SWITCHED ON
The daily technology series nobody asked for but everyone needed
The City That Watches: Smart Cities and Urban Technology
Cities are being instrumented with sensors, cameras, and data infrastructure that promise to make them run better. They will. The question of who controls the data those systems generate, and what happens when the interests of the city and the interests of the platform managing its infrastructure diverge, has barely been asked.
A city is a political community. It has residents who have rights, democratic processes through which collective decisions are made, and accountable institutions through which those decisions are implemented. A platform managing a city's data infrastructure is a corporation with shareholders, proprietary systems, and interests that may or may not align with those of the people living in the city. The relationship between these two things is poorly defined everywhere it exists.
Yesterday we brought the surveillance economy home — smart speakers listening continuously, smart TVs watching you watch them, Ring's doorbell camera network integrated with law enforcement, IoT security vulnerabilities that have been documented and not fixed, and smart home technology weaponised in domestic abuse contexts. Today we step outside and look up at the city itself. Smart cities: the urban data infrastructure that monitors traffic flows and optimises signals, tracks air quality and energy consumption, manages public transit and waste collection, and increasingly does all of this through systems built and operated by private technology companies with proprietary platforms and commercial interests that may not align cleanly with those of the residents the systems are ostensibly serving. This is the episode where everything from the previous eleven comes together at the scale of a metropolis.
01 — What Smart City Technology Actually Does
The smart city concept encompasses a broad range of technologies applied to urban management, and it is important to distinguish between the mundane, genuinely useful, and relatively uncontroversial applications and the more ambitious, more contested, and more surveillance-intensive ones, because they get conflated in both the promotional material and the criticism.
At the genuinely useful and relatively uncontroversial end: adaptive traffic signal control, which uses sensor data on traffic flow to adjust signal timing in real time and demonstrably reduces congestion and emissions; smart street lighting, which dims automatically when streets are empty and brightens when occupied, reducing energy consumption meaningfully; environmental monitoring networks that track air quality, noise, and flooding risk in real time and enable faster public health and emergency responses; smart waste management that optimises collection routes based on sensor data from bins rather than fixed schedules. These applications collect data about conditions and flows, not about individual people, and deliver clear public benefits with limited civil liberties implications. They exist in most major cities to varying degrees and their track record is broadly positive.
At the more ambitious, more contested end: real-time population movement tracking through mobile phone data and smart card transaction records; integrated surveillance camera networks with facial recognition capability; predictive policing applications that use urban sensor data to direct police resources; social credit-adjacent systems that link urban services to behavioural compliance. The spectrum runs from optimising bin collection to population surveillance, and they travel under the same brand. This requires disambiguation.
02 — The Sidewalk Toronto Cautionary Tale
Sidewalk Toronto — the project in which Alphabet's Sidewalk Labs proposed to develop a twelve-acre site on Toronto's waterfront as a technology-enabled neighbourhood from the ground up — is the most extensively documented case study in the gap between smart city ambition and democratic reality, and it deserves detailed treatment because its failure revealed something fundamental about how the relationship between cities and technology companies needs to work.
Sidewalk Labs proposed a comprehensive vision: sensor-embedded streets, modular buildings with flexible uses, data-driven traffic management, heated sidewalks melting snow automatically, automated waste management, and a range of urban services managed through digital infrastructure. The project attracted significant public attention, significant public concern, and eventually a comprehensive critique from Ann Cavoukian, Ontario's former Information and Privacy Commissioner, who resigned from an advisory role citing inadequate privacy protections. The core concerns were about data governance: who would own the data generated by the neighbourhood's sensor infrastructure, what Alphabet's commercial rights to that data would be, and whether residents would have meaningful control over the surveillance environment they lived in.
Sidewalk Labs withdrew from the project in 2020, citing economic uncertainty around the pandemic. The company maintained that the privacy concerns had been substantially addressed in its revised proposals. Critics maintained that the fundamental tension — a commercial entity with data monetisation as a core business logic operating the infrastructure of a residential neighbourhood — could not be resolved by design choices alone but required structural data governance that the project had not adequately established. The withdrawal left the question unresolved but the lesson was clear: cities cannot outsource the governance of their data infrastructure to technology companies without first establishing who owns the data, who can access it, under what circumstances, and with what accountability.
Sidewalk Toronto failed not because smart city technology does not work but because the question of who governs the data it generates was never adequately answered, and a neighbourhood's residents were not willing to find out the answer after the fact.
