Speaker
Description
Keywords: computational planning; computational legibility; operationalisation; algorithmic governance
Introduction and Research Problem
Digital planning tools are now used to analyse urban conditions, while scenario approaches allow planners to compare alternative development scenarios (Abou Jaoude et al., 2022; Son et al., 2023). Algorithmic methods add a further layer by enabling the computational evaluation and optimisation of planning alternatives (Churiakova et al., 2024; Son et al., 2023).
Once digital tools move from analysis to recommendation, the normative dimension of planning becomes harder to ignore. Planning choices concern not only existing conditions but also which futures, priorities, and competing interests are treated as desirable (S. D. Campbell, 2013). A computational model that proposes future development therefore requires some formal account of what should count as a desirable urban outcome (H. Campbell, 2006; Shu, 2026).
For normative categories such as health, security, well-being, social equality, or cultural self-realisation to enter such a model, they must be expressed through entities, relationships, indicators, or other characteristics that computation can process (Churiakova et al., 2024; Mityagin et al., 2021; Shu, 2026). Operationalisation is the process through which this translation occurs: complex normative concepts are selectively formalised and made quantitatively representable (Espeland & Stevens, 1998; Merry, 2011).
This study examines what happens to normative ideas in this process of translation. It focuses on which dimensions become computationally legible, which are represented only partially, and which remain computationally invisible.
The empirical case is ITMO University’s public IDU platform and related publications (Institute of Design and Urban Studies, ITMO University [IDU], n.d.-a). Within the value-based management approach developed in the same research environment, quality of urban life is linked to the needs and values of different socio-demographic groups, while urban services are treated as means of satisfying residents’ needs (Mityagin et al., 2021). In the Digital Master Plan, five development scenarios—Healthy City, Comfortable City/Yard, Sustainable City, Social Equality, and Business City—are translated into measurable indicators and incorporated into an optimisation procedure (Churiakova et al., 2024).
The main research question is: How are normative ideas about desirable urban development transformed when they are operationalised as variables in a computational planning model? Three sub-questions follow: Which aspects become computationally legible through operationalisation? Which receive only limited representation or remain computationally invisible? How do decisions about what is operationalised, and how it is formalised, influence model-generated recommendations?
Conceptual or Theoretical Framework
The conceptual framework of this study begins from the normative character of planning. Planning involves judgements about desirable futures, priorities, and competing interests rather than the technical application of objective knowledge alone (H. Campbell, 2006). Computational tools do not remove this normative dimension. When such judgements enter a computational model, they must be translated into forms that can be processed through calculation and optimisation (Churiakova et al., 2024; Mityagin et al., 2021).
Espeland and Stevens’ (1998) concept of commensuration helps explain this translation. It describes how qualitatively different phenomena are transformed into a common form that enables comparison. In computational planning, broad concepts such as health, comfort, sustainability, or social equality therefore enter calculation through categories, indicators, and other measurable parameters (Halla & Merino-Saum, 2022; Shu, 2026). This process makes comparison possible, but it also requires decisions about which dimensions can be represented within a shared system of measurement.
Merry (2011) similarly shows that indicators simplify and standardise complex social knowledge rather than reproduce it in full. Formalisation may therefore involve value narrowing, whereby a broad normative value is transformed into a narrower, technically implementable and measurable form (Moats et al., 2026). As a result, some dimensions of a value become available to computation while others receive only partial representation or remain outside the model.
Drawing on Scott’s (1998) concept of legibility and Shu’s (2026) concept of the model-ready city, this study develops the concept of computational legibility. It refers to the capacity of an urban phenomenon to acquire formal representation through categories, data, relationships, or indicators with which a computational model can operate. The study also introduces computational invisibility as its analytical counterpart: dimensions that do not acquire such representation and therefore cannot directly enter computation. In this sense, computational models produce a selectively legible representation of the city.
Methods
This study is a qualitative case study of ITMO University’s public IDU platform, drawing on peer-reviewed publications, public interfaces, and the author’s practitioner-researcher perspective.
Published work provides the conceptual and methodological basis. “Value-Based Management in a Smart City” describes the formalisation of values and needs of different socio-demographic groups and their relationship to urban services (Mityagin et al., 2021). “Digital Master Plan as a Tool for Generating Territory Development Requirements” describes a tool in which development scenarios are represented through measurable indicators and incorporated into an optimisation procedure (Churiakova et al., 2024).
