In the first few weeks of a nationwide shutdown spurred by the COVID-19 pandemic, former Attorney General William P. Barr issued a memorandum to the Federal Bureau of Prisons directing the use of a predictive risk assessment algorithm to prioritize early release for eligible prisoners. The algorithm, PATTERN, was developed in 2019 after the passing of the First Step Act of 2018, a piece of legislation aiming at criminal justice reform. PATTERN is a gender-specific risk assessment algorithm designed to predict recidivism – or the likelihood of re-arrest upon release – among federal prison inmates. The algorithm works by assigning inmates with a risk score of general recidivism (re-arrest for any crime) and violent recidivism (re-arrest for a violent crime) based on factors like severity of offense, criminal history, history of violence, and infraction convictions. PATTERN is the latest in a series of technologies deployed in connection with the current practice of mass incarceration in the U.S. Such technologies are attractive, according to advocates, because they are cheaper, more objective, and less biased than their human equivalents.
As scholars like Ruha Benjamin have pointed out, however, algorithms are not as neutral as their advocates claim. Rather, they automate racial injustice through relying on historical crime data (such as contact with police, involvement in gun violence, and prior arrests and convictions) which serve as proxies for race. The problem is not only that carceral algorithms reproduce racial injustice by drawing on data collection practices which have been historically biased, but also that their supposed objectivity and neutrality protect them from accountability. As LAPD spokesman Josh Rubenstein said of his police department’s PredPol algorithm, “It is math, not magic, and it’s not racist.”
The Department of Justice (DOJ) has expressed similar sentiments about PATTERN, claiming that “PATTERN’s predictive performance reduces bias and aims to improve parity across race and ethnic classifications.” Because PATTERN is designed to assign the same risk score to two offenders of different races with similar or identical criminal profiles, the DOJ suggests the algorithm is not racially biased. Despite this claim, the predictive outcomes of PATTERN indicate that it does not actually achieve racial parity. According to data provided by the National Institute of Justice, among a sample of the federal prison population, only 7 percent of Black males received a minimum-risk score compared to 30 percent of White males, 53 percent of Black males received a high-risk score compared to 29 percent of White males, and 10 percent of Black females received a high-risk score compared to 5 percent of White females. In application, this means that PATTERN predicts that Black males and females are at a higher risk of recidivism and are less qualified for early release or home confinement than their white counterparts. The algorithm can then be used to justify the continued incarceration of a population which faces the highest rates of imprisonment in this country.
Critics of carceral technologies like PATTERN are keen to point out that they are biased or unjust without always clarifying what this bias or injustice amounts to. We critics allow such words to speak for themselves as though their meaning were readily apparent to everyone. At the risk of obfuscation, we might ask, what does bias, specifically racial bias, really amount to in the case of carceral technologies?
I believe we can venture an answer to this question by turning to a concept developed by the Cameroonian philosopher, Achille Mbembe. Mbembe draws on the idea of “necropolitics” to describe the racializing power at work in historical practices like slavery in the U.S. and contemporary colonial practices like the occupation of Palestine. Racializing power in these contexts consists in a politics of death which relies on racialized distinctions between “who is disposable and who is not.” Death here does not only amount to actual, material death, but to what Orlando Patterson has called “social death,” whereby certain populations are treated as though they were already dead or non-existent.
The racial bias at work and at stake in carceral technologies like PATTERN amounts to a politics of death which disproportionately exposes and subjects Black, Latinx, and Indigenous Americans to material and social death. This is not only because the carceral system itself is necropolitical, extinguishing the social, material, and civic lives of its subjects, but compounding this is the fact that the populations who suffer disproportionate rates of incarceration are also those who are most at risk of serious infection and/or death due to COVID-19. The overcrowding of federal and state prisons has made them vulnerable sites for the contraction and spread of the virus, and as of April 16, 2021 at least 2,990 incarcerated people and prison staff have died from COVID-19.
PATTERN contributes to the racialized allocation of death because it assigns Black prisoners with higher risk scores for recidivism and is then used to justify their continued incarceration or exposure to material and social forms of death (whether through solitary confinement, contraction of a deadly virus, the death penalty, neglect by medical practitioners, or at the hands of correctional officers or other inmates). PATTERN risk scores are not simply predictions of criminality – they are judgments about who is disposable and who is not.
Exercises of power rarely go unchallenged. Inmates and detainees have staged numerous demonstrations and hunger strikes since March 2020 to contest the deadly conditions they face in prisons and detention centers in the midst of the pandemic. One detainee participating in a hunger strike at Stewart Detention Center in Lumpkin, Georgia poignantly summed up the stakes of their demonstration, “We don’t want to die here.” This mournful plea contests the necropolitics of carceral technologies, expressing a hope for a future of breath not foreshortened by state-sanctioned death.
Bonnie Sheehey
Dr. Bonnie Sheehey is an Assistant Professor of Philosophy at Montana State University. Her research draws on diverse methods of critique to address and challenge issues of racial injustice at the intersection of technology and the criminal justice system in the U.S. In addition, her work pursues the possibility of a form of critique grounded in hope to push the status and task of critical inquiry beyond its reliance on negative forms of judgment.
