Shifting Concepts of Value: Designing Algorithmic Decision-Support Systems for Public Services
Open accessAbstract
Calls for responsible design in algorithmic decision-support systems, especially those used in public services, are increasingly common. While an algorithmic system might promise greater precision and efficiency in domains such as critical care, the same efficiency is difficult to replicate in the public service domain, where caseworkers must exercise discretion in applying complex legal frameworks that directly affect individual lives. In this paper we examine the challenges in responsibly designing such an algorithmic decision-support system. We report findings from a large cross-disciplinary research project, aiming to develop an algorithmic component for municipal decision-support systems for job placement in Denmark. Our data showcases insights from how a team of data scientists, caseworkers, and system developers negotiated notions of value metrics and usefulness in a participatory design set-up. Whereas data scientists expected to focus on profiling individuals, the caseworkers instead pushed for systems that could help mitigate organizational contradictions and obscured processes in casework. We close with a discussion of challenges and future directions for participatory algorithmic systems design in municipal contexts.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Practitioners must consider and balance human and democratic values when assessing cases about citizens.
“Practitioners must consider and balance human and democratic values when assessing cases about citizens [47,62].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
Section 6.3 says job-placement caseworkers balance legal demands with a human or caring approach when working with individuals. But it treats democratic values as a concern of the design process, not as values practitioners explicitly balance when assessing each citizen’s case; it says caseworkers do not explicitly mention human values. (LLM, conf=0.86)
Experimental — accuracy not yet measuredIn several public-sector domains, there is significant debate about which predictors relate to which outcomes.
“In several domains within the public sector, there is significant debate about which predictors are associated with which outcomes [39,62,106].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
The Introduction and §5.1 (page not available in our copy) describe debate in a municipal job-placement project over risk prediction and which metrics would be useful to caseworkers. They do not establish debate across several public-sector domains about which predictors are associated with which outcomes. (LLM, conf=0.82)
Experimental — accuracy not yet measuredPrior work argues that public-sector algorithms are a distinct domain and must be characterized differently from private-sector algorithms where outcomes are well-defined.
“Prior work has argued that algorithms in the public sector is a domain in its own right [34,62,107,109,121] and must be characterized differently as compared to algorithms in the private sector where the decision outcomes are well-defined.”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
The Abstract and §2.1 describe public-service algorithmic decision support as requiring attention to caseworker discretion and value metrics that cannot be reduced to performance measures. That supports characterizing public-service algorithms differently, but our copy does not make the claimed comparison with private-sector algorithms or say that private-sector outcomes are well-defined. (LLM, conf=0.84)
Experimental — accuracy not yet measuredWhatever the outcome type, public-sector street-level bureaucrats must still use discretion, contextualize algorithmic results for each case, and act within bureaucratic constraints as ADMAPS describes.
“Moreover, regardless of the type of outcome, street-level bureaucrats in the public sector must still exercise discretion and contextualize the algorithmic results for each case and act within the constraints posed by bureaucratic processes as explained by the ADMAPS framework [2,62,91].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
The Introduction and §2.1 (pages not provided) support the need for caseworker discretion when applying bureaucratic rules to individual situations, and §5.1 describes a caseworker making her own judgment about an algorithmic red flag. The paper does not establish that this applies regardless of outcome type or explain the ADMAPS framework. (LLM, conf=0.83)
Experimental — accuracy not yet measuredADMAPS can help design algorithms for job placement centers where caseworkers must exercise discretion in applying legal frameworks, assessing resource constraints, and resolving organizational contradictions to extend unemployment benefits.
“Similarly, ADMAPS can help design algorithms for job placement centers where the caseworkers must exercise discretion in applying complex legal frameworks, assessing resource constraints, as well as resolving organizational contradictions (i.e., bureaucratic processes) to extend unemployment benefits to citizens [62].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.
Experimental — accuracy not yet measured
A person's situation can destabilize repeatedly, making successful outcomes or interventions hard to assess.
“Moreover, an individual's personal situation can (de)stabilize several times making it hard to assess what constitutes a successful outcome or intervention [62].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
These new skills are not accounted for in what constitutes professional expertise.
“Practitioners in the public sector are continually acquiring new skills as they learn to make decisions through data systems and interpret algorithmic outputs [17,21], however, these new skills are not being accounted for with respect to what constitutes professional expertise [62].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
Public-sector decision outcomes are not well-defined, and a person's life can stabilize or destabilize repeatedly, making successful outcomes hard to predict.
