Letter to Jeff Sessions, Attorney General - Booker Schatz Press DOJ, U.S. Sentencing Commission to Review Automated Methods for Determining Prison Terms

Letter

Dear Attorney General Sessions, Judge Pryor, and Sentencing Commission:

Last month it was revealed that the United States Sentencing Commission (USSC) did not comply with instructions from the Department of Justice (DOJ) in 2014 to prepare a study of algorithm-based risk assessment tools and their uses in federal sentencing and corrections. We are writing to find out why the USSC failed to complete the study and to encourage it to do so within the next six months.

The Sentencing Reform Act of 1984 requires the DOJ's Criminal Division to submit to the USSC an annual report which comments on sentencing guidelines, suggests warranted changes, and assesses the USSC's work.[1] The annual report sent to the USSC on July 29, 2014 requested a study as well as a policy statement on the proper role of risk assessment in the federal criminal justice system.[2] It expressed DOJ's concern about the use of risks and needs assessments in sentencing and corrections. Specifically, it cautioned against the growing trend of basing imprisonment terms on factors such as education level, employment history, family circumstances, and demographic information, instead of the crime committed and proven in court. Three days later, in a speech to the National Association of Criminal Defense Lawyers, then U.S. Attorney General Eric Holder expressed similar concerns and urged the USSC to study the use of data in front-end sentencing and to issue a report with recommendations based on careful, independent analysis.[3] Despite these clear and imperative instructions, the requested report was not released in 2014 or 2015.

On June 15, 2016, the Electronic Privacy Information Center (EPIC) submitted a Freedom of Information Act (FOIA) request to DOJ regarding DOJ's use of various algorithm-based risk assessment tools. (Attachment A) In March 2017 EPIC filed a lawsuit seeking release of the records requested, including a copy of the report requested in 2014 by Attorney General Holder and the Criminal Division. Last month, after having conferred with "knowledgeable Department personnel", DOJ finally acknowledged to the court in a December 14, 2017 Joint Status Report that the USSC simply did not issue the report on risk assessment tools as instructed by Attorney General Holder and the Criminal Division in July 2014. (Attachment B) We are requesting that you provide the Senate Judiciary Committee and the Senate Commerce, Science, & Transportation Committee an update about the status of that report as soon as possible. We would also appreciate information on whether the USSC ever actually attempted to conduct the evaluation as requested, and why the accompanying report was never issued.

The use of risk assessment tools in the criminal justice system is widespread and steadily increasing. There are currently several pieces of legislation before Congress which would require the Bureau of Prisons to use risk and needs assessment systems. At all levels, policy makers and others need to better understand what factors influence decisions in the criminal justice system. Risk assessments make judgments about individuals based on group tendencies, which raises questions of fairness. There are valid concerns that some algorithms may exacerbate racial disparities by relying on socioeconomic factors correlated with race or ethnicity to set bail, determine sentencing, and even contribute to ultimate determinations of guilt or innocence.[4] Given these concerns about fairness and transparency, we are troubled that the Commission has not taken the time to explore this issue.

We urge the USSC to take these concerns seriously and request that within the next six months it complete a study of algorithmic risk assessment tools and also issue a policy statement to guide jurisdictions implementing these tools.

Thank you for your attention to this important issue and for your consideration of our requests.

Sincerely,

Source