The world we live in is intrinsically inter-connected and complex. This is especially so for human systems such as politics, cities, and stock markets, where inter-relationships exist in multiple dimensions: political, social, physical, psychological and so on. We have learnt much about these relationships over the years, but more remains unknown to us.
Scientists have endeavoured to simplify and organise these relationships with laws and theories. However, the sheer number of actors and relationships makes it extremely difficult to make sense of human systems for decision-making and predictions. In recent years, complexity science has emerged as a way to understand the equations that underpin the growth of cities and the patterns within.
According to Michael Batty (2013), one of the pioneer academics in the field of complexity science for urban planning, cities are better understood as organisms than machines. Cities are complex systems by nature: individuals, firms, and institutions interact across multiple dimensions and constantly make decisions. No one authority has absolute control over individual actions, yet urban dwellers self-organise into communities and groups to engage in collective activities, and firms self-organise to produce materials and services to meet the needs of the city.
A beehive is an example of a complex system. The rules (DNA programming and stimulus-response) contained in each individual bee direct the hive in working towards a common goal (Image: public domain).
To understand how cities evolve and function, it is necessary to know what makes a system complex and the implications of this. Cities, along with other human systems, share the following key characteristics:
1. They are made up of dynamically interacting heterogeneous agents
Individual components are connected to one another through various processes. As such, a complex system is often represented as a network, where the nodes, often called agents, represent the components, and the edges represent the inter-connections (Figure 1). As components interact with one another in a non-linear manner, changes in specific components may cause disproportionate consequences throughout the system, making it difficult to establish direct causal relationships.
Figure 1: Example network diagram showing nodes and links
2. Emergent behaviour
In a complex system, the whole is greater than the sum of its parts. A city exists because of the interactions among inhabitants and their resulting collective behaviour, which are distinct from the characteristics of individual components, making its behaviour hard to predict. Many of the patterns found in cities, such as where and how people commute, cannot be adequately explained by simply adding up the effects of various factors, like distance from home, travel cost, commuting experience, and so on.
3. They are non-reducible
As a system’s characteristics are distinct from that of individual components, breaking it down to parts risks losing sight of its systemic properties. A deterministic, reductionist approach to analyse a complex system is inadequate as it risks misrepresenting or discounting the diversity and importance of interactions and dynamics that lead to the aggregate observations of the system (Baynes, 2009).
4. Adaptive in nature
Individual components in complex systems change their behaviour according to the situation. They respond to internal and external stimuli, and to their peers connected to them in the network. Over time, the system eventually possesses the quality of learning, with individual components self-organising into a hierarchical structure without an explicit mandate to do so. This is also why cities can exhibit resilience – the ability to persevere in the face of challenges – due to a series of feedback loops that result in self-regulation.
The recognition of complex systems is a breakthrough in our scientific understanding of the world, which has so far been largely influenced by Newtonian views that the world operates in a predictable, rational and consistent way. In fact, Newtonian science can explain both simple and complicated systems. A complicated system, such as a car, camera, or building, can be broken down into individual components and reassembled consistently with predictable results.
A camera is a complicated system – its components interact with one another in predictable and consistent ways. Therefore, the behaviour of a camera is highly predictable (Image: Kelly Hofer, Wikicommons)
However, Newtonian science is insufficient in explaining a large number of events in human systems that are often irreducible and unpredictable. Since the 1940s, scientists have studied and analysed complex systems across a wide range of domains. Physicist Geoffrey West, a leading researcher working on the scientific model of cities and former president of complexity science think tank Santa Fe Institute, called the idea of a “Newton’s laws of complex systems”, which would use principles and equations to make predictions, “almost ridiculous”. “That’s not what we can do for understanding the dynamics and organisations of [complex systems like] cities”, West said.
The modern scientific study of complex systems is a relatively new academic field. To date, there is no complete, unified complexity theory. Rather, complexity science resembles a set of theoretical frameworks for modelling and analysing complex systems within a variety of domains such as biology, ecology, and economics.
The varying approaches also show the potential of complexity science when applied to a diversity of fields. Cities are one of them. Understanding cities as a complex adaptive system has helped urbanists plan cities that better adapt to changes in human needs and desires.
Singapore is one such complex system that is a unique case among cities. In just a few decades, it has transformed from an entrepot port with inadequate urban infrastructure to a global city. Singapore has developed effective mechanisms in urban domains such as transport, housing, utilities, and the environment, and has addressed many urbanisation challenges that still plague many cities worldwide.
However, as it continues to develop, Singapore’s planning and development approaches may no longer be adequate to deal with increasing complexity. The unique nature of its urban transformation also means that it can hardly look to any other city for precedence in planning and development.
Such challenges require a new lens of investigation and analysis, and a new approach to devise solutions for Singapore. In recent years, technology advancements have enabled new tools and research to better manage complexity. Big data analytics, geospatial techniques, urban modelling and simulations, and complexity science are shedding more light on system dynamics in cities.
