Reviewed by
Andreas Koch
University of Salzburg
The book Understanding Complex Urban Systems edited by Walloth et al. has been published with the aim to discuss approaches that help understanding the complexities of urban systems “by overcoming data-related challenges” (p. v). It is precisely devoted to problems of data availability, quality, accuracy, and resolution. As indicated in the book’s subtitle, a promising methodological approach is supposedly given with an integrative and multidisciplinary implementation of data into adequately developed models.According to this basic theme, the seven chapters of the book firstly are dealing with methodological issues such as “big data”, “participatory data collection”, and “techniques of handling data unavailability”, followed by a dedicated focus on theoretical ideas like “nested system models”, “urban space as a system of flows”, “capability approach in data collection”, and “disruptive innovations”. Almost all the chapters are, explicitly or at least implicitly, referring to agent-based modelling or system dynamic approaches in order to achieve their scientific explorations.
The quality fluctuates significantly through the chapters, in a majority of it being problematic. One common difficulty that frames more or less all the chapters is that of a poor theoretical grounding of the themes presented. Indeed, the authors (and editors) do not put much effort in defining the core notions of the book – be it “complexity”, “urban”, “systems”, or “multidisciplinarity”. The first Chapter (Introduction) begins with an inappropriate “definition” of an urban system: “We understand that an urban system is a space in which human actions take place. Hillier and Vaughan (2007) state that an urban system is composed of two things: a large collection of buildings linked by space, i.e., “the physical city”, and a complex system of human activity linked by interaction, i.e., “the social city” (p. 2). Besides the problem of tautology – urban system is a space and is composed of space that is already space – the definition can be applied to any settlement. The particularity of an “urban” settlement is not mentioned at all (which is, among others, given with a huge social, ethnic, life-style diversity; conflicts between competing milieus about how to use public spaces; a huge diversity of economic activities, along with highly differentiated labour markets). This problem gets even worse, because of the editors’ claim to “first understanding the way urban systems actually evolve, before making models and choosing data” (p. 5). If research is (correctly) organised this way, then theories/definitions and models derived from these theories/definitions are necessary.
Convincing chapters of this book are those which have some theoretical and critical debates as a (temporary) starting point in order to further investigate methodological challenges of data restrictions – the common problems of calibrating, verifying, and validating the model(s) in use – within a specific model environment. The Chapters of Scheutz and Mayer on “Combining Agent-Based Modeling with Big Data Methods to Support Architectural and Urban Design” as well as of Yang and Day on “Operationalizing the Capabilities Approach for Modeling Household Welfare Shifts in Urban Systems” can be mentioned here as positive instances.
Other chapters are less convincing, partly due to a positivistic, uncritical epistemology, and partly due to a missing link to the book’s title and thus intention. For example, Gebetsroither-Geringer and Loibl present a self-proclaimed “new” approach of using participatory data as an empirically based ingredient for decision-support systems in urban planning. They contrast their own approach with a traditional approach of creating “attractivity maps” of neighbourhoods that derived from official population data and statistics. Though a critical discussion of such an approach is quite fruitful, it is scientifically unfair to criticise another approach while favouring one’s own. In addition, the two approaches cannot be compared directly, because the authors’ data gathering approach is based on hypothetical attitudes of the respondents (in their questionnaire citizens are asked about which urban areas they “like most”, “could imagine to move to”, “would not want to live in at all”). Hence, while the “traditional” approach represents objective facts irrespective of individual aspirations, the “new” approach exclusively considers subjective opinions. Moreover, the authors’ approach does not come with any information about what “like most” or “could imagine” actually means – does it refer to the social or physical neighbourhood, the transportation or social infrastructure, the spatial nearness to workplaces, or to all of the above (and if so, weighted or not)? Gebetsroither-Geringer and Loibl, however, conclude that the attractivity maps based on their approach represent the “preferences of citizens at the point of time when the poll was carried out” (p. 41). This seems to be a highly problematic conclusion, since it is unproven. The authors do not have any information about the factual behaviour nor the factual knowledge of the respondents – asking people “would” or “could” questions is nothing more than hypothetical knowledge of an allegedly stereotype nature.
Another chapter with some problematic implications is given with Walloth’s topic of “Big Data or No Data: Supporting Urban Decision-Making with a Nested System Model”. The problem is not so much the conceptualisation of a “Nested System Model” (which is not new) as is the line of argumentation – to praise the own approach and deny the benefits of the counter approach. Walloth begins with the statement that big data “must seek outliers, i.e. irregularities or deviations from the existing pattern” (p. 67). Apart from the fact that this can be stated to any data analysis (and thus is not an argument against big data), it is the way how Walloth further specifies “irregularities or deviations”. He suggests and describes without any further explanation three “types of data”: (i) those that represent “the average”; (ii) those that represent the “unlikely”; and (iii) those that represent the “novel” or the “unpredictable”. What is missing, among others, is a definition of “average”, “deviation”, and “unpredictability” as well as an evaluation of how these allegedly distinct “data sets” are intertwined, which they actually are from both a statistical and a content point of view. Nonetheless it is important to raise a critical awareness to the problems of big data, but it does not fully refute its usefulness. After the presentation of this critical evaluation of big data Walloth proceeds with an introduction of his nested system approach, which is mainly about a differentiation of system behaviour in terms of temporality and/or process speed (he distinguishes slow ecological systems from fast technological systems, to mention just the two opposing poles of systems and their process times). This topic, however, has no connection to the big data evaluation topic discussed before. The description of nested systems is then followed by an attempt to influence these systems – at a qualitative level. Based on two case studies (Airbnb in Berlin and Uber in Brussels) Walloth concludes with different political “answers” to the problems that have been raised by Airbnb and Uber (which reveals a strict linear thinking, but this is another problem of the chapter). Decision-making in and with nested systems “requires researchers who carry out case studies to observe, evaluate, and understand which activities belong to which system, and/or to apply methods of (Big Data) analysis to analyse the periodicity of events in time series, prior to influencing the complex system” (p. 77). This might be a valid procedure, but it remains an open question how users might adequately deal with the data; the title “Big Data or No Data” is anything but helpful in this respect.
To conclude: it is quite difficult to detect the ultimate intention of the book. The complexity of urban systems is well stated but not so well described and explained; the examples given in the chapters do not represent urban complexity in a comprehensive manner. The claim of “integrating multidisciplinary data” is only partly implemented; for example, a discussion about sensor data would be very important – what does sensor data represent? What might be derived from sensor data? Is it possible to link them to other data types such as social media data? All this would be beneficial in order to (better) understand complex urban systems.
Return to Contents
of this issue