STORY:"Machine bias: Federal Judge Unseals New York Crime Lab’s Software for Analyzing DNA Evidence," by Lauren Kirchner, published by ProPublica on October 20, 2017.
SUB-HEADING: "We asked the judge make the source code public after scientists and defense attorneys raised concerns that flaws in its design may have resulted in innocent people going to prisoning to prison."
SUB-HEADING: "Investigating Algorithmic Injustice."
SUB-HEADING: "New York City’s crime lab has been a pioneer nationally in analyzing especially difficult DNA samples. But the recent disclosure of the source code for its proprietary software is raising new questions about accuracy."
SUB-HEADING: ProPublica seeks source code for New York City's disputed DNA software: We're asking a federal court for the code behind a technique that critics say may have put innocent people in prison."
GIST: "A federal judge this week unsealed the source code for a
software program developed by New York City’s crime lab, exposing to
public scrutiny a disputed technique for analyzing complex DNA evidence. Judge Valerie Caproni of the Southern District of New York lifted a protective order in response to a motion by ProPublica,
which argued that there was a public interest in disclosing the code.
ProPublica has obtained the source code, known as the Forensic
Statistical Tool, or FST, and published it on GitHub; two newly unredacted defense expert affidavits are also available. “Everybody who has been the subject of an FST report now gets to find
out to what extent that was inaccurate,” said Christopher Flood, a
defense lawyer who has sought access to the code for several years. “And
I mean everybody — whether they pleaded guilty before trial, or whether
it was presented to a jury, or whether their case was dismissed.
Everybody has a right to know, and the public has a right to know.” Caproni’s ruling comes amid increased complaints by scientists and
lawyers that flaws in the now-discontinued software program may have
sent innocent people to prison. Similar legal fights for access to
proprietary DNA analysis software are ongoing elsewhere in the U.S. At
the same time, New York City policymakers are pushing for transparency
for all of the city’s decision-making algorithms, from pre-trial risk
assessments, to predictive policing systems, to methods of assigning
students to high schools. DNA evidence has long been a valuable tool in criminal
investigations, and matching a defendant’s genetic material with a
sample found on a weapon or at a crime scene has impressed many a judge
and jury. But as new types of DNA analysis have emerged in recent years
to interpret trace amounts or complex mixtures that used to be dismissed
as hopelessly ambiguous, the techniques are coming under fire as overly
ambitious and mistake-prone. An article ProPublica co-published with The New York Times
on Sept. 4 detailed the growing doubts about the Forensic Statistical
Tool, which New York City created to determine the likelihood that a
given defendant’s DNA was present in a mixture of multiple people’s
genetic material. According to the crime lab’s estimates, FST was used
to analyze crime-scene evidence in about 1,350 cases over about 5 1/2
years. It was phased out at the beginning of this year in favor of a
newer tool. A coalition of New York City defense lawyers has called for a review
of all cases that may have been affected by either FST or a second
disputed analysis method, called high-sensitivity DNA testing. The state
inspector general, which acts as the lab’s ombudsman, has received the
lawyers’ request but has not yet announced whether she will launch an
investigation.The crime lab, which is part of the Office of the Chief Medical
Examiner, did not oppose ProPublica’s motion, but maintains its support
of its technology......... See the Source Code
You can see the code for the Forensic Statistical Tool on GitHub. ProPublica’s motion, filed on Sept. 25
with the help of the Media Freedom and Information Access Clinic at
Yale Law School, argued that the judge should vacate that protective
order because of “the profound importance of this technology to the
integrity of the criminal justice system, and the overriding public
interest in transparency.” “This ruling finally enables ProPublica to gain access to the code in
order to report on this matter of vital public concern,” said Hannah
Bloch-Wehba, a supervising attorney in the MFIA clinic, following the
judge’s order. “As law enforcement agencies increasingly rely on
algorithmic tools in the criminal justice system, it is all the more
important that the press and public have access to the information
critical to understand what the government is doing and hold it
accountable.” FST was invented by employees of the crime lab and programmed by
software consultants. The lab began using it in 2011 to analyze complex
mixtures of DNA left behind at crime scenes. About 50 jurisdictions as
far away as Bozeman, Montana, and Floresville, Texas, also sent samples
to New York City for testing. When defense attorneys challenged FST’s
results in court and sought access to the program’s source code, the
crime lab has previously refused, saying it was proprietary. Although almost all judges have allowed FST results as evidence in
court, one state judge, Mark Dwyer of Brooklyn, ruled them inadmissible
in two cases in 2014. Dwyer, now presiding in Manhattan, excluded FST
evidence from two more cases this week. While prosecutors in both cases
said DNA evidence analyzed with FST showed that the defendants violated
gun possession laws, Dwyer said in court on Oct. 16 that his doubts
