Abstract
In this Editorial, we draw on our collective experience working across a variety of disciplines and application domains at the Alan Turing Institute to place Digital Twins on metaphorical trial, asking whether they have the potential to offer solutions to big, complex, wicked 1 social and urban challenges, or whether their application should remain limited to improving jet engines (Trimble, 2022) or, perhaps, urban transportation systems (Feng et al., 2023). This is an important and timely question to ask, as the popularity of Digital Twins is increasing precisely at a time when analytical and data-driven solutions are clearly needed to address questions of critical societal importance. The ‘wicked’ questions that we might want to ask include: why, in the 21st century, in a wealthy nation like the UK, are 4.7 million people living in food poverty (Francis-Devine et al., 2003)? Or, how does it happen that half of all children born in 2009–2010 in Liverpool, a major UK city, were referred to Children’s Services before the age of five (Bennett et al., 2024)? In this piece, we discuss some of the barriers that are preventing Digital Twin technology from helping with these questions and highlight current efforts, including those as part of the Turing’s Urban Analytics programme, that may help to overcome these obstacles.
The questions we pose above may appear overly narrow and not the appropriate domain for Digital Twins (DTs). Consider, however, that these tangible and measurable outcomes result from complex interactions between, for example, food systems, financial systems, care networks, transport and infrastructure networks, and many other quantifiable components of urban systems. Consider also that DTs are frequently touted as a technocratic solution to urban wellbeing challenges. If DTs cannot address the immediate wellbeing challenges of urban inhabitants – for example, food, shelter, health, and fuel – what good are they? And, more importantly, what is needed to bridge the gap between DT potential and what they currently achieve?
For those who may not have noticed, DTs are having a moment (Batty, 2018; Tomko and Winter, 2019). Adopted by quantitative geographers, urban analysts, planners and civil engineers alike, the term ‘Digital Twin’ has become almost ubiquitous in certain circles. They have certainly generated substantial interest in the pages of this journal – the forthcoming ‘Digital Twins for Cities’ special issue, for example – as well as in other journals such as Nature Computational Science’s ‘The Increasing Potential and Challenges of Digital Twins’ (Nat Comput Sci, 2024). In an extraordinary crossover, DTs have captured the imaginations not only of academic researchers, but also policymakers and practitioners. Enthusiasm spans academic disciplines, domains, and practice. DTs have come of age at a critical juncture, able to leverage the advantages of 21st century data (big, smart, and frequent), computing (power, speed, and distributed systems), and modelling (dynamic, integrated, and flexible) to understand and resolve complex challenges, whether related to transport, infrastructure, or land use.
Although ‘riddled with ambiguity’ (Batty, 2024), the term DT broadly refers to a precise computer representation, or virtual copy, of a machine or system. Unlike typical computer models, a ‘twin’ is likely to be composed of numerous interconnected digital objects that are combined to form a coherent virtual representation of the underlying system (Wagg et al., 2024), often supported by real-time data streams. Building on considerable promise in engineering applications (Ferrari and Willcox, 2024) and potentially ‘revolutionising industry’ (Tao and Qi, 2019), DT approaches have been widely embraced by those who work on human-focused systems (e.g. traffic and mobility or land use). They are also undeniably popular in urban planning and management (Ferré-Bigorra et al., 2022). Nevertheless, these solutions are largely driven by infrastructural, engineering-oriented data, and modelling perspectives, for the most part failing to fully embed humans within the system. DT applications are generally related to the physical components of urban systems, as if cities existed independently of the actions and preferences of their human inhabitants.
This is a problem. Net-zero energy transitions, environmental sustainability, and equitable urban development, to take just a few examples, are complex, wicked, problems that DTs might reasonably be expected to address, given the hype they have generated. A solution cannot be credible if it fails to account for the role of humans and, particularly, the ways in which systemic changes alter – and are altered by – human behaviours and preferences. This presents a significant challenge to the viability of DTs for responding to the grand challenges facing society and raises some important questions. For example, do the current shortcomings of DTs for addressing wicked societal challenges represent a failure of DTs themselves? If they are failing, does the responsibility fall to the researchers who have not created sufficiently innovative solutions or more vocally articulated the limitations of DTs for wicked social and urban problems? Or are current efforts making adequate progress and the missing key ingredient is patience?
The defence: Evidence in support of digital twins
DTs are compelling to urban and social scientists for several reasons. Perhaps the most exciting development paving the way for DTs beyond engineering applications is that of City DTs. City DTs are virtual replicas of cities designed to simulate urban processes by integrating discrete models of various city components and connecting them to real-time data feeds (Papyshev and Yarime, 2021). Examples are emerging in places like Turin (Boccardo et al., 2024), Zurich (Schrotter and Hürzeler, 2020), and Singapore (Shahat et al., 2021). By dynamically integrating many aspects of urban systems, City DTs can enhance urban service provision, provide real-time adjustments when systems are impacted, and can elucidate unexpected ways in which different elements of an urban system are interconnected.
City DTs showcase a number of strengths of DTs:
The case for the prosecution
DTs have also generated a great deal of scepticism, with many in the social sciences discounting them as mostly hype (e.g. Nochta et al., 2021). Some of this may be disappointment that DTs seem unable to address many critical social and urban challenges. But there is also a strong vein of valid recognition of DT weaknesses that has emerged.
A few key shortcomings we identify are: The absence of social processes in DTs is not entirely unexpected, though. People are ‘messy’ and exhibit ‘annoyingly difficult-to-predict behavior’ (Fotheringham, 2023), which makes them extremely hard to model. And even when we can understand and model human activities reliably, the required data to properly constrain their behaviour are either non-existent, stored privately by large corporations, or are so sensitive that we may not want to make use of them anyway (Papyshev and Yarime, 2021). Coupled with this is the difficulty that social systems are often associated with long-term dynamics (in contrast to the typical short-term dynamics of many DT applications) that are very difficult to account for in models that run for longer periods of time (Nat Comput Sci, 2024).
Our verdict
Society faces enormous and complex challenges – challenges that urban and social researchers have an important role to play in addressing. Solutions will require complex and integrated models, data, theory, stakeholder engagement, and expertise. At this time, however, DTs are not up to the task. Although a number of supposed ‘City Digital Twins’ have emerged, they are typically more akin to highly detailed 3D maps than models of societal and social processes. However, our purpose with this Editorial is less to counsel scepticism and more to encourage realistic expectations, with a dose of cautious optimism. After all, the evidence in favour of DTs is strong and positive – and revealed preference suggests urban and social researchers continue to be captivated by DTs, finding them conceptually and practically appealing for a range of applications. There are also encouraging developments that may be slowly bringing us closer to the realisation of ‘true’ societal twins. Considering the efforts of our own group, we have conducted work that attempts to better understand and quantify the uncertainty in DT-like approaches for simulating people (McCulloch et al., 2022), we have combined a range of different approaches to estimate the spread of disease (a precursor to a system akin to a DT) (Spooner et al., 2021), we are developing the tools to allow us to integrate disparate models of human systems, 2 and we are working directly with policy makers on projects that aim to move towards the realisation of DTs, where people and behaviour play crucial roles. Even if a ‘true’ societal DT is not on the foreseeable horizon, at the very least they offer a visible rallying point and a potential common ground that brings together necessary expertise and technologies.
If we think about a related domain, that of transport systems (which is arguably the closest system to the wicked ones that we mention here), it has a many decades long history in the development and application of advanced models for understanding and prediction. Wicked systems do not (yet?) have the same history of expertise on which to draw in the development of DTs. So perhaps our final verdict is not that DTs have failed to deliver solutions to wicked social problems, but that we need to be patient.