Some jurors made Parks’s acquiescence the deciding factor in their vote to convict her of murder by arson. In “Burned,” his riveting account of the fire and its repercussions, Edward Humes sums up those jurors’ reasoning as follows: “There was no way any mother — any mother who wasn’t a killer, that is — would hear her kids were okay and then leave the scene without first seeing them for herself.”
Yet the prosecution had not made that argument, perhaps because the defense could easily have explained such behavior by a mother in Parks’s situation. (On her lawyer’s advice, Parks herself did not testify.) For example, she may have had a foreboding that the kids were not okay, so putting off seeing them was a way to keep hope alive.
Unfortunate as the jurors’ mind-reading attempts may have been, Humes’s main objection to the Parks trial has to do with evidence-gathering. Before he is done with it, the small fire that killed Parks’s kids grows into a conflagration threatening the integrity of the American criminal justice system.
Arson experts’ opinion that Parks was the perpetrator rested in large part on their mapping of the fire’s path. Humes charges them with a faulty understanding of flashover, which occurs when a fire gets so hot that “every flammable surface in the room not already burning will ignite in rapid succession.”
The same year that Parks went on trial, an experiment was conducted at the Federal Law Enforcement Training Center in Glynco, Ga. After setting on fire two “stand-alone fake rooms,” the experimenters asked veteran arson investigators to examine each burned room and pick out the quadrant where the fire had started — “not the area of origin itself,” Humes emphasizes, “but just the quarter of the room that contained it.” Although the participants had expected the assignment to be “child’s play,” more than 90 percent of the time they chose a wrong quadrant. This and other tests have confirmed what a few arson experts were already coming to believe: that flashover can make domestic fires impossible to sort out afterward.
Although flashover had occurred in the Parks apartment, investigators confidently read patterns they discerned in the burnscape as implicating the mother. (Her then-husband had been out working a night shift and thus was not a suspect; Humes seems to regard their 4-year-old son’s habit of playing with matches as the likeliest cause of the fire.) In the investigators’ defense, they were adhering to the consensus at the time, but they, along with most other arson experts, have since clung to their old certainties and refused to question verdicts based upon them.
Widening his scope, Humes notices that other forensic standbys are being discredited, too. Comparing bite marks on a body with those made by a suspect’s teeth has been shown time and again to be a highly subjective exercise, often producing false positives. Hair and fiber comparisons are sketchy, too, as are the judgment calls involved in matching fingerprints. Even lineups from which witnesses are asked to pick out the culprit can be flawed. How do we know all this? From later comparisons of DNA samples, which speak for themselves.
Common to all these problems, Humes argues, is cognitive bias arising from the way crime investigations are often handled. The cops and the technical specialists work together, so that the latter enter a crime scene with a pretty good idea of what the cops expect — and want — them to find. Humes’s proposed solution is to adapt to police work the double-blind procedures used in scientific research. This would mean, for example, that those in charge of a lineup would have no idea which member of the line had been nabbed by their fellow cops; all too often, Humes suggests, the lineup conductor’s tone of voice or body language serves as a tell.
One might think that judges would take note of the growing number of wrongful convictions uncovered by DNA analysis and do something about it. But even bite-mark evidence, the least trustworthy of all in Humes’s opinion, is still being admitted because judges are so enamored of precedent. And the Trump administration has abolished an Obama-era commission charged with recommending solutions to forensic failures.
The California Innocence Project looked into Parks’s case and managed to get her a habeas corpus hearing. At that hearing, the defense’s expert witness testified that dramatic changes in fire science over the past quarter-century have made Parks’s conviction untenable; the prosecution’s expert downplayed the importance of those changes. Parks’s lead attorney, Raquel Cohen, came away from the hearing believing, in Humes’s words, that “every prosecution expert in the . . . case . . . has been influenced by unconscious cognitive bias.”
On Nov. 2, 2018, the habeas judge rejected Parks’s request for a new trial, ruling in essence that the expert witnesses had fought to a draw. Cohen wants to appeal — if, that is, she can raise the $100,000 it would cost to fund a project that might sway the appeals court: a staged re-creation of the apartment fire.
Humes makes no mention, however, of what might be the strongest weapon in Cohen’s arsenal: “Burned” itself, a powerful brief not only for Parks but also for a recognition of the weaknesses in forensic science generally.
A Story of Murder and the Crime That Wasn't
By Edward Humes
Dutton. 306 pp. $28