Paul Kix’s riveting feature in the New Yorker today is one of the best things I’ve read in awhile. Far too often in today’s system, it’s easier to score political points by being “tough” on crime than being “smart” on crime, and the collateral damage is unimaginably tragic. I suspect this is one of the unfortunate aftereffects of all the reform in the 1990s to address the violent crime issues of the day.

In a dissenting opinion in 1981, Supreme Court Justice William Brennan wrote that “eyewitness identification evidence is notoriously unreliable.” Dozens of scientific studies support this claim. Nevertheless, eyewitness testimony continues to be used widely, and many criminal cases hinge on it almost exclusively. Since 1989, two hundred and eighty people have been exonerated of sexual-assault charges in the U.S. Nearly three-quarters of those wrongful convictions relied, in whole or in part, on a mistaken identification by an eyewitness.

Psychologists have long recognized that human memory is highly fallible. Hugo Münsterberg taught in one of the first American psychology departments, at Harvard. In a 1908 book called “On the Witness Stand,” he argued that, because people could not know when their memories had deceived them, the legal system’s safeguards against lying—oaths, penalties for perjury, and so on—were ineffective. He expected that teachers, doctors, and politicians would all be eager to reform their fields. “The lawyer alone is obdurate,” Münsterberg wrote.

It’s for people like Tim Cole that organizations like the Innocence Project exist and are worth our time.