For many years, experts such as Elizabeth Loftus have demonstrated experimentally and statistically the fundamental inaccuracy of eyewitness identifications, one of the time-honored foundations of the criminal justice system.  Five years ago, the Innocence Project gave real-world weight to these findings with an analysis showing that 71 percent of wrongful convictions involved some kind of mistaken eyewitness identification, both in and out of court, and that of that 71 percent, more than half involved an incorrect in-court identification. The result of these findings has been a move to limit the use of in-court eyewitness identifications.

A 2016 state supreme court decision in Connecticut held that witnesses cannot be asked for an in-court identification unless they knew the defendant before witnessing the crime or have already successfully identified the defendant in an out-of-court procedure, or the perpetrator’s identity is not contested.

In Massachusetts in 2014, the state’s top court largely banned the practice for cases in which witnesses had been anything short of unequivocal in identifying the defendant before the trial. It’s possible that Colorado will soon be joining them.

According to the Innocence Project, first-time in-court identification increases the risk of wrongful conviction. They argue that the powerful theatrics of pointing to the defendant can sometimes overcome the shortcomings of a weak case. The courtroom layout—with the accused seated next to defense counsel—leaves little doubt about who the state is trying to prosecute, and the defendant may be the only person in the room who matches the witness’s description. Instead of relying on what he remembers from the crime scene, a witness might guess or simply point to the defendant because he seems like the most obvious choice in the courtroom setting.