Urban Science for augmented cities
The term ‘Urban science’ is increasingly used to describe a new generation of approaches to understanding cities and urban areas. There are now journals of Urban Science – and Computational Urban Science – and both undergraduate and postgraduate university degrees in Urban Science linked to Planning, Computer Science and Urban Analytics. But what, exactly, is urban science? How does urban science differ from other kinds of science, or other kinds of urban knowledge? And how could urban science help support understanding of urban fields including the built environment?
Figure 1: Machine learning? A concept associated with ‘urban science’.
Defining urban science
Science can be considered a system for understanding the world, based on evidence gained through methods such as observation, experiment and hypothesis-testing. Hence, urban science in its most general sense could be any science (or use of scientific method) for supporting the understanding of cities/urban areas (and potentially how they have developed and how they might be designed or planned). In this general sense, urban science could include anything from studying street trees to provide shade and air filtering, or the effect of permeable surfaces on run-off, to understanding materials for construction, or energy in buildings.
Seen this way, urban science in a general sense is seamless with any other science; it is just an urban-oriented version of science – or application of sciences to the urban domain – and not a ‘special’, anthropocentric kind of science (in the way that social science sometimes distinguishes itself from other kinds of science).
But nowadays, a new wave of ‘urban science’ or ‘city science’ has come to the fore as an academic discipline, or at least, a domain in which inter-disciplinary scholarship is pursued. This has a more specific focus, and set of connotations, than science in general.
According to Lobo et al (2020)’s report to the Santa Fe Institute, “Urban science seeks to understand the fundamental processes that drive, shape and sustain cities and urbanization. It is a multi/transdisciplinary approach involving concepts, methods and research from the social, natural, engineering and computational sciences, along with the humanities.”
More specifically, Michael Batty (2019) and Luis Bettencourt (2021) have emphasized the idea of generalizable laws that seem to govern cities, as complex systems; and in Batty’s view, this narrower sense of urban science can exclude things like the physics of buildings or technology of construction, but can be considered more to do with the science of flows, of people, goods and information.
Urban science has been popularly defined as “an interdisciplinary field that studies diverse urban issues and problems… Urban science uses a computational understanding of city systems to evaluate how they work and how they… grow and change.” The term ‘computational’ gives a hint of the kind of contemporary slant on the topic that we sense is intended by the term.
MIT’s Urban Science web entry states “We integrate data analysis, visualization, sensors, and artificial intelligence into a planning, design, and policy-making context” – in effect, the science is used to support the urban understanding and intervention. A longer list of topic keywords at MIT’s City Science lab includes artificial intelligence, augmented reality, machine learning, robotics and the internet of things.
That said, urban science is not just about computational technology. After all, Leonhard Euler did not need a computer to model the Königsberg bridge problem that became a foundation of graph theory. The conception that travel patterns could be predicted – if not explained – in a way analogous to Isaac Newton’s Inverse-Square Law of Gravitation did not need to wait for the advent of digital technology.
Figure 2: Euler’s Königsberg bridge problem (left), and its representation as a graph (right). Urban science is often based on abstract models but these do not necessarily require modern computation nor big data.
Attributes of urban science
Rather, urban science can be seen as a way of thinking about and understanding the urban, that could be regarded as differing from traditional ways of urban knowledge. In effect, we could recognize a suite of related attributes of urban science:
· Computational (hardware, software) – including cutting-edge computation, artificial intelligence and machine learning – this is reasonably considered technological;
· Big data – where we learn from statistical regularities in large samples or data sets, rather than extrapolating or generalizing from local individual observations on the ground. In a way this is an epistemological issue rather than a technological one, as we can imagine a manually compiled equivalent – a sort of steampunk or rather paper-and-ink-punk version of big data laboriously compiled from hand-written scripts;
· Non-human sensing of the city – beyond human capability or reach – this could be considered experiential or phenomenological. Again this is not just about technology per se – such as using cameras or air quality sensors instead of our own eyes and noses. Rather, it is about how we get at what’s happening via different media – for example, using the geospatial coordinates and linguistics of social media posts to detect the movements and mood of crowds on the day of an event, to take just one example;
· Abstractions and models – conceptualizing cities as complex systems, or networks and flows… to some extent treating the human realm, almost as if were a problem of physics, forces and particles. In a sense the city becomes analogous to a huge human machine, its working parts propelled by human motivations and actions. This is conceptual rather than technological;
· Finally, pervading all we have the scientific mindset itself, which can help to propel a progressive accumulation of knowledge, featuring testing and validation, and open to challenge and correction. Here, the quantification of problems and establishing matters of statistical significance and sensitivities are pursued in a way that could be considered in contrast to more traditional ways of urban scholarship.
