Algorithms and Criminal Sentencing

HandcuffsI am both proud and embarrassed to admit, I am a full blown data nerd. I am that person who loves to track just about everything on an app so I can look at the data and understand how fast (or slowly) I am running, how much I am spending by category and how much data my daughters use and who is at fault for causing data overage charges. I am sure it comes as no surprise that data analysis regularly manifests itself in the work place; I pride myself on running the firm in a lean and efficient way. When making decisions, I frequently evaluate options based on productivity, profitability, ROI, etc. Ask me, how many unique visitors we had on our website or what percent of the fee dollar we spend on office space and I will happily oblige. Without question, the thing I like most about of data is using historic trends to help predict future performance.

Even I recognize, however, that data has it limits when predicting future human behavior. For example, I cannot expect one of my daughters to behave a certain way, based on her older sister’s behavior. That theory has crashed and burned several times. Advertisers routinely incorrectly predict what I may want from the store based on what I bought.

Using predicative algorithms has also grown in popularity in court rooms across the country.

Courts have begun to rely on algorithms to help predict the likelihood of recidivism, when determining sentences and setting parole. Presumably, courts give heighted sentences to individuals who are likely to reoffend based on the predicative analysis. This term, the Wisconsin Supreme Court will weigh in on the constitutionality of using data to help determine criminal sanctions.

In State v. Loomis, Eric Loomis is challenging a 2013 sentence that was influenced by COMPAS analysis (Correctional Offender Management Profiling for Alternative Sanctions). Mr. Loomis plead guilty to operating a vehicle without the owner’s consent and attempting to flee police; he was sentenced to six years in prison and five years of supervision. In handing down the sentence, circuit judge, Scott Horne, told Mr. Loomis, “[t]he risk assessment tools that have been utilized suggest that you’re extremely high risk to reoffend.” Northpointe, the maker of COMPAS will not reveal how the algorithms score defendants, citing the proprietary nature of the tool, but one question inquires, “[d]id a parent figure who raised you ever have a drug or alcohol problem?” Some legal scholars suggest that these tools inherently discriminate against indigent people and other groups, while some prosecutors argue that using data in this way helps to make more informed decisions. I am not a criminal lawyer, but I will nonetheless be interested to learn if the Wisconsin Supreme Court determines that a defendant’s due process rights are violated when a circuit court uses predictive data when sentencing defendants.

For more information:

http://www.wsj.com/articles/wisconsin-supreme-court-to-rule-on-predictive-algorithms-used-in-sentencing-1465119008

http://theweek.com/articles/627570/many-prison-inmates-futures-depend-racially-biased-algorithms