Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor

Eubanks describes visiting Kim and Kevin Stipes in Tipton, Indiana. Their daughter Sophie has cerebral palsy, among other conditions, which in total cost the family more than $6000 a month in medical expenses. After a mix-up, the local Family and Social Services Administration (FSSA) cut of Sophie Stipes’s Medicaid benefits for inaction, even though the office failed to inform the Stipes of mandatory paperwork. Sophie’s case was only addressed after the Stipeses protested at the governor’s office and received TV coverage.

The Stipeses were emblematic of what happens when welfare is privatized. In 2006, Indiana Republican governor Mitch Daniels “instituted a welfare reform program that relied on multinational corporations to streamline benefits applications, privatize casework, and identify fraud” (45). Daniels’s actual agenda was funneling public taxpayer funds to private business: “By the time of the automation, the FSSA had halved its public workforce and was spending 92 percent of its budget buying services from outside vendors” (47).

After Indiana instituted an automated referral system made by IBM and ACS, the state’s welfare system fell apart: Private call center employees were not equipped with the proper training or emotional resiliency required of their new jobs, which meant applicants were stranded. Errors surged, as a “result of inflexible rules that interpreted any deviation from the newly rigid application process, no matter how inconsequential or inadvertent, as an active refusal to cooperate” (51). The label “refusal to cooperate” is a common welfare decision used to deny the claims of many applicants, regardless of facts.

Daniels’s administration argued that caseworkers spent too much time on manual data collection, and that they had too much leniency to help their clients avoid red tape—empathy and leeway that Daniels termed fraud. The initial goals of the automated system were to maximize efficiency and eliminate fraud by shifting from a relationship-based system to a task-based one. As a result, “No one worker had oversight of a case from beginning to end” as the system was “designed to sever” caseworker-client relationships (53). The system also severed relationships between offices. This meant departments no longer shared files, so applicants received demands for decades-old, irrelevant documents, as well as repetitive requests for documents they had just submitted to a different office. The lack of data sharing and coordination was far below any proper data management standards. By the summer of 2009, the error rate tripled: “12.2 percent of those applying for food stamps were being incorrectly denied” (63). Moreover, the new system wasn’t cracking down on welfare fraud at all.

Eubanks also visited Jane Porter Gresham, who worked as an Indiana FSSA caseworker for 26 years. After automation, Gresham lost access to her cases; she now only responded to individual tasks assigned by the Workflow Management System (WFMS): “Staff were responsible for responding to tasks that dropped into their queue in the WFMS. Cases were not handled in the county where applicants lived” (62). The process was dehumanizing as caseworkers lost any sense of vocation or professionalism and instead “became slaves to the task system” (63). In Muncie, Indiana, affected people gathered for a Town Hall Meeting to protest the new automated system, which discriminated against the poor and the disabled: “In the old days, they used to be able to call their caseworker and find out what piece of paper they were missing, or what signature line they forgot to sign, or whatever the problem was. Now they don’t have anyone to call” (70). Not only did the new system create additional errors, but it also had no recourse of fixing errors or giving alternative routes to applicants who questioned decisions.

No mechanism could force private companies to correct, upgrade, or even support their system. When FSSA demanded that IBM submit a “corrective action plan to improve 36 different service deficiencies,” IBM argued that “nothing in their contract required that they respond to a corrective action plan” (71). Even when Indiana sued IBM in 2010, IBM won the breach of contract suit and was awarded $52 million.

Evidence from the trial is pretty damning. According to the state’s representatives, “The IBM Coalition workers were so far behind in processing applications that they would often recommend denial of an application to make their timeliness numbers look better, but then would tell the applicant to appeal the decision” (72-73). Additionally, procedures unfairly targeted Black women, perpetuating a longstanding racist and sexist view. Historically, welfare rules were selectively interpreted to block African American women from claiming benefits until the rise of the welfare rights movement in the 1970s. Now, even when “eligibility modernization was tested on primarily white communities, Black families still felt its worst effects” (80), like the increasing difficulty of accessing food stamps.

Today, Indiana uses a hybrid eligibility system, which has electronic data processing systems as well as in-person caseworkers, although the priority is on the former. Fewer mistakes are filed, but it is unclear whether this is from user and caseworker disillusionment (and subsequent abandonment) of the system, rather than the system’s eventual success.