Applications of urban science
When asking ourselves how urban science can support understanding urbanism and the built environment, we can answer in the following way. First, in common with any science, urban science can help support understanding of individual urban phenomena, and can also be used to build more general, integrated theories of urbanism.
Secondly, we can see particular areas where the specific attributes of ‘urban science’ in the contemporary sense can help. These are not merely technological, but get to the heart of understanding cities for people, for example:
· Use of pervasive sensing coupled with big data could get more diverse and objective data on the city than extrapolating from the idiosyncratic, subjective impressions of individual ‘ambulant academics’;
· Exploitation of advanced mapping, data mining and machine learning to detect previously unseen patterns of urban phenomena, including human behaviour;
· Simulation and visualization of proposed futures – including via augmented reality – could help make city and local plans more accessible to lay people who may not be used to interpreting conventional plans and cross-sections; the ability to ‘fly’ virtually though proposed developments; and the possibility for potential users to generate their own ideas and plans, and share them publicly or among their peers, could help ‘democratise’ proposal-making and shaping of urban places (Marshall et al, 2019; Karadimitriou et al, 2022).
Figure 3: Screenshot of 3D model of a residential courtyard featuring in an online platform allowing residents share, adapt and comment on user-generated designs.
Built Environment journal is interested in cross-cutting research agendas arising from urban science, such as new technology (hardware and software and social media); the interface between data science and issues of equity and social justice; synergies arising from linking pervasive artificial sensing and the human experience – or machine learning and human learning, for that matter; and the development of new urban concepts and urban models that are ‘native’ to the digital age, going beyond those developed from the topology of Euler or physics of Newton. In these days when lab-grown brain cells play video games, the time seems ripe for a new wave of urban science for augmented twenty-first century cities, that combines the best of human and artificial intelligence.
This article is based in part on an online lecture delivered to MIT’s Department of Urban Studies and Planning, Norman B. Leventhal Center for Advanced Urbanism (LCAU), 12th October 2022.
Batty, M. (2019) On the confusion of terminologies. Environment and Planning B: Urban Analytics and City Science, 46: 997-998.
Bettencourt, L. (2021) Introduction to urban science. The MIT Press, Cambridge, Mass.
Karadimitriou, N., Magnani, G., Timmerman, R., Marshall, S., & Hudson-Smith, A. (2022). Designing an incubator of public spaces platform: Applying cybernetic principles to the co-creation of spaces. Land Use Policy, 119, 106187. doi:10.1016/j.landusepol.2022.106187
Lobo, J., Alberti, M., Allen-Dumas, M., Arcaute, E., Barthelemy, M., Bojorquez Tapia, L.A., Brail, S., Bettencourt, L., Beukes, A., Chen, W.Q., Florida, R. (2020) Urban science: integrated theory from the first cities to sustainable metropolises. Report to the National Science Foundation. Santa Fe Institute, Santa Fe.
Marshall, S., Hudson-Smith, A., & Farndon, D. (2019). Digital participation – taking 'planning' into the third dimension. Town and Country Planning, 88 (1), 11-14.
Image 2 (right): Representation of the Seven Bridges (Source: author, all rights reserved)
Image 3: 3D model of a residential courtyard (Source: Incubators of Public Spaces project consortium, all rights reserved)