a job, drive a car. Chaos at the scene means chaos in life too.

The model made its way into books and films, becoming the myth of an expert who looks at photos and describes a person almost as if from their passport. In reality, a profile is a probabilistic description of a group: approximate age, whether they have previously come to the police's attention, whether they live near the scene, whether they act alone. It contains no name or address, and it cannot.

In 2004, researchers examined about 100 cases of American serial murderers and tried to find those two types in them. No clear split emerged: signs of "organized" behavior appeared almost everywhere, while "disorganized" details were mixed in. A year earlier, another experiment showed something more troubling: police officers judged the same profile accurate for 2 different offender descriptions - the text was general enough to fit both.

The most famous failure happened in the fall of 2002 near Washington. While the shooter was killing people at gas stations and parking lots, public theories focused on a lone offender. There were 2 shooters, and the car was different.

Still, some parts of profiling work today - just not mind-reading, but statistics. Geographic profiling takes crime locations and estimates where a person is most likely to live or work: people rarely travel far and rarely attack right on their doorstep. It does not give an address; it changes the order in which investigators search neighborhoods. Even here, simple rules sometimes work no worse than specialized programs.

So the honest answer sounds ordinary: profiling is a set of assumptions about whom to check before everyone else. It saves weeks when it narrows the pool correctly, and costs weeks when an assumption is taken for fact.