Policy-making in this environment has to take into consideration the interplay of interconnected and combinatorial factors from across social, environmental, economic and technological domains. These complexities are compounded by challenges that Singapore faces, which include a rapidly changing society, emergence of disruptive technologies, limited resources, climate change and more recently, a global pandemic. Against this backdrop, policymakers have to be aware of the effects of their policies and decisions on other sectors of society.
Complexity science offers a new perspective to the understanding of urban issues and dynamics in the city. While theories and applications of complexity science are still being developed, established research guides us to observe cities at three levels.
At the macro level, a city can be treated as a holistic system. Work by Luis Bettencourt and Geoffrey West, in particular, illustrate emergence in cities from a systems perspective, showing quantitatively that the whole is greater than the sum of its parts. This approach sees the city as an organism that consumes energy, grows in size, and produces waste. In their work, cities manifest macroscopic super-linear (gradient>1) scaling patterns in socioeconomic properties as a result of the social interactions of agents.
Some examples are the super-linear scaling of wages and creative industries as a city grows (Figure 2). On the other hand, space occupied by urban infrastructure, such as roads, scale sub-linearly (gradient<1) as the city grows, demonstrating more efficient and optimal use of resources (Figure 3). These macro observations can be used to gauge cities’ actual performance against expected performance given their size, and highlight inefficiencies and areas for improvement.
Figure 2: Wages and creative industries scale super-linearly as cities grow in size (Bettencourt et al., 2007).
Figure 3: The scaling of a) road infrastructure and b) socioeconomic output with respect to population growth (Bettencourt, 2013). The red lines are the best fit to the data points, the yellow lines describe the theoretical prediction, while the black lines indicate proportional linear scaling (y=x).
At the mezzo, or community, level, a city consists of many individual systems. Michael Batty (2013) highlighted the fundamental basis of complexity science in analysing cities, explaining that they are made up of flows of people, goods and information in various physical, social and digital networks.
Planners model these networks to understand their robustness and resilience. This helps to answer important questions such as whether a disruption to a small section of a network would bring about the failure of the entire transport system; how to facilitate the exchange of ideas to foster an innovative and enterprising environment; and perhaps more pertinently, how to predict the spread and impact of a virus transmission within a community.
Mapping these networks can also reveal patterns that might inform the way we plan and design cities, and allocate resources. There are a wide variety of approaches to describe and analyse complex systems, such as network analytics that focuses on relations between actors, choice modelling that aims to model the decision processes of individuals, and agent-based modelling that simulates the actions and interactions of adaptive agents.
Network science has been widely applied in urban studies to take into consideration the interactions between individuals and other systems. For example, Markus Schlaepfer’s work sought to establish a generalisable model of flows and place-visiting patterns, which can explain the hierarchy of needs and activities (Figure 4).
Figure 4: The number of visitors to a place (corresponding to the brightness of each cell grid) should scale inversely with the square of visiting distance or frequency, where r=distance from the place and f=frequency of visitation (Schlaepfer et al., 2020).
Network analytics has been done to analyse spatial structures in Singapore using public transport smart card data in 2010, 2011 and 2012 (Zhong et al., 2014). The results show an emerging polycentric urban form over the years, with the appearance of new sub-centres and communities mostly consistent with what has been laid out in the Master Plan (Figure 5).
Figure 5: Identified urban centres in Singapore in 2011 based on how each node in the network attracts flows to itself (Zhong et al., 2014). The areas in red represent the detected centres.
Lastly, at the disaggregated level, cities are emergent outcomes of interactions among adaptive agents. Individual human beings and organisations constantly interact, encounter barriers, make decisions and adapt accordingly. Agents respond to the environment and their peers, and are engaged in feedback to planning and policy interventions. The processes through which individual agents perceive and respond to their environment are not entirely clear, but they shape their perception of quality of life. This recognition that agents are dynamic and adaptive demands the consideration for adaptive capacity in exercising policies and people-centric planning.
The availability of data from Internet of Things (IoT) systems in cities, mobile phones and other devices create new opportunities for applying complexity science to understand urban issues and create solutions. For example, transaction data and goods flow data show us how different sectors benefit economically from being spatially clustered, and inform policies to support the growth of economic nodes and sectors. Amenities can also be better planned through the use of data that capture visitorship, user behaviour, and what influences the choice of amenities they visit.
The strength of complexity science, as compared to traditional tools and approaches to urban issues, is not the predictions it can make. Rather, it presents useful tools for positive intervention in urban management (Baynes, 2009) and offers new dimensions for consideration such as learning, adaptation, system dynamics and complex interactions, which have been typically excluded from conventional planning tools.
Complexity science is not a panacea for all urban issues. Meanwhile, it requires a multi-disciplinary effort among urban practitioners and researchers to understand the urban dynamics and sub-systems, and trade-offs in achieving certain objectives. Only then will we be able to gain greater insights into the needs of the city and its people, and better plan and provide the services required.