An article from Ben Paynter at Good Magazine about Gary Well’s work in the Austin Police Department to use a computer program to improve crime witnesses’ identification of suspects.
“It’s an experimental protocol designed by Gary Wells, the guru of eyewitness reliability—or rather, unreliability. The director of social sciences at the American Judicature Society’s Center for Forensic Science and Public Policy, Wells has been working on lineups since the 1970s, but in the past 20 years exonerations of hundreds of prisoners based on DNA evidence—after many had been convicted in part based on good-faith eyewitness testimony—have made his task all the more urgent. Wells doesn’t want to merely understand witness identification. He wants to fix it.”
“Attorney General Janet Reno asked Wells to head a task force on new lineup guidelines for states, and he proposed new practices drawn from his research. All lineups should be blind, he said—the cops administering them shouldn’t know who the suspects or fillers are. There should only be one suspect per lineup. Witnesses should be clearly advised that a suspect might not be in the lineup. And statements of confidence should be recorded verbatim at the time of the pick, because witnesses with any uncertainty have been known to talk themselves into their choices as time passes.”
“In 2006, Wells designed a new study protocol. The tests wouldn’t just be blind but “computer blind”—the computer itself could offer prerecorded instructions to ensure lineups were done uniformly. After officers created a lineup, the photos would also be digitally shuffled so they couldn’t pass along the location of their suspect to anyone running the lineup. That eliminates the chance of lineup administrators giving off any cues—subtle nods, coughs, or the suggestion to pay closer attention to any one photo—that might be used, unconsciously, of course, to tip witnesses off to prime suspects. The computer would even randomly decide whether to run a sequential lineup or a simultaneous one.”