Wisconsin's Supreme Court justices unanimously agreed to allow the state to continue using a computer program to help judges determine how criminal defendants should be sentenced.
The state's high court ruled Wednesday that the computer program, COMPAS, was appropriately used in the case of Eric Loomis, who is serving six years in prison for driving a stolen vehicle and fleeing from police.
All seven justices unanimously upheld the decisions of the state’s lower courts, affirming the sentencing process for Loomis and the use of COMPAS. The court denied Loomis' request for a new sentencing hearing.
Anne Walsh Bradley wrote the opinion and Shirley Abrahamson and Patience Roggensack filed separate concurring opinions.
Here are some questions and answers about the case and the debate over using predictive computer programs in criminal sentencing.
What was Loomis’ argument in this case?
Loomis said that the use of COMPAS to sentence him, in part, violated his due process rights because it considers gender when analyzing how likely someone is to re-offend, it does not allow for an individualized sentence and it violates a defendant’s right to be sentenced based upon accurate information, in part because the proprietary nature of COMPAS prevents him from assessing its accuracy.
What is COMPAS?
COMPAS, which stands for Correctional Offender Management Profiling for Alternative Sanctions, is a software program made by Northpointe, Inc., a private company. Wisconsin's Department of Corrections started using the program in 2012.
How does it work?
When most defendants are booked in jail, they respond to a COMPAS questionnaire. Their answers are fed into the COMPAS software to generate several scores including predictions of “Risk of Recidivism” and “Risk of Violent Recidivism.”
The risk scores are displayed in a bar chart, with three bars that represent pretrial recidivism risk, general recidivism risk and violent recidivism risk. Each bar indicates a defendant's level of risk on a scale of one to ten. Judges are given these scores and use them in determining how to sentence a defendant.
So, is a software program deciding how long someone goes to jail in Wisconsin?
Not exactly. COMPAS is just one factor of several that a judge might consider when sentencing someone. The justices essentially said it is intended to be one useful tool in sentencing, not the only tool.
What did the justices say in their opinion?
Ann Walsh Bradley wrote that, overall, software like COMPAS gives the courts a more comprehensive picture when determining a defendant's sentence. She affirmed the need for judges to understand the program's limitations and advised them to "exercise discretion when assessing a COMPAS risk score with respect to each individual defendant."
"Ultimately, we disagree with Loomis because consideration of a COMPAS risk assessment at sentencing along with other supporting factors is helpful in providing the sentencing court with as much information as possible in order to arrive at an individualized sentence,” Bradley wrote.
The court rejected the argument that COMPAS discriminates between defendants based on gender:
“COMPAS's use of gender promotes accuracy that ultimately inures to the benefit of the justice system including defendants,” Bradley wrote. "Considering gender in a COMPAS risk assessment is necessary to achieve statistical accuracy. The State argues that because men and women have different rates of recidivism and different rehabilitation potential, a gender neutral risk assessment would provide inaccurate results for both men and women."
In her separate, concurring writing, Roggensack clarified that the court was not advising lower courts to rely on COMPAS alone for a sentence it prescribes.
"While our holding today permits a sentencing court to consider COMPAS, we do not conclude that a sentencing court may rely on COMPAS for the sentence it imposes," she wrote.
What are the problems with COMPAS?
The algorithms and information COMPAS uses to make its risk assessments are proprietary and therefore private, shielding it from scrutiny over its accuracy or fairness. Critics say that makes it an unreliable tool.
"It is a big problem that the algorithm used to decide whether a person is 'high risk' is kept secret from the public, the parties and the court," said Cecelia Klingele, a University of Wisconsin Law School professor who studies sentencing policy.
"Being labeled 'a highly probable risk' by COMPAS can have major, real-life consequences in terms of what sentence a person gets, how long that sentence is, and what restrictive conditions are placed on anyone serving a sentence in the community," she said. "But what that 'risk' means is a mystery.
"The public and the courts have to blindly trust the company’s own assurances about the reliability of its product. Maybe those assurances are legitimate, but without greater transparency, there is no way to know. Meanwhile, the liberty of real people hangs in the balance."
What have studies said about this tool?
A ProPublica review of risk scores in Florida found that only 20 percent of people predicted to go on and commit more violent crimes actually did so. It also found that the program treats black defendants different than whites.
In that review, ProPublica also noted that:
"Wisconsin has been among the most eager and expansive users of Northpointe’s risk assessment tool in sentencing decisions. In a 2012 presentation, corrections official Jared Hoy described the system as a 'giant correctional pinball machine' in which correctional officers could use the scores at every 'decision point.'"
ProPublica's report noted that "Wisconsin has not yet completed a statistical validation study of the tool," which is a study on how statistics help COMPAS draw its conclusion about a person’s likelihood to reoffend.
ProPublica added that the state has not said when such a study might be released.
Do other states use COMPAS or programs like it?
Yes, but at often different points during a criminal case. COMPAS is not the most commonly used program, said Klingele.
"The National Center for State Courts has been encouraging corrections to adopt tools for predicting individual defendants’ risks of re-offense for a number of years. In some states, risk assessment results are only used within corrections, to set security levels or decide whether to grant parole," she said.
In Wisconsin, risk assessment scores are presented to judges in order to influence sentencing, which is a more controversial practice, she said.
"Traditionally courts have sentenced people based on the punishment they deserve for what they’ve already done, and not for what they might (or might not) do in the future."