Back to Newsroom
16-Jun-2017
Computers are helping justice
Pittsburgh, PA
On June 13, 2017, the New York Times published a lawyer's op-ed article that unfairly misrepresented Cybergenetics work for a just and safe society. On June 16, Cybergenetics Dr. Mark Perlin submitted this more balanced op-ed piece. He also sent them two letters. The Times did not respond or correct the record.
Four summers ago, a serial rapist was brutally attacking minority women and children in their Bakersfield, California homes. Entering before dawn, he would bind his victims with duct tape and zip ties, raping young working mothers and molesting their daughters.
Kern County forensic scientists analyzed DNA left on items found at three crime scenes. They developed a DNA profile from a zip tie dropped in the road by the fleeing rapist. A DNA database linked the tie to Billy Ray Johnson, the defendant named in Rebecca Wexler's June 13 op-ed article "Computers Are Harming Justice."
Thirty-seven crime scene samples from clothing, surfaces and bodies had small amounts of DNA. They were largely mixtures of three or four people. But the crime lab's FBI software didn't yield useful information. Fortunately, the county lab had Cybergenetics TrueAllele® computer system for analyzing mixture data.
The Bakersfield lab and the Pittsburgh-based company separately analyzed the data. Their computers unmixed the mixtures, identifying who had left DNA in the samples. The independent match results agreed. Following his conviction, the "sadistic monster" was sentenced to life in prison without parole, plus 423 years.
Most DNA evidence is a mixture of two or more people. Crime labs generate good mixture data, but for twenty years have interpreted that data incorrectly. Their FBI-provided software either gave no result, or calculated the wrong match statistic.
A 2005 Commerce Department inter-laboratory study documented widespread mixture interpretation failure and inconsistency. In 2013, Commerce examined "improved" FBI methods, showing most labs falsely identified someone who was not in the mixture.
Failed DNA evidence leads to criminal injustice. When match statistics are wrong, convictions are unsound. Unreported exculpatory DNA can falsely imprison the innocent.
In 1989, five men were terrorizing northern Indiana with "bump and rape" crimes. They would hit a car, and then rob, beat or rape the driver. Darryl Pinkins was misidentified and wrongfully convicted of rape and robbery.
A 2001 DNA analysis of the victim's jacket and sweater showed mixtures. Limited FBI-style interpretation found two unknown DNA profiles. But these were insufficient for exoneration, since they did not clear Pinkins.
Cybergenetics pro bono computing on the same data uncovered five profiles. Three were brothers. In 2016, the prosecutor accepted the TrueAllele science, and released Pinkins from prison. Exculpatory DNA data, dormant for fifteen years, had finally seen the light of computer interpretation.
Cybergenetics developed TrueAllele over ten years, starting in 1999. The system's accuracy was established through extensive testing. Seven of the 34 validation studies are published in peer-reviewed scientific journals. TrueAllele results have been reported in hundreds of cases across 37 states.
New York City's Medical Examiner had TrueAllele reanalyze World Trade Center DNA to identify victim remains. Seven crime labs, from California to Virginia, use their in-house system to interpret mixtures. Prosecutors, defenders, police and innocence groups rely on TrueAllele when FBI technology fails. Cybergenetics provides free services to help them.
Opponents can challenge the computer's reliability, asking a judge to keep the evidence from the jury. Since 2009, there have been a dozen such admissibility challenges. TrueAllele has prevailed every time, based on scientific validation.
TrueAllele is transparent. Cybergenetics has long disclosed the underlying mathematics. In a criminal case, the company provides both sides with a 4 GB disk containing validation studies, published papers, educational videos, electronic data, and free software for reviewing results. The opposing side can test the system on their own data at no cost. Lawyers argue, scientists test.
Computer programmers write human-readable software text called "source code." A compiler tool translates this text into a computer-executable application program, which can process data.
Crime labs use many software programs. They don't have or read source code text. Rather, they run their application programs on input data to test reliability and compute results.
Government recognizes that privileged information benefits society. Compelling lawyers to betray clients, or forcing reporters to reveal confidential sources, harms the public. The same law protects trade secrets, needed by companies to innovate essential technology in a competitive world. All courts have ultimately ruled that the law does not require TrueAllele source code disclosure.
TrueAllele mixture analysis is objective. All DNA data are entered, and the computer doesn't know the defendant's profile. But other mixture interpretation methods can introduce human bias that affects the outcome.
Nick Hillary was accused of killing a twelve-year old boy in Potsdam, New York. The case hinged on the boy's fingernail DNA, a mixture of the victim and fragments of a foreign cell. When FBI software failed to give an answer, the prosecution turned to recent New Zealand statistical software.
The software operators chose their data in a way that implicated Hillary. Using more of the data, the same software exonerated Hillary. I assisted the defense team pro bono, explaining the foreign software's limitations on very small amounts of DNA. The judge precluded the evidence, and Hillary was acquitted.
Reliable computer mixture interpretation helps society promote justice, prevent crime and reduce waste. But not everyone embraces better technology.
Defense lawyers try to keep unfavorable DNA evidence out of court (source code is one ruse). Some crime labs prefer choosing their data to using objective software. Decades of failed DNA mixtures have harmed tens of thousands of innocents, creating a multibillion-dollar injustice liability.
We must overcome such entrenched interests. Better science is needed for better justice.