The case involved South Carolina bank robbery. Witnesses were unable to ID the robber and the bank's surveillance video turned out to be too grainy to help. When the FBI finally arrived at a suspect, the man countered that the robber had been his brother. With only low-quality video to go by, how could investigators tell which brother had committed the crime?
The short answer is that they could not. Richard Vorder Bruegge, one of the FBI's most influential image examiners, reported that the FBI's suspect had a similar "overall shape of the face, nose, mouth, chin and ears" but didn't actually declare him a match to the grainy video. He admitted that the video was too low a resolution to fully identify the suspect.
That's not what federal prosecutors said in court filings, according to the nonprofit investigative journalism outfit ProPublica. They said that Vorder Bruegge would testify that the images of the bank robber could be shown definitively to be the FBI's suspect. A judge ruled the testimony would be admissible. The suspect, led to believe the government had very strong evidence against him, pled guilty.
ProPublica found at least three cases where Vorder Bruegge's lab results said one thing but the courts were told another -- either by prosecutors or by Vorder Bruegge himself.
These findings aren't meant as an attack on Mr. Vorder Bruegge, but rather to point out serious issues with yet another one of the FBI's forensic techniques. ProPublica published an investigation in January showing that the FBI's image analysis techniques are not rooted in science.
Unfortunately, it seems that many forensic techniques lack sufficient scientific underpinnings. In 2009, the National Academy of Sciences issued a review of several common forensic techniques of questionable validity. In 2016, the President's Council of Advisors on Science and Technology laid out similar concerns as a follow-up to the 2009 report. Feature-comparison methods like image analysis were especially criticized as lacking sufficient study.
When expert testimony is admitted into evidence, the expert must first demonstrate to the court that the techniques used are widely accepted within the relevant scientific community. Since forensic techniques like image analysis lack sufficient scientific study, you might expect they would be inadmissible as evidence.
Yet judges continue to allow unproven forensic techniques to be used as evidence -- perhaps because they are common and familiar. Or worse, because television shows like "CSI" make these techniques seem highly scientific and reliable.
Criminal defendants should not have to overcome evidence that is being oversold by prosecutors. However, many common forensic techniques are oversold as reliable scientific evidence when they simply are not.