| Popis: |
Large-scale public sector information systems (PSIS) that administer social welfare payments face considerable challenges. Between 2014 and 2023, an Australian government agency conceived and implemented the Online Compliance Intervention (OCI) scheme, widely referred to as Robodebt. The scheme's primary purpose was to apply digital transformation in order to reduce labour costs and increase recovery of overpayments. Among its key features were a simplified, but inherently erroneous, estimation method called income averaging, and a new requirement that welfare recipients produce documentation for income earned years earlier. Failure by welfare recipients to comply with mandates resulted in the agency recovering what it asserted to be overpayments. This article presents a case study of Robodebt and its effects on over 1 million of its clients. The detailed case study relies on primary data through Senate and other government hearings and commissions, and secondary data, such as media reports, supplemented by academic sources. Relevant technical features include (1) the reliance on the digital persona that the agency maintains for each client, (2) computer-performed inferencing from client data, and (3) automated decision-making and subsequent action. This article employs a socio-technical systems approach to understanding the factors underlying a major PSIS project failure, by focusing on the system's political and public service sponsors; its participants (users); the people affected by it (usees); and the broader economic, social, and political context. Practical and theoretical insights are presented, with the intention of highlighting major practical lessons for PSIS, and the relevance of an articulated socio-technical frame for PSIS. |