03 — The Surveillance Camera Question
London has one of the highest concentrations of surveillance cameras of any city outside China — approximately 942,000 CCTV cameras across Greater London, roughly one camera for every nine people. This figure includes private cameras in shops, offices, and residential buildings as well as public cameras operated by Transport for London, the Metropolitan Police, and local councils. The network is extensive, partially integrated, and the subject of ongoing controversy about its effectiveness as a crime deterrent relative to its civil liberties cost.
The addition of live facial recognition to elements of this camera network — which the Metropolitan Police has been deploying at specific events and locations since 2019 — has significantly escalated the civil liberties implications. Facial recognition in public spaces converts an already extensive camera network from a passive recording system, accessed retrospectively for investigation purposes, into an active real-time identification system that can flag individuals of interest as they move through the city. The legal framework governing this deployment remains contested: the Court of Appeal found in 2020 that South Wales Police's use of facial recognition lacked sufficient safeguards, but did not declare the technology unlawful in principle, and deployment has continued under various governance frameworks that critics argue remain inadequate.
China provides the most extensive example of what a fully integrated urban surveillance and facial recognition infrastructure looks like at scale. The Skynet and Sharp Eyes systems cover hundreds of millions of cameras integrated with AI analysis capability, linked to national ID systems, and used for everything from traffic enforcement to political monitoring. The Chinese system is not a cautionary tale about where Western smart cities are inevitably heading — the political contexts are substantially different — but it is a demonstration of the technology's capability when deployed without the civil liberties constraints that democratic societies should be imposing, and it informs the concern about incremental deployment in democracies where each step seems modest and the cumulative effect is substantial.
04 — What Good Urban Data Governance Looks Like
The smart city concept does not have to mean the surveilled city. The technologies involved are mostly tools whose impact depends on governance choices, and several cities have made governance choices that are worth examining as models.
Barcelona's smart city initiatives have been explicitly framed around data sovereignty — the principle that data collected from citizens in public spaces belongs to the city and its residents, not to the technology vendors providing the infrastructure. The city's Urban Data Office manages data governance, procurement contracts require vendors to provide open-source software and city-controlled data infrastructure rather than proprietary locked systems, and the city has been an active voice in European policy advocating for public data infrastructure as an alternative to commercial platform dependency. The approach is more expensive and more administratively complex than simply procuring turnkey systems from large technology vendors. It is also more consistent with the principle that urban data infrastructure is public infrastructure, like roads and sewers, and should be governed accordingly.
Amsterdam has developed a smart city initiative built around citizen participation and transparency, with a public registry of all sensors deployed in the city — what they measure, who operates them, what data they collect, and what it is used for. This is a minimum standard of transparency that is not universal and that enables the kind of informed public debate about specific applications that most cities currently cannot have because the information is not available. The registry approach is replicable at low cost and should be a standard requirement for urban sensor deployment globally. It is not.
05 — The City as a Platform — and Why That Is the Wrong Metaphor
The dominant framing in the technology industry's engagement with cities has been the city-as-platform: urban space as an operating environment for services that can be optimised, personalised, and monetised through data, with technology companies providing the platform layer on which the city runs. This metaphor has driven significant investment, some genuine innovation, and a set of governance assumptions that are poorly suited to what a city actually is.
A city is not a platform. It is a political community with accountable institutions, democratic processes, and residents who have rights that cannot be overridden by terms of service. The relationship between a city and its residents is fundamentally different from the relationship between a platform and its users — citizens cannot easily leave, they did not meaningfully choose the infrastructure, and they have democratic standing to demand accountability that users of commercial platforms do not. Urban data infrastructure managed by commercial entities on proprietary platforms creates accountability gaps that have no equivalent in conventional infrastructure: when the water company fails, there is a regulatory framework with defined accountability and legal remedies. When the smart city platform vendor changes its data practices or terms of service or exits the market, the governance vacuum can be immediate and consequential.
The path forward is not rejecting urban technology — the genuine benefits of data-driven urban management are real and significant — but insisting on a model where cities own their data infrastructure, require open and interoperable systems from vendors, publish comprehensive transparency registries for sensor deployments, prohibit real-time biometric surveillance in public spaces without specific democratic authorisation for specific purposes, and build procurement requirements that treat data governance as infrastructure governance rather than as an afterthought in a technology contract. These are not radical positions. They are the minimum conditions for urban technology that serves residents rather than extracts from them.
Tomorrow we are making a deliberate turn from the systemic to the personal — specifically to something that does not get discussed enough in technology coverage because it sits at the intersection of technology and human psychology in ways that are hard to quantify. Digital addiction, attention, and what the deliberate engineering of compulsive behaviour in digital products is doing to the human capacity for sustained thought. See you then.
Switched On is a daily technology series covering the ideas, systems, and arguments shaping the digital world. Opinionated. Witty. Occasionally wrong. Always worth the argument.
