The folly of Smart Cities: lessons for Digital Twins.

Adam Greenfield's 2013 pamphlet 'Against the smart city' is an indictment of a movement. It reveals the hubris of self-interested technology companies, their lack of self-awareness, and their tendency to unintentionally promote authoritarian views of society. In internet speak, Adam 'owns' the Smart City movement. Why is this relevant in 2021? Because, aside from a few cosmetic differences, the Digital Twin movement could easily fall victim to many of the same mistakes.

I'm going to summarise Greenfield's key arguments as applied to Digital Twins below, but by all means read the pamphlet, it's only 100 pages long and worth the small investment just to enjoy his acerbic prose first-hand.

I'm also going to use quotes liberally in this post because Greenfield has a way of concisely expressing complex ideas that I admire greatly, and which require little adaptation to apply to Digital Twins. Rather than trying to appropriate his thoughts as my own, I have instead merely tried to shape them into a narrative that applies to Digital Twins.

So here are 5 things that the Digital Twin movement must learn from the failure of Smart Cities...

1. Digital Twin technology is limited.

Technology breaks. By putting technology front-and-centre in the fabric of our built environment we create a whole new genre of failure modes that simply don't exist in conventional urban areas. The same will be true if we use Digital Twins to govern how our infrastructure operates. Greenfield writes that we risk trapping ourselves in "'code/space'... places whose possibilities are activated by the arcane workings of computational systems, to the degree that they are incapable of functioning as intended should those systems default."

This doesn't mean that we should seek to avoid augmenting the world around us with technology. But that in doing so, Greenfield argues, we should "design with the particularity, and the particular limitations, of specific technical systems firmly in mind."

We should also be aware that technology becomes redundant quickly. Nobody wants to live in the Smart City equivalent of a Blackberry. As Greenfield writes, "the problem with over-specification is simply that it leads to a curiously static conception of the future: once we get there, we stay there."

"The proximate future is a place we never quite get to."

Whilst Digital Twins have less of a real-world presence than some Smart City technology, we should still bring the same humility to what can be accomplished with technology. We should avoid creating Digital Twins that ignore intangible 'soft' factors in favour of technocratic 'optimisation'. We should accept that our Digital Twins will grow redundant, and work with prevailing technological trends even where they may undermine the purity of our Digital Twin vision.

As an example, Greenfield observes in the Smart City literature that "it's positively weird how rarely these visions invoke the on piece of networked information technology that city-dwellers all over the planet already have ready to hand." Perhaps this is because the smartphone emerged whilst the Smart City movement was gaining pace and so didn't fit neatly into the command-and-control vision that Smart City designers had in mind, and so they conveniently neglected it.

In short, we should design for the future, without trying to design the future. We should create modular systems that are intended to evolve and change. And we should never kid ourselves that our work is finished.

2. Life is complex, and stochastic, and Digital Twins should reflect that

Smart Cities like Masdar and Songdo are lonely. By taking a technology-focused approach to sculpting the spaces in which we live our lives, they miss out on the crucial cultural factors that make our cities vibrant liveable.

Behind much Smart City (and thus Digital Twin) marketing, Greenfield argues, lies "unreconstructed logical positivism, which... holds that the world is in principle perfectly knowable, its contents are enumerable and their relations capable of being meaningfully encoded in the state of a technical system without bias or distortion." The architects of Smart Cities have attempted to quantify our lived experience as mere variables in an equation, and have then attempted to solve that equation (for example by minimising commute time, or maximising traffic flow and green space).

"The designers of informatic systems have historically treated the environment in which their products and services are used as an abstraction, 'pure background'."

By expanding logical positivism beyond the urban environment to entire national infrastructures, the Digital Twin movement risks oversimplifying on a spectacular scale. Society does not have an optimisation function "what may be perfectly appropriate in a hierarchical, highly structured organisation with known, quantifiable goals is fundamentally unsuitable to the protean entities we know as cities." We cannot successfully approach improving real world outcomes through information technology if we bring a "discomfort with unpredictability, a positive terror of the unforeseen and emergent."

Instead we must become comfortable with Digital Twins that treat outcomes as probabilistic. When applied to real-world decision-making this means not treating the world as deterministic and seeking to make the 'best decision', but actually building uncertainly into our models, calibrating against real world outcomes, and seeking to avoid unlikely but detrimental outcomes. This necessitates a Rawlsian approach that minimises injustice, rather than a Utilitarian approach the seeks to ensure the greatest aggregate good.

3. Digital Twins should empower normal people

If there is one theme that emerges repeatedly across Greenfield's writing it is the tendency for technology to reinforce (and indeed exacerbate) existing societal inequality. The internet that we use today is a far cry from the egalitarian online communities that emerged in the early 90's, let alone its countercultural 'Whole Earth Catalog' roots.

