Book Review: Automating inequality: How high-tech tools profile, police, and punish the poor The International Journal of Information, Diversity, & Inclusion, 2(4), 2018 ISSN 2574-3430, publish.lib.umd.edu/IJIDI/ IJIDI: Book Review Eubanks, V. (2018). Automating inequality: How high-tech tools profile, police, and punish the poor. New York: St. Martin’s. ISBN 978-1-2500-743-17. 272 pp. $26.99 US. Reviewer: Dawn Betts-Green, Florida State University, USA Book Review Editor: Norda A. Bell, York University, Canada Keywords: automation; data; poverty; surveillance Publication Type: book review ith technological ubiquity and improvements comes the misguided notion that automated systems are more objective and less prone to error than the human element. While this may be true with calculations, when it comes to decisions involving the multilayered human experience, it becomes abundantly clear that this is not universally true. Increasingly, there has been a push for higher levels of automation and decreasing human involvement in the provision of government services such as Social Security, veterans’ benefits, and welfare. Automation provides a way to slash payroll spending and, ostensibly, improve services by making them more objective, and, ultimately, reduce the instance and possibility of fraud. Despite this intention, the systems designed to manage food stamps, housing assistance, and many other services accessed by the poor seem only to increase difficulties faced by the very people the services were designed to help. Virginia Eubanks traces this embracing of technology in terms of poverty services back to the idea of scientific charity and casework. During the 19th and early 20th centuries, caseworkers spent a great deal of time identifying and separating the “deserving poor” from the “unworthy poor.” Eubanks refers to Josiah Quincy III’s 1821 definitions of the “impotent” and “able” poor, which divided them into those that were unable to work (e.g., infants and the physically disabled) and those that were able yet were “just shirking” (17). Using these definitions, the only people who were deemed worthy to receive aid were those completely incapable of work, regardless of their circumstances. She notes that “our new digital tools spring from [these] punitive, moralistic views of poverty” (16) which grew directly out of the scientific casework movement, and its mission to use data-driven methods and standardized criteria to evaluate need—an idea meant to improve service work from the poorhouse model used for centuries prior to this time. Poorhouses, operating in the U.S. since the early 1600s, were the primary mechanism the government used to regulate poverty. Poor people were sent to--some voluntarily entered-- poorhouses in order to receive aid, but these institutions were intentionally harsh in order to deter people from entering as a form of escape. Families were separated, and those who had civil rights (white men) had none while living in the poorhouse. Those who ran the poorhouses often found ways to personally profit from the labor of the inmates, and government funding for the institutions eventually dwindled. Consequently, a more efficient and less expensive model of aid was needed, ultimately engendering the scientific charity movement. Under this model, the poor were no longer spoken of as families or people in need or to be helped, but instead, “cases” to be solved. This lens dehumanized the problems experienced by the poor and further distanced caseworkers from those they were attempting to help. This provided fertile ground for W http://publish.lib.umd.edu/IJIDI/, Automating Inequality The International Journal of Information, Diversity, & Inclusion, 2(4), 2018 ISSN 2574-3430, publish.lib.umd.edu/IJIDI/ ideas such as eugenics to take hold as well as perpetuated the notion that the poor needed to be policed in order to be worthy of the help proffered by the state. In addition to the dehumanization and criminalization of the poor, ideals of what a worthy family unit looked like also solidified. Particularly with New Deal programs, white male earners were privileged and presented as the only really desirable head of household, making women’s work, nonwhite families, and family units without the “traditional” structure suspect in terms of need and worth. These groups were seen as part of the “undeserving” poor, which, in the present, results in the close monitoring and reporting of these families for supposed infractions that may go unnoticed in more “traditional” family units. Eubanks divides the book into chapters focusing on individual cases of automation gone wrong and its harmful effects to the poor. While each case addresses its own specific problems and concerns, the primary thread weaving through them all is the process and problem of dehumanization. The implementation of systems in Indiana designed for medical and state assistance, in California for homeless assistance, and in Pennsylvania for child abuse reporting distanced needy families and individuals from the very people trained to evaluate individual cases and situations to determine and distribute assistance. In each case, the people