Abstract Scholars across Computer-Supported Cooperative Work (CSCW) increasingly focus on the topic of care when investigating data-driven technologies in contexts of re-humanizing technology design and usage. Previous studies have shown how care work eludes complex bureaucratic systems shaped by data, digitalization, and a restrictive political agenda. This research aims to understand how asylum stakeholders enact care as an aspect of asylum casework, while navigating what is largely acknowledged by NGOs, nation states, and the EU to be a broken asylum system (von der Leyen). We investigate care as a relational aspect of casework in which knowledge and technology of the implicated caseworker and asylum seeker are attuned to one another in a way that takes the unaccountable into account (following Mol 2010). We add to studies of care in CSCW by empirically expanding the research sites of care and data work. In this multi-sited ethnographically informed study, we conducted interviews (n = 19) and 160 h of observational studies amongst: 1) Danish Red Cross care workers; 2) Danish Refugee Council legal counsellors; and 3) Danish Immigration Service case officers. We contribute empirically grounded insights into the meanings of care in a datafied asylum context. We find that care is enacted by caseworkers in moments of ambivalence, translation, and attentiveness to “new substantial information” relevant for asylum decision-making. We find that these relational aspects of care in asylum casework impact the production of data about the asylum seeker. We end with a discussion of how a care perspective increases our sensitivity as CSCW researchers towards understanding the conditions for producing quality data and documentation in casework.

In asylum decision-making, legal authorities rely on the criterion "credibility" as a measure for determining whether an individual has a legitimate asylum claim; that is, whether they have a well-founded fear of persecution upon returning to their country of origin. Nation states, international institutions, and NGOs increasingly seek to leverage data-driven technologies to support such decisions, deploying processes of data cleaning, contestation, and interpretation. We qualitatively analyzed 50 asylum cases to understand how the asylum decision-making process in Denmark leverages data to configure individuals as credible (or not). In this context, data can vary from the applicant's testimony to data acquired on the applicant from registers and alphanumerical data. Our findings suggest that legal authorities assess credibility through a largely discretionary practice, establishing certainty by ruling out divergence or contradiction between the different forms of data and documentation involved in an asylum case. As with other reclassification processes [following Bowker and Star 1999], credibility is an ambiguous prototypical concept for decision-makers to attempt certainty, especially important to consider in the design of data-driven technologies where stakeholders have differential power.

Public services increasingly presume a new and more active role for individuals to play in datafied society. While design efforts increasingly attempt to include stakeholders, such attempts are often limited to professional perspectives. Little is known about how individuals who are subject to these solutions experience the increasing use of data about them. One example of public services is job placement. Taking design fiction as our approach, we invited individuals enrolled in job placement (n=20) to reflect on the ‘work of the unemployed’, a fictive scenario where individuals make themselves eligible for support through sharing data. The fiction addresses power dynamics. The study shows how approaches, such as design fiction, are effective at including marginalized communities through changing the conditions for design. Showcasing the fictional outlook, and how the ‘design experience’ can be disempowering if not qualified through a deeper critique, the paper contributes to agendas on design justice.

The research outlined in this paper is focusing on understanding asylum decision-making and outcome variations across the Nordic countries. As a preliminary study, we extract data from an open dataset1 of decision summaries from the Danish Refugee Appeals Board. The approach we propose, offers points for discussion of how prototyping context and participation can help raise questions about such data and engage stakeholders. Combining the application of Machine Learning (ML) and Natural Language Processing (NLP) with participatory methods (e.g. critically designed artefacts) enable us to 1) move beyond “obvious” ML-application areas, 2) through sensemaking of data with stakeholders, and 3) co-develop approaches to data science from a CSCW-perspective.

This paper presents a design fiction: a research prototype of a platform for unemployed individuals trading their personal data. The design fiction questions the ramifications of an understanding of data as individual property by showing a near-future speculative scenario of what government-driven job placement could look like. Which are the kinds of accountability and agency that could be leveraged in the context of job placement if data of unemployed individuals are considered property that can be traded with the public sector in return for support? An algorithm classifies the performance of unemployed individuals based on the data they upload. This way the algorithm becomes the central mechanism for accountability and control instead of the caseworker, who acts as an arbitrator between the individual citizen and algorithm. Our purpose with this paper, and the speculative research prototype, is to create a space for reflection on dilemmas in relation to data property, accountability, and agency in public services.