The CSI effect is a positive trend in criminal law. In a nutshell, lawyers now see an unwillingness on the part of jurors to convict with only circumstantial or eyewitness testimony, and jurors demand more forensic evidence before they are willing to convict defendants.

Before the advent of shows like CSI, jurors put too much faith in eyewitness testimony. Slowly, the message is getting out that people aren’t all that reliable in what they think they remember. And there isn’t a correlation between how confident witnesses seem or claim they are in their memories, and how accurate their memories actually are. A good lawyer can help a witness exude confidence, while other witnesses are just naturally confident.

Problems with eyewitness memory are compounded when it comes to identifying people. Anyone can identify what their close friends look like, but identifying someone you’ve only seen once—for the few moments it takes to commit a crime—is hard. It gets harder if the two people are from different ethnicities. A confident witness can convince a jury that they remember a criminal’s face just as well as they remember their own mother’s face, but in all likelihood, they’re wrong.

Unfortunately, the Supreme Court doesn’t see this as a problem (or at least the Court doesn’t see it as a violation of the Constitution.) In a case decided earlier this month, Perry v. New Hampshire, the Court took a step in the wrong direction as far as letting bad eyewitness testimony into trials. The Court ruled that judges do not need to exclude eyewitness identifications, even if the identification procedure was unnecessarily suggestive, as long as the police did not arrange the identification procedure. There are many ways eyewitness identifications can be “unnecessarily suggestive.” For example, the police can have the identification take place on the street and show the witness someone they arrested, and have handcuffed and laying on the ground. Witnesses want to be helpful, so they’re likely to confirm what the police already suspect, even if the witnesses aren’t entirely sure. If the police arrange for the eyewitness to make the suggestive identification, courts are supposed to exclude that evidence. But the Supreme Court said it’s okay to use unnecessarily suggestive identifications if the eyewitness just happens to spot the suspect in police custody.

During arguments before the Court, the Innocence Project showed that in 2/3 of the cases where they had used DNA to exonerate a wrongful conviction, an eyewitness had incorrectly made an identification during trial. Most of the convictions that the Innocence Project managed to exonerate occurred prior to the widespread use of DNA during trials. Therefore, the trials also predate the CSI effect. One hopes that the CSI effect would have prevented some of the wrongful convictions, but we can’t be sure.

Although the Court doesn’t think it is a violation of the Constitution to let bad evidence into trials, individual judges can let juries know that there are problems with eyewitness identifications, but most won’t do so. What is worse, many judges won’t let defendants use experts to tell to juries the problems with eyewitness identification. Some judges seem to think such testimony confuses juries.

Only one Justice, Justice Sotomayor, dissented from the Court’s opinion, so this is not a decision that is going to be reversed any time soon. In the meantime, I’ll hope that pop culture continues to teach jurors that when it comes to evidence, hard science is best.