Public platform interfaces are treated as primary digital material. The catalogue of applied services provides access to the tools available on the platform (IDU, n.d.-a), while the Population Value Model links social groups and value categories to life situations, urban services, and parameters of pedestrian or transport accessibility (IDU, n.d.-b).
An operationalisation episode is used as the unit of analysis. Each episode traces a transition from a normative concept or life situation to formal categories and measurable characteristics. The analysis follows what is retained in this transition, what becomes measurable, and what is only partially represented or left without formal expression.
The author’s participation in the Digital Master Plan project as an urban planner and architect within an interdisciplinary team provides an additional reflexive perspective on how professional planning knowledge was translated into formal and computational representations.
Results or Expected Results
Preliminary analysis reveals two interconnected levels of operationalisation. In the Population Value Model, normative categories are translated into life situations, and then linked to specific urban services and measurable accessibility parameters (IDU, n.d.-b). Pregnancy illustrates this process clearly: it is represented through access to a pharmacy within a 500 m walk, an antenatal clinic and polyclinic within a 1,000 m walk, and maternity and baby-goods shops within 30 minutes by transport. A complex life situation thus becomes computationally legible through a defined set of services and accessibility thresholds.
A similar mechanism appears in becoming a victim of crime, formalised through access to the police within 2,000 m walking distance and to a prosecutor’s office within 30 minutes by transport. Despite involving very different needs and vulnerabilities, both situations enter the model through the same formal structure: selected services plus measurable spatial or temporal accessibility. Other dimensions receive only partial formal expression or remain computationally invisible.
A second level appears in the Digital Master Plan, where broader normative scenarios are translated into variables of the optimisation function. Looking across the five published scenarios, the same types of variables recur: service provision, accessibility, diversity, land-use characteristics, and access to public transport (Churiakova et al., 2024).
The Sustainable City scenario shows this particularly clearly. Its conceptual framing includes economic, social, environmental, and institutional dimensions, whereas its computational representation relies on a more limited set: green-space provision and area, walking access to public transport, service diversity in residential areas, and social-infrastructure provision (Churiakova et al., 2024). Sustainability therefore becomes computable through selected spatial and service characteristics, while other dimensions of the broader concept remain outside direct computational representation.
These examples show that operationalisation does more than measure pre-defined normative concepts: it helps constitute their computational content. Pregnancy becomes computable as services and accessibility thresholds, and sustainability through green spaces, public transport access, service diversity, and social infrastructure provision.
The comparison of the two levels reveals a common pattern, termed here convergence through operationalisation. Normatively different categories and scenarios—from pregnancy and security to social equality and sustainability—are translated into recurring spatial-service variables. This shared formal language makes heterogeneous normative concepts comparable and amenable to computation.
A practical boundary follows from this translation. Dimensions with formal representation can enter calculation; dimensions without one remain only partly represented or computationally invisible. Normative content is therefore not removed from the model, but carried into it in a particular operational form.
Across both levels, computational legibility depends on what has been formalised: entities, services, indicators, and measurable spatial relationships. Model-generated recommendations inherit the same boundary because optimisation can respond directly only to characteristics already defined through variables and objective functions. Operationalisation therefore affects not only how a desirable city is represented, but also which forms of urban improvement the model can recognise and recommend.
Discussion and Conclusion
The case shifts attention from how well a model performs optimisation to what has first been made available for optimisation. Operationalisation is therefore part of planning itself, not merely a technical preparation stage. By the time a normative idea enters the model, some of its dimensions have already been retained, narrowed, or excluded.
Computational invisibility is not always a simple consequence of missing data or technical limits. It may also arise earlier, when decisions are made about which categories, indicators, and relationships belong in the formal representation of the city. These choices matter politically because they shape what the model is able to treat as a condition, a problem, or a possible improvement. What remains outside the computational representation cannot directly enter optimisation and therefore has a weaker capacity to shape model-generated recommendations.
Assessment therefore cannot stop at accuracy or optimisation efficiency. It also needs to consider the formal representation on which computation rests: what categories capture, what they exclude, and whether those choices can be questioned or revised.
This matters especially in questions of equality and accessibility, because similar spatial conditions may correspond to very different lived possibilities. The same proximity to a service can have different implications depending on gender, physical mobility, care responsibilities, or other dimensions of everyday urban experience (Kern, 2020). When such differences remain computationally invisible, they cannot directly influence optimisation and may consequently be reproduced or reinforced in model-generated recommendations.
