New Jersey Appellate Court Holds Defendant Entitled to Source Code of Novel Probabilistic Genotyping Software Upon Showing of Particularized Need
Police officers in Jersey City, New Jersey, pursued Corey Pickett and Jonathan Ferrara down a side street after observing the two men fire handguns into a crowd. After arresting the two men, the police retraced their paths and recovered a Colt .45 caliber semiautomatic and a Smith and Wesson .38 caliber revolver. Pickett was indicted on several offenses, including first-degree murder.
Swabs for DNA evidence from the handguns were sent to a laboratory for testing. The samples failed to meet the criteria for traditional DNA analysis. The State then forwarded the testing data to a private laboratory in Pittsburgh, Pennsylvania, operated by Cybergenetics Corporation (“Cybergenetics”). Cybergenetics, using its novel proprietary TrueAllele computer software program, identified Pickett as the source of the DNA on the Smith and Wesson. Relying on the evidence generated by the TrueAllele software, the State intended to call Dr. Mark Perlin as an expert. (Perlin is the cofounder of Cybergenetics.)
Pickett filed a motion requesting a hearing pursuant to Frye v. United States, 293 F. 1013 (D.C. Cir. 1923) (“Frye hearing”) to challenge the reliability of the scientific evidence. Prior to cross-examining Perlin, Pickett sought discovery of TrueAllele’s source code and related testing documentation.
The State objected, arguing that the source code was a “trade secret.” Cybergenetics is a for-profit company, and the source code is valuable to competitors and could be sold for profit. Perlin declared that ample materials had already been provided, including “over thirty validation studies and publications.” Perlin contended those reviews were adequate enough to enable cross-examination to determine if TrueAllele is reliable. The State asserted that disclosure of the trade secret and potential for exposure were unnecessary.
Pickett countered with a declaration from defense expert Nathaniel Adams, who explained that a novel program like TrueAllele is highly prone to error. Unlike traditional DNA analysis, probabilistic genotyping performed by software like TrueAllele examine much smaller quantities of DNA that are mixtures of more than two contributors. The source code is necessary to determine if the program design adequately accounts for these conditions when calculating the probability and when generating data that are then subjectively analyzed by the forensic analyst.
The parties could not reach agreement on any type of examination of the source code or terms of a protective order. The Superior Court, Law Division, denied the request for the source code. The Appellate Division granted leave to appeal. Several groups joined in amici, including the Innocence Project.
The Court observed that the Fifth and Sixth Amendments to the U.S. Constitution and article I, section 10 of the New Jersey Constitution, ensure criminal defendants “a meaningful opportunity to present a complete defense.” State v. Garron, 827 A.2d 243 (N.J. 2003). Forensic computer software generates expert evidence, and Rule 702 of the New Jersey Rules of Evidence permits introduction and cross-examination of expert witnesses. The proponent of the expert evidence must establish: (1) the subject matter of the testimony is beyond the ken of the average juror, (2) the field of the inquiry is at a state of the art such that an expert’s testimony is sufficiently reliable, and (3) the witness has sufficient expertise to offer the testimony. State v. J.L.G., 190 A.3d 442 (N.J. 2018). Factor two was at issue in this case.
The Frye test requires trial judges to determine whether the particular science underlying the proposed expert testimony has “gained general acceptance in the particular field in which it belongs.” Id. The three ways to establish general acceptance are: (1) expert testimony, (2) authoritative scientific and legal writings, (3) and judicial opinions. Id.
The Court scrutinized those three categories with the benefit of lessons learned from the software errors discovered in relation to a 2017 study of Forensic Statistical Tool (“FST”) (a probabilistic genotype software developed and used by the New York City Office of Chief Medical Examiner) that was produced as result of adversarial testing of FST when ProPublica won an order for the release of FST’s source code. It was only after release of the source code that scientific analysts were able to discover errors that caused the software to overestimate the likelihood of guilt. Steven M. Bellovin et al., “Seeking the Source: Criminal Defendants’ Constitutional Right to Source Code,” 17 Ohio State Tech. L.J. 1 (2021). These errors were not detected during any of the previous studies of FST performed without the source code.
Similarly, code errors and miscodes were discovered in STRmix (TrueAllele’s competitor). STRmix had been developed, validated, and used in criminal prosecutions. When the source code was reviewed by independent analysts, they discovered that the program produced false results in 60 cases, according to a report by David Murray titled “Queensland Authorities Confirm ‘Miscode’ Affects DNA Evidence in Criminal Cases,” published March 2015, in the Courier-Mail. Adams (Pickett’s expert) had reviewed STRmix’s code in 2015 and was able to identify potential issues in its source code that negatively affected the functioning of the software, and those issues could not have been learned from any other source.
In the instant case, TrueAllele had been subjected to at least 30 validation studies that concluded the software was reliable – but none were performed with benefit of the source code. Additionally, in six of the studies, Perlin himself participated. This was an impediment to reliable validation of a method: Scientific validity requires scientific evaluation by other scientific groups that did not develop the method, according to the PCAST Report. Consequently, Pickett provided a rational basis for discovery of the source code, the Court determined.
As to prior judicial opinions, the Court observed that 18 courts have rejected calls to allow independent evaluation of TrueAllele’s source code. In Commonwealth v. Foley, 38 A.3d 882 (Pa. Super. Ct. 2012), the court accepted Perlin’s assertion that validation studies are the best tests of the reliability of source codes and reasoned that “scientists can validate the reliability of a computerized process even if the ‘source code’ underlying that process is not available to the public.” But in light of the above discussion, this reasoning is not sound, the Court explained. The Court further explained that subsequent courts placed great emphasis on the observation made in Foley, without further scrutiny, creating a “house of cards.” (See opinion for citations.)
The Court then balanced Pickett’s need for the source code with Cybergenetics’ interest in protecting its trade secrets, emphasizing that when it comes to balancing the rights of the accused against other interests, New Jersey errs on the side of disclosure. State v. Chun, 943 A.2d 114 (N.J. 2008). The Court acknowledged that there “must be teeth to the protective order” and listed several safeguards, including license suspension, disciplinary actions, as well as criminal contempt. To provide guidance for the lower court establishing appropriate parameters of a protective order in this case, the Court referenced Model Protective Order §§ 7.2, 7.3 from the Northern District of California (whose docket includes the most complex and financially consequential patent cases in the world).
In summary, the Court concluded that Pickett had articulated a particularized need for the proprietary source code and related information for use at the Frye hearing by (1) demonstrating a rational basis for ordering the State to attempt to produce it, including through expert testimony supporting the claim for disclosure; (2) providing specificity for the information sought; (3) showing through examples from other jurisdictions that the company’s intellectual property can be safeguarded by a protective order; and (4) demonstrating that source-code review is particularly crucial to evaluating the unique technology at issue.
As a digital subscriber to Criminal Legal News, you can access full text and downloads for this and other premium content.
Already a subscriber? Login
Related legal case
State v. Pickett
|2021 N.J. Super LEXIS 17 (2021)