“Decision outcomes in the public sector are not well-defined and as previously noted, a person's life can stabilize or de-stabilize several times making it hard to predict what constitutes a successful outcome or intervention [62].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
Prior public-service work found caseworkers sought algorithmic systems to mitigate organizational contradictions and clarify processes.
“The central role of bureaucratic processes is evident from prior work conducted in public services where caseworkers pushed for algorithmic systems that could help mitigate organizational contradictions and clarify organizational processes [62].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
Differing socio-technical imaginations of public-sector algorithms are consistent with existing literature.
“Differing socio-technical imaginations of algorithms in the public sector are aligned well with existing literature [20,62,103,107,112].”
From A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare (2021) · cites this paper as [62]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 44 · Cited by 97
Research works printed but not linked: 11
- [2]2. Karla Bardillo-Urquiola, Xinru Page and Pamela J. Wisniewski. 2019. Risk vs. restriction: The tension between providing a sense of normalcy and keeping foster teens safe online. In Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI), Article 267.
- [4]4. Pernille Bjørn & Ellen Balka. 2007. Healthcare Categories Have Politics Too: Un- packing the Managerial Agendas of Electronic Triage Systems. In Proceedings of The European Conference on Computer-Supported Cooperative Work (ECSCW), 371-390.
- [10]10. Anna Brown, Alexandra Chouldechova, Emily Putnam-Hornstein, Andrew Tobin, and Rhema Vaithianathan. 2019. In Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI), Article 41.
- [14]14. Susanne Bødker, Pelle Ehn, Dan Sjögren & Yngve Sundblad. 2000. Cooperative Design -Perspectives on 20 Years With 'the Scandinavian IT Design Model'. Invited paper in Proceedings of the 11 th Nordic Conference on Human-Computer Interaction (NordiCHI), 1-11.
- [24]24. Elizabeth Kaziunas, Mark S. Ackerman, Silvia Lindtner & Joyce M. Lee. 2017. Car- ing through Data: Attending to the Social and Emotional Experiences of Health Datafication. In Proceedings of the ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW), 2260-2272.
- [33]33. Hanne Marie Motzfeldt. 2019. Socialrådgivere og Fagforeninger: Grib dog red- ningskransen! Uden for Nummer, 39(19), 32-41.
- [40]40. Samir Passi & Steven J. Jackson. 2017. Data vision. Learning to see through algorithmic abstraction. In Proceedings of the ACM Conference on Computer- Supported Cooperative Work and Social Computing (CSCW), 2436-2447.
- [41]41. Anette C. M. Petersen, Lars Rune Christensen & Thomas T. Hildebrandt. 2020. The role of discretion in the age of automation. Computer Supported Cooperative Work: An International Journal (JCSCW).
- [44]44. Anthony T. Pinter, Pamela Wisnewski, Heng Xu, Mary Beth Rosson, and John M. Carroll. 2017. Adolescent online safety: Moving beyond formative evaluations to designing solutions to the future. Note in Proceedings of the ACM Interaction Design and Children Conference (IDC), 352-357.
- [46]46. Juho Paä kkonen, Matti Nelimarkka, Jesse Haapoja & Airi Lampinen. 2020. Bu- reacracy as a lens for analyzing and designing algorithmic systems. In Proceedings of the ACM Conference on Human Factors in Computing (CHI), Article 651.
- [48]48. Toni Robertson & Ina Wagner. 2013. Simonsen, J. & Robertson (Eds), T. Routledge International Handbook of Participatory Design, Routledge, 64 85.
- Authors
- Holten Møller, N., Shklovski, I., & Hildebrandt, T. T.
- Affiliations
- Department of Computer Science University of Copenhagen, Denmark
- University of Copenhagen (DK)
- Venue
- Proceedings of the 11th Nordic Conference on Human-Computer Interaction: Shaping Experiences, Shaping Society
- Publisher
- ACM
- Published
- 2020-10-25
- Pages
- 1-12
- OpenAlex
- W3095415574
- Cited by
- 97
- References
- 44
- Citations by year
- 2026: 16 · 2025: 14 · 2024: 21 · 2023: 18 · 2022: 21 · 2021: 7
- FWCI
- 11.93
- Type
- conference paper
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Nordic Conference on Human-Computer Interaction (publishedVersion)
- Research at the University of Copenhagen (University of Copenhagen) (publishedVersion, cc-by)
- Funders
- Innovationsfonden (7050- 00034A)
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Social Sciences › Sociology and Political Science › Information Systems Theories and Implementation
- Social Sciences › Political Science and International Relations › E-Government and Public Services
- Keywords
- algorithmic decision support, responsible design, public services, job placement, value metrics, participatory design, algorithmic transparency, data scientists, organizational contradictions
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
