Menu
Contact

Back to Newsroom

23-Jun-2020

The government wants to take away your right to use independent forensic software


Washington, DC

On September 17, 2019, California Democrat Mark Takano introduced HR 4368, the inaptly named "Justice in Forensic Algorithms Act." The Wall Street Journal (WSJ) wrote about the pending measure on June 1, 2020. The House bill would harm scientific innovation, replace DNA truth with State-approved software, and transfer judicial powers from experienced impartial courts to an unaccountable federal agency.

DNA evidence has the power to bring the light of scientific truth into the courtroom, promoting justice for both the guilty and innocent. Most DNA items are a mixture of two or more people. Human analysts find these data hard to decipher, so they rely on sophisticated computer software for accurate interpretation.

Government failure, small-business innovation

Twenty years ago, the federal government imposed ineffective DNA mixture software on forensic science. The State’s software either gave the wrong answer, or couldn't solve the problem. The National Institute of Standards and Technology (NIST) Applied Genetics Group compounded the disaster with procedures that forced forensic failure in hundreds of thousands of criminal cases. Widespread injustice followed, acquitting violent criminals and imprisoning innocent people.

Twenty years ago, private-sector Cybergenetics solved the DNA mixture problem. The Pittsburgh-based innovator's TrueAllele® software can reliably "unmix" mixture data for accurate DNA match statistics. The company transparently publishes its methods so that others can understand the scientific breakthrough. TrueAllele crime laboratories and Cybergenetics casework services find truth in formerly "inconclusive" DNA evidence, and reverse injustice.

DNA evidence and the law

For nearly a hundred years the courts have been a gatekeeper, deciding the reliability of scientific evidence. When a defendant challenges a new forensic method, impartial judges hear from lawyers and scientists to rule on its admissibility, shielding jurors from unreliable evidence.

Evidence laws and the courts safeguard trade secrets, fostering needed innovation that makes the world a better place. Scientific transparency is ensured by disclosing underlying algorithms, sharing testing results, and providing software that others can test. Publicly releasing confidential source code doesn't help assess reliability, because the text isn’t used in testing. But its release can wreck an innovative company.

On April 13, 2006, Pennsylvania dentist John Yelenic was slashed to death in his Blairsville home by an unknown intruder. Suspicion fell on State Trooper Kevin Foley, who was living with Dr. Yelenic’s estranged wife Michele. Ineffective government forensic software weakly linked Foley to a small amount of mixed DNA under Yelenic’s fingernails. Making its first criminal justice appearance, TrueAllele found a far stronger match statistic of 189 billion, connecting Foley to the murder.

In 2009, the defense challenged TrueAllele. Relying on scientific studies, Presiding Judge William Martin admitted TrueAllele as reliable evidence that the jury could hear. Foley was convicted, and sentenced to life in prison. "Yelenic provided the most eloquent and poignant evidence in this case," said prosecutor Anthony Krastek. "He managed to reach out and scratch his assailant."

Foley appealed the verdict. In February, 2012, the Pennsylvania Superior Court sustained his conviction. They noted that "TrueAllele has been tested and validated in peer reviewed studies." Regarding trade secrets, "scientists can validate the reliability of a computerized process, even if the ‘source code’ underlying that process is not available to the public," wrote the court. "TrueAllele is proprietary software. It would not be possible to market TrueAllele if it were available for free."

What the bill does

The pending HR 4368 bill has five main provisions:

  1. The bill would annul trade secrets, eliminating innovation in forensic software.
  2. It would shift responsibility for forensic evidence admissibility away from impartial judges, transferring historic judicial powers to a federal executive agency.
  3. The diverse scientific community that tests forensic software would be replaced by a monolithic government unit.
  4. Federal testing standards for forensic software that were instituted five years ago would be rebranded as if they were something new.
  5. Legal defense teams could disclose company-crippling trade secrets, without incurring any consequences.

The bill would radically alter the federal rules of evidence and trade secret law. Judges, with help from lawyers and scientists, could no longer decide the reliability of forensic evidence. Forensic innovation would lose trade secret protection, abolishing software that finds truth through better science. Prosecutors and defenders could not present their own DNA evidence in court. Instead, they would be forced to accept the results of limited government-sponsored software.

A detailed side-by-side analysis steps through each line of the bill, comparing current status with the impact of its proposed changes. Supporting facts are provided.

How the bill affects you

Suppose you have been accused of a crime, and the State claims to have "DNA proof" of your guilt. How would HR 4368 affect your rights? Would the bill work in your favor or work against you? Let’s consider each provision of the pending legislation.

First, the bill would abolish trade secret protection for forensic algorithms. You could get the source code for the DNA interpretation software. But you can already get source code confidentially from the developer; you’ve gained nothing. However, without trade secret protection, you lose software alternatives for challenging the State. The private-sector innovators who can help you have all left forensic science.

Next, the admissibility rules would have changed. You can no longer challenge the State-sponsored software used against you, even when it’s dead wrong. The State has censored alternatives that might prove your innocence; you can no longer defend yourself. Lawyers and judges can’t help you; science and software can’t help you. HR 4368 silences helpful DNA evidence, and cancels your due process right to be heard.

Third, the bill would let you test DNA software used against you. But you can already do that – DNA software companies will give you their testable software at no cost. The bill gives you nothing new. But under HR 4368, State "testing" approves the State’s DNA software used against you, and eliminates alternatives that might help you.

A fourth bill provision would establish forensic testing standards. But these forensic DNA standards have already been around for five years. HR 4368 changes nothing.

Finally, the bill would allow defense teams to disclose forensic software trade secrets, without facing any consequences. Lawyers could destroy innovative companies and put them out of business. Then, better software that might help you would never be developed. Instead, HR 4368 puts your life and liberty in the hands of a State-sponsored software monopoly that you cannot counter or challenge.