about the program’s acceptance in the scientific community persist,
especially since the New York lab is no longer using it, and no other
lab has adopted it. New information about the development of the FST source code and some
of its purported weaknesses surfaced this past July in the cases before
Dwyer in an affidavit by Eugene Lien, a technical leader in the DNA
lab, whom the prosecution was using as an expert. After the lab started
using FST for casework in early 2011, he and his colleagues discovered a
problem with the program’s math that could skew a test’s results,
according to Lien. “Because of this, the FST program was taken offline
and portions of the software were re-coded,” he wrote. The lab did a “performance check” of the new version before resuming
casework with it in July 2011, he went on, but lab officials did not
inform the state oversight commission about the change, nor did they run
another full validation study on the program. The letter to the state inspector general from the group of
defense lawyers cited Lien’s account, saying it contained “damning
admissions” about the lab’s lack of transparency. They also theorized
that the recoding Lien described could itself have led to one problem
identified by Adams — the exclusion of potentially valuable data from
FST’s calculations of likelihood ratios. Characterizing Adams’
criticisms as merely cosmetic rather than substantive, the lab has
contended that FST calculations were reliable. Besides ongoing criminal cases in New York City involving FST,
Caproni’s decision to unseal the source code may also affect another
legal fight for access to a proprietary DNA software system. The American Civil Liberties Union and the Electronic Frontier Foundation
intervened in a case in California’s appeals court on Sept. 13 in
support of a defendant’s right to review the source code behind a
commercially available DNA analysis program called TrueAllele. “It’s a major credit to the court, the parties and ProPublica that
the source code used in Mr. Johnson’s case will now be subject to public
scrutiny,” said Brett Max Kaufmann, a staff attorney for the ACLU who
is working on the California appeals case. “We urge other courts to
follow this example when hearing cases involving similar types of
evidence.” Outside the courtroom, some New York City lawmakers are seeking more
public review of algorithms and their impacts. On Oct. 16, the New York
City Council’s Committee on Technology held a hearing about a proposed bill
calling for all city agencies to publish online the source codes for
algorithms that they use in decision-making. As an example of the danger
of relying on algorithms, witnesses and a committee report cited
ProPublica’s 2016 investigation that found racial bias in a software program used by courts to decide whether it’s safe to let defendants out on bail. “These tools seem to offer objectivity, but we must be cognizant of
the fact that algorithms are simply a way of encoding assumptions, and
their design can be biased, and that the very data they possess can be
flawed,” the bill’s author and the committee chair, James Vacca, said at
the hearing. “I have proposed this legislation not to prevent city
agencies from taking advantage of cutting-edge tools, but to assure that
when they do, they remain accountable to the public.” The committee heard from defense lawyers and others who support the
bill as well as representatives from Mayor Bill DeBlasio’s Office of
Data Analytics and the city’s Department of Information Technology and
Telecommunications, which both oppose it in its current form. After the
hearing, Vacca told ProPublica that he would revise the bill to address
criticisms he had heard about confidentiality concerns, and also to
clarify that the proposal applies to both programs developed by
third-party vendors and software developed in-house by city employees.
Vacca said he is determined to pass a law on this issue before the end
of his term."
The entire story can be found at:
https://www.propublica.org/article/federal-judge-unseals-new-york-crime-labs-software-for-analyzing-dna-evidence
PUBLISHER'S NOTE: I am monitoring this case/issue. Keep your eye on the Charles Smith Blog for reports on developments. The Toronto Star, my previous employer for more than twenty incredible years, has put considerable effort into exposing the harm caused by Dr. Charles Smith and his protectors - and into pushing for reform of Ontario's forensic pediatric pathology system. The Star has a "topic" section which focuses on recent stories related to Dr. Charles Smith. It can be found at: http://www.thestar.com/topic/c harlessmith. Information on "The Charles Smith Blog Award"- and its nomination process - can be found at: http://smithforensic.blogspot. com/2011/05/charles-smith-blog -award-nominations.html Please
send any comments or information on other cases and issues of interest
to the readers of this blog to: hlevy15@gmail.com. Harold Levy;
Publisher; The Charles Smith Blog;
The entire story can be found at:
https://www.propublica.org/article/federal-judge-unseals-new-york-crime-labs-software-for-analyzing-dna-evidence
PUBLISHER'S NOTE: I am monitoring this case/issue. Keep your eye on the Charles Smith Blog for reports on developments. The Toronto Star, my previous employer for more than twenty incredible years, has put considerable effort into exposing the harm caused by Dr. Charles Smith and his protectors - and into pushing for reform of Ontario's forensic pediatric pathology system. The Star has a "topic" section which focuses on recent stories related to Dr. Charles Smith. It can be found at: http://www.thestar.com/topic/c