seeking help are treated only as a set of data. This data is then run through algorithms designed by computer scientists with very little, if any, training at all in the nuances of evaluating human situations. For example, in the case of the child abuse reporting system, there is very little ability to override the inclusion of reports. Human caseworkers can separate false or obviously vindictive reports from those actively needing investigation, but the automated system includes all reports, regardless of veracity, inflating the need of one family over others that may actually warrant monitoring. These reports, whether true or not, are tracked through all of the household members and maintained so that if anyone applies for assistance in the future, the entirety of that history is attached and is factored in the decision for assistance. Regardless of whether it was a simple need for food stamps when they were a small child, these records affect scores given to applicants for aid by the system. A past use of welfare or aid of any kind flags the individual and the family unit for monitoring; thereby artificially driving up applicants’ need scores and flagging them for surveillance. Systems like these rely completely on the information entered, and so assessments are based on criteria that, yes, can be indicators of child neglect or abuse, chronic homelessness, or situational health risks, but can also simply be indicators of poverty. This can often lead to the criminalization of homeless populations and of poor parents, creating situations in which they become less “worthy” of assistance because the very reasons they need help are seen as criminal and self-inflicted based on scores calculated by the systems. For example, the data suggests that you are chronically in need of government assistance because you demonstrate consistent markers of poor parenting and life skills. However, this data is only pulled from interactions with government agencies. Families in higher income classes may also demonstrate the same markers, but, unlike the poor, they have access to private helping professionals and services such as therapists, recovery programs, and child service workers, which are not tracked or reported by the government. Eubanks has written here a useful primer, not only for those in academia researching issues around information systems and access and poverty studies, but also for the general public interested in these topics. In order to improve the situations outlined, those working with and under these systems should be able to grasp the historical and current functioning of the systems. 88 http://publish.lib.umd.edu/IJIDI/, Automating Inequality The International Journal of Information, Diversity, & Inclusion, 2(4), 2018 ISSN 2574-3430, publish.lib.umd.edu/IJIDI/ This work is accessible, an attribute not found in many academic studies, but is of immense value for both general and academic readers. The author does not ignore the implications and problems identified from her research, but instead calls for better collaboration between the system designers and end users, in this case social workers, to ensure that algorithms and systems can better accommodate nuanced and situational need. It is difficult to imagine a way to correct the problems Eubanks so eloquently outlines in this book. Automation and technology are not going to fade away; they will, in fact, increase exponentially in coming years. The only solutions lie in shifting our notions of poverty and technology. As the author explains, technology is not neutral. These systems are designed by fallible humans and can be improved by increasing understanding of the intended end users and beneficiaries of the systems and reminding the designers that their systems are “not aimed at data points, probabilities, or patterns, but at human beings” (213). We must also, as a society, shift the rhetoric about poverty. The poor and impoverished are not “a people apart,” living in distant, separate areas of the world, entirely divorced from the economic and daily interests of “normal” people. Poverty is also not a monolithic entity full of people who caused their situation by neglect or apathy; it is a “borderland” (205), adjacent to every single person regardless of socioeconomic status, beliefs, or political leanings. It is “we” rather than “us vs. them.” Finally, as with so many problems, one of the most important solutions is to listen to the people the systems are being designed to help. Designing systems with a “savior” mentality will not benefit anyone save the egos of those employing it. Dawn Betts-Green (cdb07f@my.fsu.edu) is a PhD candidate in Information at Florida State University's iSchool. Her dissertation focuses on LGBTQ resources and patrons in rural libraries in the U.S. South. Her research interests include diversity in LIS education; graphic novels; intellectual freedom; social justice and radical librarianship; young adult and children’s services; rural and small public libraries in the U.S. South; and information needs of diverse populations. She is also a Point Foundation LGBTQ Scholar. 89 http://publish.lib.umd.edu/IJIDI/, mailto:cdb07f@my.fsu.edu