The impact of independent forensic software

TrueAllele is an innovative, accurate, neutral, and transparent truth-seeking tool for criminal justice. The software provides a reliable scientific alternative to failed government-sponsored DNA analysis software programs. TrueAllele is used by both prosecutors and defenders, and by police and innocence groups. The technology helped identify victim remains in the World Trade Center mass disaster.

TrueAllele helps the innocent. Johnny Lee Gates, a Southern black man, spent over forty years in Georgia prison for a crime he most likely did not commit. Judges decried the case’s racial injustice, but under the law only TrueAllele could set him free. Based on newly discovered DNA evidence, he was released on May 15, 2020.

If passed, HR 4368 would block justice from using better science, and keep many innocent men and women in prison. TrueAllele has helped exonerate a dozen innocent men. Mr. Gates resides in the district of the Hon. Hank Johnson, chairman of the bill’s Judiciary Subcommittee on Courts, Intellectual Property, and the Internet.

The police use Cybergenetics’ free screening service to find DNA leads when crime labs can’t. The TrueAllele software helps focus their investigation on people who are connected to crime scene biological evidence. In the company’s last hundred police screenings, TrueAllele has cleared a hundred and fifty suspects through DNA.

TrueAllele lets prosecutors follow the DNA to spare the innocent and convict the guilty. In 2010, California teenager Norma Lopez was murdered as she fought off a sexual attacker. Blocked by ineffective government software, the crime laboratory reached out to Cybergenetics. TrueAllele connected mixed DNA on the girl’s earring to Jesse Torres. On March 13, 2019, a Riverside jury convicted Torres; they recommended the death penalty. Ms. Lopez lived in the district of Rep. Takano, sponsor of HR 4368.

Putting government over people

The pending bill harms innocent people and due process rights – but it helps NIST. The bill would give the bureaucracy more funding and power, at the expense of science and justice. The federal agency misleads the public, and doesn’t understand the technology they seek to control. NIST favors foreign imitators over American innovators.

HR 4368 would let NIST anoint a single State-sponsored DNA software solution, silencing legal challenges and scientific alternatives. Under the bill, NIST would waste tens of millions of taxpayer dollars on outdated, overpriced, unneeded "studies" that delay criminal justice and suppress evidence.

The GAO report

In May of 2020, the Government Accountability Office (GAO) issued a report on forensic technology, focusing on algorithms used in federal law enforcement. They included probabilistic genotyping (PG), the powerful TrueAllele method for analyzing complex DNA evidence Cybergenetics invented twenty years ago. PG uses data variation for accurate identification, whereas failed government methods discard data and its variation.

Dr. Mark Perlin, Chief Scientist and Executive of Cybergenetics, served on the GAO’s panel of experts this January. In March, he gave the GAO written answers to their questions about PG methods. In April, he sent them line-by-line comments on the PG section of their draft report, correcting many errors, and providing an accurate replacement section and figure.

The WSJ interviewed Dr. Karen Howard, author of the report and Director of the GAO's Science, Technology Assessment and Analytics team. Her comments about forensic algorithms are not relevant to PG software for DNA analysis.

"It is a well-known issue that it’s difficult to know what’s under the hood of an algorithm," said Dr. Howard. Her statement may apply to some algorithms, such as the deep-learning neural networks used in facial recognition. However, it’s not true for most computer algorithms, which are explained in textbooks and scientific articles. PG algorithms are well-documented in many papers, reports, user manuals, courses, and on-line lectures.

"Does it have any bias baked in?" asked Dr. Howard. Her concern may be with social science software used for crime prediction, bail decisions, or facial recognition. These algorithms often rely on historical trends having ethnic or socio-economic bias. However, PG software does not have such bias; LR match statistics adjust for ethnicity. Automated PG software eliminates human decision-making that might taint DNA results.

Dr. Howard said it is unclear how some algorithms actually work, "particularly an AI-based algorithm that is learning as it goes, you don’t always know what it’s learning." That opacity certainly arises with deep-learning AI methods, whose inner workings are hidden. But PG software doesn’t learn from past cases. The programs use transparent statistical methods to analyze DNA data. They do not "learn as they go."

Why source code isn’t needed

In a recent podcast, Rep. Takano said, "a forensic algorithm is source code." That is incorrect. 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.

At the end of the WSJ article, Dr. Perlin explained why source code isn’t needed (in fact, it can’t even be used) to test the accuracy of forensic algorithms: "You don’t learn how a car works by reading its blueprints; you take it for a test run." Lawyers read, scientists test.

Dr. Perlin told the WSJ that leaked trade secrets can bring innovation to a standstill. "Under this legislation," he asked, concluding the article, "what company would spend ten years developing reliable technology that can help society?"


Links


News and Media Reports


Pending House Bill HR 4368


GAO Forensic Algorithms Report


  • Forensic technology: algorithms used in federal law enforcement - GAO
  • Dr. Perlin's comments on draft GAO report - Notes on draft section
  • Dr. Perlin's rewrite of PG section for GAO report - New section
  • Cybergenetics reworked PG figure for GAO report - New figure

NIST Reliability Issues


TrueAllele Criminal Cases


  • The Blairsville slaying and the dawn of DNA computing - Book chapter
  • Commonwealth of Pennsylvania v Kevin James Foley - Appellate ruling
  • Cybergenetics DNA evidence helps convict California killer of teenager Norma Lopez - News
  • Johnny Lee Gates freed after 43 years in prison after TrueAllele reanalyzes DNA evidence - News

Cybergenetics Resources


Back to top