Pseudoscience and Dubious Forensics

A dentist tells a jury there is a one-in-a-million chance anyone else left a bite mark. An FBI analyst says a hair at the crime scene came from the defendant. A federal agent swears that a set of household pliers, and no other tool ever made, cut the wires in a pipe bomb. Jurors trust testimony like this because it sounds like science. But much of it was never tested, and some of it has sent innocent people to prison for decades.

In this TalksOnLaw interview, Joel Cohen sits down with M. Chris Fabricant, Director of Strategic Litigation at the Innocence Project and author of Junk Science and the American Criminal Justice System, to examine how unvalidated forensic techniques enter the courtroom, why judges keep admitting them, and what it takes to undo a conviction built on them.

Science Born at the Crime Scene

Fabricant explains that the Innocence Project studied its DNA exonerations to find what went wrong. Eyewitness misidentification was the leading factor, faulty forensics second, and false confessions third. Forensic techniques, he says, differ from mainstream science because they were not developed in laboratories. They grew out of crime scenes, often from a single case. Bite-mark evidence traces to a California case in which Walter Marx was accused of killing his landlady and biting off the end of her nose. The judge said he could see the match with his own eyes, and the decision was cited as precedent across the country for years, though no research had tested the method. Fabricant counts at least 31 wrongful convictions and indictments tied to bite-mark evidence since then.

The Steven Chaney Case

Fabricant's leading example is Steven Mark Chaney, a Dallas union ironworker convicted in the killings of John and Sally Sweek, a couple who sold cocaine from their apartment. The police had many possible suspects and no eyewitnesses. A pathologist noticed a U-shaped mark on John Sweek's arm, and a forensic dentist later testified that there was a one-in-a-million chance anyone but Chaney made it. That figure leaned on a small published study claiming to prove that human teeth are unique. Chaney had nine alibi witnesses, but the jury convicted him anyway.

From Frye to Daubert

Fabricant walks through the two tests courts use to screen expert evidence. Under Frye, a technique comes in if it is generally accepted in the relevant scientific community. In practice, he says, courts let the practitioners define that community, so dentists vouched for bite marks and nothing was ever rejected. The shift to Daubert came from civil litigation, when corporate defendants pressed for protection against speculative science in product-liability cases. Daubert makes judges the gatekeepers and asks about peer review, testability, and error rates. Fabricant describes cadaver studies by Peter and Mary Bush at SUNY Buffalo that he calls devastating to bite-mark analysis: even on undistorted bites, the same teeth did not leave marks that looked the same, and teeth that had not made a bite sometimes matched it better. His central criticism is that Daubert is applied rigorously when money is at stake but loosely when liberty is, partly because of resources and partly because everyone in the courtroom tends to assume the defendant is guilty.

Cadaver Dogs and Hair Microscopy

Fabricant describes a Detroit case in which a father was convicted of killing his two-year-old daughter even though her body was never found. The evidence included a cadaver dog's alert, which passed a Daubert hearing despite no peer review and no known error rate. He recounts a study in which detection dogs "found" explosives in the boxes their handlers had been told to expect, though none of the boxes contained any. He then turns to microscopic hair comparison, which the FBI made routine in the mid-twentieth century. Analysts compared hairs by eye, without measurements or statistics, and sometimes made up odds like one in 10,000. In 2012 the FBI and Justice Department acknowledged that their examiners had overstated hair evidence for decades. That led to a joint review with the Innocence Project and the National Association of Criminal Defense Lawyers covering more than 3,000 cases, a review in which Fabricant remains involved.

Blood Spatter and Shoe Prints

In a quick round, Fabricant says blood-spatter analysis depends on fluid dynamics and physics, but many analysts rebuild entire events after 40 or 80 hours of training. He makes clear that he is not attacking the field itself, only how it has often been used. With shoe prints, he notes that different shoe sizes often share the same sole. He adds that even a cautious "could have made" opinion becomes one more seemingly objective link toward conviction.

