This week a US federal judge unsealed the source code for a software application used by New York City's crime lab to help analyse DNA evidence from crime scenes. The Forensic Statistical Tool (FST) was developed by the office of the city's Chief Medical Examiner. It is used to substantiate the statistical likelihood that someone's DNA profile matches DNA from a sample that may be tiny or degraded, or represent more than one person.
Providing an advanced service that other labs could not match, the New York City crime lab processed samples not only from the New York police but also for about fifty other jurisdictions in the USA, the New York Times reported. This authority is now in dispute, however, as the validity of these specific methods for DNA analysis has come under scientific criticism. As a result, hundreds of criminal cases may have to be revised.
Considerable public interest
The public release of the FST source code (now available on GitHub) was requested by ProPublica, an online newsroom specialising in investigative journalism. Although the source code had already been made available to the defence in a recent case, parties were barred from sharing or discussing it. ProPublica's motion asked the judge to lift this order because of the considerable public interest, after previous attempts to get the code directly from the medical examiner's office were denied.
In another case, two other nonprofit organisations, the Electronic Frontier Foundation (EFF) and the American Civil Liberties Union (ACLU), had asked for access to the source code of a proprietary application for DNA analysis. The developer of the software, however, successfully contended that its code is a trade secret.
These and other examples support the idea that all public agencies should publish for public scrutiny the source code for algorithms that they use in their decision-making.
One year ago, the French association Droits des Lycéens finally managed to obtain the source code of an algorithm that influences students' choice of university after the Baccalauréat exam. It took the association a seven-month battle against the French Ministry of Education, however.
At about the same time, the French Law for a Digital Republic came into force. This states that public agencies should publish the rules defining the major algorithmic processes used in the realisation of their tasks when they are used for individual decisions. It was the first law in France to address source code explicitly. The first applications opened up under this legislation were the tax and benefits calculators used by French policymakers.
Before the implementation of the new law, people had to use the Freedom of Information law if they wanted to get access to the source code of software used in public administration. A computer scientist successfully took this route when he wanted to study the source code of the income tax software used by the Directorate of Public Finances, although he had to take his case all the way to the French Administrative Court.