By creating a command-and-control mindset to the urban environment, Smart Cities risk exacerbating the biases in law enforcement and public services that already exist. To avoid such outcomes it's not enough to adopt technocratic pretensions of being unbiased, rather our Digital Twins must actively seek "to amplify the abilities of citizens and communities to determine the conditions of their own existence."

"The sense that citizens themselves may wish to available themselves directly of the information being gathered on their behalf is almost surreally absent from the [literature]."

In practice this means building decision-making logic (algorithms) into our Digital Twins that reflects the values of the communities that they effect, rather than merely echoing existing government or corporate measurements. As Greenfield writes, "it would be unconscionable if these systems were anything less than fully accountable to the communities and individuals whose lives they do so much to shape."

Crucially, these arguments apply doubly to the use of individual's data. We all know that massive data collection will be the lifeblood of Digital Twins, like Smart Cities before them. However, Greenfield argues that we cannot adopt an 'ends justify the means' approach to data collection on this scale, "if we are to embrace ubiquitous data collection and the other technics of computational oversight... they [must] be placed at the full disposal of an engaged citizenry, with the understanding that such tools should be used to provoke debate rather than forestalling it."

Many of the arguments for both Smart Cities and Digital Twins are premised around some supposed public good. However, we cannot put ourselves in a situation where the arbiters of what constitutes public are the very same people seeking to profit from Digital Twins.

4. Digital Twins will become political

When we see the world through a technocratic logical-positivist filter, it is hard to understand how our investment in Digital Twins could ever be conflated with something as shallow and unbecoming as politics. Digital Twins, like Smart Cities, don't have opinions. They are merely ways of improving people's lives through technology. And yet what is politics if not a debate over the division of scarce resources? And what is an effective Digital Twin if not a better way of using scarce resources? If successful, the motivation of public sector Digital Twins will be suspect, the biases of their programming will be a source of public debate, and it will be difficult to separate the technology from the people who seek to wield them.

"The authorship of an algorithm intended to guide the distribution of civic resources is itself an inherently political act."

For technology to rise above politics, Greenfield advises that we check our privilege. We must acknowledge that because we are interacting with the real world, social outcomes will be impacted our Digital Twins. The risk of negative impacts demands that we "[design] with concerns about power, privilege and justice at their very heart," and in doing so "seek not some shallow optimum but a fruitful balance of contending prerogatives."

Where the Smart City movement descended into folly was where it treated real world outcomes as "a simple matter of keeping performance indicators balanced between nominal thresholds." Our experience of our built environment will never be as simple as an SLA for how quickly our waste is picked up, or how many screens we have on the walls of our home. As the ambition of Digital Twins is arguably even greater than that of Smart Cities, we must be wary of applying the type of 'KPI and SLA' mindset that we use within our organisations to wider society. Interacting with the real world will mean understanding and making difficult trade-offs, as Greenfield argues, we ought to be wary of claims that the application of some single master algorithm could result in an Pareto-efficient distribution of resources."

We must consequently be open to being surprised by our Digital Twins, to uncover unintended consequences of their logic, and to change course quickly if it becomes apparent that we are doing more harm than good.

5. Digital Twins will be abused

"the data is never just the data"

Unexpected consequences are one thing, but if Digital Twins are anywhere near as powerful as advocates suggest, then we must also be ready for amoral and immoral users. It is an unfortunate truth that many of the technologies that feature in both the vision of Smart Cities and Digital Twins have most extensively leveraged by non-Democratic regimes. The need for extensive data collection to support these technologies require a degree of data stewardship and ethics that is often lacking in practice. As Greenfield writes, "[smart] technologies mesh particularly well with an authoritarian government's interest in monitoring dissenters, anticipating likely sources of resistance and forestalling... acts perceived as challenging the government's claim to legitimacy."

The urge to measure everything, to cover the world with sensors and the internet with cookies, will inevitably improve the accuracy of Digital Twins, but it also creates a substantial paper trail that's ripe for abuse. As Greenfield argues, "every act in the smart city is a formal one... every behaviour is observed." This is not a world that ordinary people have necessarily consented to live in. Moreover, it's not the type outcome that we have been primed to expect when Smart Cities and Digital Twins are billed as making the built environment better for humans, rather than as an intrusion of government into henceforth ungoverned aspects of society.

It's as if someone took Minority Report as a shopping catalogue or a punch list rather than a vision of dystopia.

I'm not a Smart Cities expert, but I suspect that the lasting impact of the Smart Cities movement won't be in the megaprojects like Songdo and Masdar. Instead, the value from making our cities smart will come from incremental improvements to existing services that reduce the burden of maintenance and failure. Applying the same logic to Digital Twins means focusing on what we can control (e.g. how we anticipate and/or respond to disruption and failure across complex infrastructure systems). In turn this means ignoring what we can't, or shouldn't, control (e.g. culture, society, people's behaviours).

I was lucky enough to interview Adam Greenfield recently for the Digital Twin Fan Club podcast, so keep your ears peeled for that episode when it comes out. 'Against the smart city' is available as a print-on-demand paperback here.