After a successful appeal by the defendant, state attorneys have filed a motion to reconsider the use and evaluation of probabilistic genotyping software TrueAllele in the case of Corey Pickett, a man charged with murder in New Jersey.
While Cybergenetics’ TrueAllele has had over 25 admissibility cases in multiple states and federal court, New Jersey is not one of them. The Pickett case is the first time a New Jersey court has addressed the use of probabilistic genotyping software.
At an evidence hearing in 2019, the court heard TrueAllele’s admissibility testimony The court fell in line with other state courts that have moved to protect the software’s source code as a “trade secret.” However, defense attorneys for Pickett have argued that they need access to the source code to commence an independent review. An appeal was granted in February 2021 by the appellate court, sending the case back to the trial court.
“In appropriate circumstances, especially where civil liberties are on the line, independent source code review is critical when determining reliability at a Frye hearing. These case studies illustrate that software is not immune from error. Fundamental due process and fairness demand access,” wrote Judge Fasciale in the appellate division of the Superior Court of New Jersey.
Days later, prosecutor Esther Suarez filed a motion to reconsider, taking legal aim at three specific arguments used in the February appeal decision: 1) widespread use of probabilistic genotyping software; 2) the 2016 PCAST report; and 3) relevant rulings from other jurisdictions.
Case background
Pickett was charged in connection with a fatal shooting that occurred in Jersey City, New Jersey. On April 17, 2017, Pickett and an alleged accomplice opened fire into a crowd, killing one man with a shot through the head and injuring a 10-year-old girl, who suffered a non-fatal wound to the stomach when a bullet entered the car she was waiting in. Police apprehended Pickett and his alleged accomplice, and recovered two handguns and a ski mask.
Traditional DNA analysis indicated Pickett as a major source contributor for the DNA profiles lifted from the ski masks. Remaining samples were sent to Cybergenetics for analysis. TrueAllele then identified Pickett as a source of DNA found on one of the handguns as well as a ski mask. Ferrara could not be identified as a contributor to any of the samples.
The motion to reconsider
The appellate decision said errors in the source code of probabilistic genotyping software have been found, and that there is no reason to assume that TrueAllele is immune from these errors since “peer-reviewed studies are not a substitute for source-code review.” However, pointing to a paper by John Buckleton et. al, the state says discovery of miscode has not been found by code review, rather by empirical testing. Buckleton—who is the creator of STRmix, TrueAllele’s competitor—and his team uncovered a minor issue in the Forensic Statistical Tool probabilistic genotyping software created and used by the Office of the Medical Examiner of New York City.
“This minor and largely innocuous routine was rediscovered by testing and subsequently confirmed in the code. In our experience this is the normal sequence. The testing identifies an unusual behavior in the software, the cause of which is subsequently found in the code once both a suitable test example is available, and a portion of the code comes under scrutiny,” the paper, published in WIREs Forensic Science, reads.
The appellate decision also references the 2016 PCAST report that takes aim at the methods used in DNA analysis of complex mixture samples. The state argues the report has been denounced by both forensic science and law enforcement, including the Department of Justice announcing that it would not follow the report’s recommendations. It should be noted that TrueAllele, like STRmix, has been validated and is in compliance with the FBI’s Scientific Working Group on DNA Analysis Methods (SWGDAM).
Lastly, the motion to reconsider points to the relevant rulings from other jurisdictions in which similar claims regarding the TrueAllele source code were rejected. The Pickett case is unusual in that there was no hearing or chance for cross-examination. The state argues there have been a number of other courts that have addressed the source code issue after cross-examination of experts, and all come to the same conclusion. The motion references eight cases in seven states—Ohio, Virginia, Pennsylvania, Washington, Tennessee, Georgia and Nebraska—that say TrueAllele meets the Daubert standard, validation studies show its reliability, and it can be evaluated without access to the source code.
Source codes as trade secrets
On a broader level, the issue comes back to the question of if the source code of probabilistic genotyping software—current and future—should be considered a trade secret. Cybergenetics has entered trade secrets claims in more than 10 cases where TrueAllele’s findings were used as evidence. Thus far, the courts have sided with the company, every time.
In previous cases, Cybergenetics has granted use of its software to expert witnesses for the defense for testing. The company has also provided source code under a protective order. The company offered the same to the defense in the Pickett case, but the two sides could not come to an agreement on the extent of the protections. Cybergenetics offers inspection of the source code by an expert defense witness, but requires that expert to accept responsibility for any legal and financial consequences in the event of a breach. Experts for Pickett’s legal team did not agree to that liability.
But Perlin, CEO of Cybergenetics, is adamant about protecting his source code in what he calls a “highly competitive commercial environment.” He also says the source code is just not necessary to test the accuracy of any his software—nor anyone else’s.
“An algorithm describes a procedure. A programmer writes in a computer language, translating the algorithm into source code text. A compiler turns the text into executable software that runs as a smartphone, laptop or other computer app. Algorithms are shared, software is tested. Since software pirates can easily copy text files, trade secret law protects source code confidentiality,” Perlin explains. “You don’t learn how a car works by reading its blueprints; you take it for a test run. Lawyers read, scientists test.”
TrueAllele’s competitor, STRmix, has never filed a trade secrets claim. The algorithms to both software programs have been published in peer-reviewed literature.