She took a picture of the man who attacked her. It didn't matter.

Last year, on a cool September afternoon around two o'clock, a friend who lives in my building was walking to the post office in downtown Brooklyn when she was attacked by a stranger. She was on the phone when she vaguely noticed someone around her. Suddenly he was right in front of her, imitating her movements as she tried to step away. He assumed the position of a linebacker and tackled her to the ground, leaving her on the sidewalk with several injuries.

He walked away, but it wasn't long before he turned around and came back. At that moment, my friend Laura, a petite artist in her fifties (who asked that I not use her full name because she continues to feel vulnerable), was safe in the nearest building. From behind a glass door she was able to take a photo of the man with her mobile phone. And there was even more visual evidence: A nearby security camera had recorded the attack, the footage of which my friend eventually viewed in the company of detectives at the 84th Precinct.

The incident struck me not only because it happened in the middle of the day to someone I know and care about, in what is considered a very safe part of Brooklyn, but also because of what followed procedurally and what it revealed about the still increasingly questionable place of technology in modern law enforcement.

On October 23, five weeks after the attack that left Laura with bruises on her lower back, a chipped tooth and abrasions on her elbow and forearm, she was called to the police station to identify a suspect. There were no real people in the lineup; instead, she was confronted with a presentation of eight photos of different men who, she said, looked unnervingly similar.

“The idea that I might wrongly accuse someone weighed on me,” she told me later. Although she was able to quickly eliminate five of the eight, she found it difficult to distinguish between the remaining three; each of them had a point at the top of its skull, she noticed, and eyes that turned downward.

She has made a selection. Detectives then told her that her attacker was number 5; she had chosen the wrong man. She had not recorded any details about her attacker's appearance during the incident itself, but she had looked at the photo she had taken. He appeared to be in his twenties or early thirties and was wearing patched jeans, white sneakers and a black parka. He had a blank stare, a small, noticeable bump over his right eyebrow and a small scar over his left eyebrow. If you looked closely, you could see a cigarette in his left hand.

Detectives told her that despite the photo and video evidence, her mistake would prevent them from taking anyone into custody. When I asked New York Police Department sources to explain why victim identification, known to be so unreliable, would trump visual imaging — which in this case included a high-resolution iPhone photo taken immediately after the attack was taken – a spokesperson responded with: an email stating: Detectives are “working closely with prosecutors to build the best possible prosecution,” including “taking several investigative steps” to “procure an arrest '.

In other words, regardless of the clarity of the portrayal, human determination remained the gold standard, and in the absence of an accurate one, the case was considered too weak to achieve a conviction.

This implicit understanding of living beings as the most reliable witnesses to traumatic events contradicts decades of social science research. According to a report by the Innocence Project, titled “Re-evaluating Lineups: Why Witnesses Make Mistakes and How to Reduce the Chance of Misidentification,” empirical and peer-reviewed research “reaffirms what DNA exonerations have proven true: human memory is fallible.” For all the disadvantages of living amid the ever-present surveillance of the 21st century, one advantage would likely be the ability to correct precisely these errors in human observation.

Memory formation consists of three phases: encoding, storage, and retrieval. “If someone is in a moment of stress — when someone has attacked them — that stress affects both the encryption and storage functions,” explains Alexis Hoag-Fordjour, a criminal justice professor at Brooklyn Law School. There is a frustrating lag, she said, between developments in technology, pathology, social sciences and science in general and what is happening in the law.

There is no uniform approach to the use of photographs and security camera footage when making decisions in criminal cases. “The irony,” said Alex Vitale, a sociologist who has studied policing for 30 years, is that if Laura had died, police “would have been perfectly happy” to arrest the suspect “in the absence of positive eyewitness identification.” ”

Some lawyers, like Julie Rendelmanformerly deputy chief of the homicide unit in the Brooklyn district attorney's office, argues that misidentifications should not necessarily prevent prosecution from progressing, or, as she explained, “that the level of crime should be relevant to the filing of the case. ”

What happened to Laura was especially psychologically disruptive. The release of her attacker is the kind of outcome that will infuriate those who see New York as an increasingly dangerous place where law and order has been subjugated to the ostensible virtues of progressive reform. But it would also rankle with progressives who see insufficient attention to the psychological well-being of homeless and other marginalized people as the problem animating our sense of anxiety.

Detectives asked her if she wanted to press charges, and she did so, she reasoned, so the man who attacked her could get help.

Lincoln Restler, the city council member representing the area where the attack occurred, said the decision to prosecute an attack like this “should be made by the district attorney's office every time,” adding, “If if that case is heard, then the court, a judge, could refer the alleged attacker to treatment services, even housing,” assuming that is what is needed.

As Mr. Vitale suggested, “It's not that we don't know who these people are that are responsible for these quality of life problems.” The challenge, as he put it, is that “we don't know what to do about it.”