How Chaney Went Free

Freeing Chaney without DNA evidence, Fabricant explains, required several things to happen. The National Academy of Sciences' 2009 report concluded that no forensic method except nuclear DNA analysis has been shown to reliably link evidence to a single source. The American Board of Forensic Odontology stopped claiming that a bite could be matched to one person to the exclusion of everyone else. The Dallas County Conviction Integrity Unit, started under District Attorney Craig Watkins and later led by Patricia Cummings, reinvestigated the case with the Innocence Project. It found an alternative suspect and evidence that the injury was believed to have happened days before the murders. The original dentist recanted the one-in-a-million figure. Finally, Texas enacted its "junk science writ," Article 11.073, which allows relief when the science behind a conviction has been contradicted. The Texas Court of Criminal Appeals declared Chaney actually innocent after 28 years in prison. Fabricant adds a caution: unreliable evidence does not prove innocence, but it does mean the trial was unfair.

Toolmarks and Jimmy Genrich

Fabricant then discusses Jimmy Genrich, convicted in a series of Grand Junction, Colorado, pipe bombings that killed two people. Police searched his apartment and found no gunpowder or bomb-making materials. Apart from his writings, the main evidence was testimony from an ATF examiner who linked marks on bomb components to Genrich's wire cutters, wire strippers, and slip-joint pliers, to the exclusion of every other tool. The examiner tied one of the tools to a bomb from a time when, Fabricant says, Genrich had an undisputed alibi. Almost 40 scientists later signed an amicus brief calling the testimony invalid.

Incentives, Bias, and Reform

Fabricant argues that forensic disciplines seek acceptance from courts, not scientists. Once courts admit a technique, there is little reason to do the basic research. Judges who see the problems often still admit the evidence as going "to weight, not admissibility" and rely on cross-examination, which he says cannot undo the authority of an expert the court has endorsed. He points to Itiel Dror's study in which experienced fingerprint examiners were given their own earlier casework with different case details, and two-thirds changed their conclusions. His proposed reforms: validate methods before they are used in court, as is done with toothpaste and aspirin; create independent oversight like the Texas Forensic Science Commission and national standards through NIST; separate crime labs from police budgets, following the model of Houston's independent lab; and shield analysts from case details that could bias them.

What to Know Now

Several of the cases Fabricant discusses have since moved forward. The Texas Court of Criminal Appeals formally declared Steven Chaney actually innocent in December 2018. The Texas Forensic Science Commission, after reviewing his case, became in 2016 the first state body to recommend a moratorium on bite-mark evidence. In 2023 a Mesa County judge ordered a new trial for Jimmy Genrich, finding the toolmark testimony unreliable and likely decisive. The Colorado Court of Appeals affirmed in 2025, and the Colorado Supreme Court declined review. In April 2026, prosecutors concluded a retrial was no longer viable, and the murder charges were dismissed. Genrich remains in prison on separate convictions for use of an explosive device and assault, which carry a 72-year sentence and which his lawyers continue to challenge. The FBI hair review reported in 2015 that examiners' testimony contained erroneous statements in at least 90 percent of the trial transcripts reviewed. Texas courts have also applied Article 11.073 to "shaken baby" convictions, overturning Andrew Roark's in 2024. In October 2025 the court stayed the execution of Robert Roberson, whose capital conviction also relied on that diagnosis, and sent his claim back to the trial court, where it remains pending as of 2026. The State continues to defend the conviction.

Courts have also tightened the rules. Federal Rule of Evidence 702 was amended effective December 1, 2023, to make clear that the party offering an expert must show by a preponderance of the evidence that the testimony is reliable and that the opinion stays within what the method can support. The Advisory Committee cited forensic overstatement as one reason for the change. In Abruquah v. State (2023), Maryland's highest court held that a firearms examiner may testify that bullets are consistent with a particular gun but may not give an unqualified opinion that the gun fired them. Forensic practitioners dispute how far such limits should go, and many courts still admit firearms and toolmark evidence with only modest restrictions. The Innocence Project counted 375 DNA-based exonerations nationwide through 2020 and reports that misapplied forensic science contributed to nearly half of them. The National Registry of Exonerations, which tracks all exonerations, DNA and non-DNA, finds false or misleading forensic evidence in roughly a quarter. Fabricant's 2022 book follows three men convicted of capital murder on since-discredited forensic evidence.