On an April night in 1989, Jo Ann Parks survived a house fire in the Southern California city of Bell that claimed the lives of her three small children. Two years later, investigators announced that the seeming accident was actually a monstrous crime in which the 23-year-old waitress had set several fires, then barricaded her 4-year-old son inside a closet so he could not escape. Convicted through the power of forensic fire science, Parks was sentenced to life in prison without parole. She has steadfastly proclaimed her innocence.
Nearly 30 years later, a revolution in the understanding of fire has exposed many of the scientific certainties of the era as guesswork in disguise — including forensic evidence used to convict Parks that was flat-out wrong.
At her trial, the prosecution’s lead arson expert incorrectly claimed that a common and very intense burning known as flashover did not occur in the fire. Flashover can obscure a fire’s origins and make an accident look like arson, and its absence allowed the prosecution to argue with powerful certainty that the fire was deliberately set. Research since then has led to more accurate ways of looking for evidence of flashover and a greater understanding of its misleading effects, and even prosecution fire experts acknowledge now that it did occur in Parks’ case.
Yet she remains in California’s women’s prison in Chowchilla.
Welcome to the real world of forensics, where the wizardry lionized by the “CSI” television empire turns out to have serious flaws. The science of bite-mark comparisons, ballistic comparisons, fingerprint matching, blood-spatter analysis, arson investigation and other common forensic techniques has been tainted by systematic error, cognitive bias (sometimes called “tunnel vision”) and little or no research or data to support it. There is, in short, very little science behind some of the forensic “sciences” used in court to imprison and sometimes execute people.
Welcome to the real world of forensics, where the wizardry lionized by the “CSI” television empire turns out to have serious flaws.
The rigorously researched and peer-reviewed newcomer to forensics, DNA matching, has thrown into sharp relief the lack of scientific rigor in many other forensic disciplines. According to data gathered by the National Registry of Exonerations, of the 2,363 inmates exonerated of murder or other serious felonies since 1989 (most commonly through DNA), 553 were convicted with flawed or misleading forensic evidence—nearly one out of four.
Forensic science’s shortcomings have left the justice system alternately in a quiet panic or massive denial. The issue was first brought into the spotlight by a highly critical report from the National Academy of Sciences in 2009, which found a dearth of scientific backing for most forensics methods other than DNA. It cited evidence that “faulty forensic science analyses may have contributed to wrongful convictions of innocent people.” That report was followed by an even more blistering presidential commission report in 2016, which found serious errors and junk science in a host of commonly used forensic methods tying suspects to crimes.
Even the seeming infallibility of fingerprint evidence took a big hit. Multiple experts at the FBI’s vaunted Latent Print Unit incorrectly matched a Portland, Ore., attorney to prints found at the scene of the 2004 Madrid train station bombing. The prints actually belonged to an Algerian terrorist. A form of cognitive bias — finding what you expect to find— has been blamed because the FBI examiners had received extraneous information about the lawyer converting to Islam, and they were also told that a respected senior agent had already declared a match.
Closer to home, the murder conviction of Bill Richards of Mojave led the California Legislature to confront the problem of junk science in the courtroom — but only after an innocent man served 22 years in prison for supposedly killing his wife. After two hung juries failed to convict Richards, prosecutors found a bite-pattern expert who tilted the scales by matching a mark on the victim’s hand to Richards’ crooked teeth. Years later, attorneys at the California Innocence Project based in San Diego requested testing of samples from the murder weapon, which uncovered DNA that did not belong to either Richards or his wife. The expert recanted and admitted there was no scientific basis for any of his bite-mark findings in the case. But Richards’ release was delayed for eight years after prosecutors argued successfully that only factual testimony, not expert opinions, can be false evidence under California law. It took new legislation to change the definition of false evidence to include disproven or recanted expert opinions before Richards walked out of prison in 2016.
The change in expert witness law also allowed the California Innocence Project attorneys to attempt to overturn Parks’ conviction for burning her children to death. Her attorneys argued that false forensic expert opinion was used against her at her original trial.
But the fire investigation community remains divided on how to handle cases such as Parks’. Experts testifying for her during a recent habeas corpus hearing said the misleading fire patterns and destruction wrought by flashover made it impossible to determine the origin and cause of the fire, and that she deserved a new trial. Prosecution experts argued they could still read the fire patterns like a book despite their concession that flashover had occurred. They offered little in the way of research or error rate studies to back that assertion.
In November, Los Angeles County Superior Court Judge William C. Ryan ruled that the new flashover evidence was insufficient to win Parks a new trial because the experts cannot agree on its significance. “The world of fire science and fire investigation is a complex area rife with differing opinions and contentious debates,” Ryan wrote.
His decision is being appealed.
Whatever the outcome of the Parks case, it is one of many demonstrating the profound difficulty the justice system has in separating good science from bad. The National Academy of Sciences has suggested raising the bar for expert testimony by requiring hard data and error rates for all forensic disciplines.
Right now the bar is shockingly low. One expert in the recent Parks hearing testified that his analysis of door hinges showed that she had barricaded her child in a closet, using a technique he had never attempted before and for which he cited no scientific data. This lack of scientific rigor in the courtroom has to change.
A commission formed by President Obama to study solutions to flawed forensics was disbanded by the Trump administration. It may be time for the states, individually or in partnership, to undertake this effort. The stakes are too high to maintain the status quo.
Edward Humes is a Pulitzer Prize-winning journalist and the author of 15 nonfiction books, including his latest, “Burned: A Story of Murder and the Crime That Wasn’t.”