Governance issues arise before the final planning decision. Once computational models become part of planning practice, influence also lies in decisions about which categories structure the model, who may revise them, and whose objections can alter them. The next stage of this research will examine how this authority is distributed among developers, planning professionals, municipalities, and local communities.
This distribution of authority may be especially consequential in more centralised governance systems. Municipalities may be responsible for implementing planning decisions without having an equivalent capacity to alter the computational categories on which those decisions depend. Further research will examine whether institutional centralisation creates such a constraint and, if so, how it affects municipal capacity to modify, challenge, or supplement categories embedded in computational planning systems.
The implication is not to reject computational planning, but to treat formalisation as an explicit, reviewable, and contestable part of the planning process. Doing so requires the normative assumptions, exclusions, and representational limits of computational models to remain visible and open to professional and public scrutiny.
References
Abou Jaoude, G., Mumm, O., & Carlow, V. M. (2022). An overview of scenario approaches: A guide for urban design and planning. Journal of Planning Literature, 37(3), 467–487. https://doi.org/10.1177/08854122221083546
Campbell, H. (2006). Just planning: The art of situated ethical judgment. Journal of Planning Education and Research, 26(1), 92–106. https://doi.org/10.1177/0739456X06288090
Campbell, S. D. (2013). Sustainable development and social justice: Conflicting urgencies and the search for common ground in urban and regional planning. Michigan Journal of Sustainability, 1, 75–91. https://doi.org/10.3998/mjs.12333712.0001.007
Churiakova, T., Starikov, V., Sudakova, V., Morozov, A., & Mityagin, S. (2024). Digital master plan as a tool for generating territory development requirements. In T. Guarda, F. Portela, & J. M. Diaz-Nafria (Eds.), Advanced research in technologies, information, innovation and sustainability: Third international conference, ARTIIS 2023, Madrid, Spain, October 18–20, 2023, proceedings, Part II (pp. 45–57). Springer. https://doi.org/10.1007/978-3-031-48855-9_4
Espeland, W. N., & Stevens, M. L. (1998). Commensuration as a social process. Annual Review of Sociology, 24, 313–343. https://doi.org/10.1146/annurev.soc.24.1.313
Halla, P., & Merino-Saum, A. (2022). Conceptual frameworks in indicator-based assessments of urban sustainability—An analysis based on 67 initiatives. Sustainable Development, 30(5), 1056–1071. https://doi.org/10.1002/sd.2300
Institute of Design and Urban Studies, ITMO University. (n.d.-a). Система комплексной оценки [Integrated assessment system] [Web page]. Retrieved September 11, 2026, from https://dc.idu.actcognitive.org/applied-services
Institute of Design and Urban Studies, ITMO University. (n.d.-b). Ценностная модель населения [Population value model] [Web application]. Retrieved September 11, 2026, from https://dc.idu.actcognitive.org/models-population/human-society/demonstration
Kern, L. (2020). Feminist city: Claiming space in a man-made world. Verso.
Merry, S. E. (2011). Measuring the world: Indicators, human rights, and global governance. Current Anthropology, 52(S3), S83–S95. https://doi.org/10.1086/657241
Mityagin, S. A., Gornova, G. V., Drozhzhin, A. I., & Sokol, A. A. (2021). Value-based management in a smart city. International Journal of Open Information Technologies, 9(12), 104–110. https://doi.org/10.25559/INJOIT.2307-8162.09.202112.104-110
Moats, D., Ruckenstein, M., Brogård Kristensen, D., Eidenskog, M., Elhadj, E., Lehtiniemi, T., Møhl, P., Pretnar Žagar, A., Tanninen, M., & Velkova, J. (2026). The dark side of public values in algorithmic systems. Dialogues on Digital Society. Advance online publication. https://doi.org/10.1177/29768640261457507
Scott, J. C. (1998). Seeing like a state: How certain schemes to improve the human condition have failed. Yale University Press.
Shu, T. (2026). Upstream of value disconnects: The model-ready city and the material conditions of urban AI. Dialogues on Digital Society. Advance online publication. https://doi.org/10.1177/29768640261479430
Son, T. H., Weedon, Z., Yigitcanlar, T., Sanchez, T., Corchado, J. M., & Mehmood, R. (2023). Algorithmic urban planning for smart and sustainable development: Systematic review of the literature. Sustainable Cities and Society, 94, 104562. https://doi.org/10.1016/j.scs.2023.104562