The Problem with Smart Cities

8 min readSep 21, 2021

If we do not confront the direction Smart Cities are heading in, they will benefit the rich and powerful, destroy communities, and exacerbate the climate emergency.

Hi, here’s why you should trust me.

I am a Software Engineer at Microsoft’s Azure IoT (Internet of Things) group. We are a group focused on enabling IoT solutions through PaaS and SaaS offerings including Azure IoT Hub and Azure IoT Edge and Azure IoT Central. (PaaS and SaaS just mean we write code to give to other people either to write code on top of to create their own applications, or to click through to create the own solutions.) I focus specifically on the core software infrastructure to connect devices to our cloud backend. I know a thing or two about MQTT, write in Python and Javascript predominantly although I studied Electrical Engineering with a focus in Data Science, and on the side I obsessively study Urban Design and Smart Cities. So I know a thing or two about IoT, Smart Cities, and the ecosystem surrounding them.

So what’s this about?

There’s a fundamental problem with the way many cities, countries, companies, and global leaders approach Smart Cities. And I’m really worried about it. While there’s going to be a lot of back-patting and “good job team!”s to be had at conferences and office email threads, and a lot of money to be made along the way, there’s a real, ongoing risk that this whole… experiment, the whole IoT Smart Cities Smart Technologies space, will actively make the cities worse. It comes down to the fact that oftentimes “Smart” is conflated with “Technology”, and “Smart Cities” end up looking pretty, well… dumb. Let me explain.

What is a Smart City?

Before we hop into it, I want to ask you, earnestly, if you are in the Smart Cities space especially, to think what exactly is a Smart City. If you think the answer is easy, you’re wrong. And if you think the answer has just to do with your IoT product that you’re trying to pedal to the mayor of Montreal or president of Egypt, you’re also wrong. The term Smart City has bounced around over time, and it’s origins are murky. Some claim IBM and Cisco were the first to coin the term in the mid-2000s, associating it with tech-enabled city planning and coordination departments. For example, IBM worked with the city of Rio De Jaineiro, ultimately winning the “World Smart City Award” in 2013. Despite much hullaballoo, the NASA Mission-Control looking command center ultimately failed to create much in the way of change.

Others in the Smart City space, myself included, seek to compartmentalize the role of tech in a Smart City as one component of many components in a multifaceted effort. The Wikipedia page on the Smart City reflects this, pointing that some consider Bletchley Park one of the first Smart Cities or Smart Communities. Bletchley sought to bring together some of the greatest minds to solve pressing problems, at the time decoding Nazi radio transmissions, and did so in a novel, community oriented way.

The Many ‘Smarts’ of a Smart City

A common transgression of Big Tech is bending any problem in such a way that their services can then solve it. The work in Smart Cities is no exception. Part of why the problems of cities haven’t been solved for thousands of years is that they are extraordinarily complicated. Engineers may be tempted to believe that cities can be made Smart through IoT and the quantization of everyday life into discrete sensor data, but beneath the 3D Digital Twin modelling and clean data processing on well understood trends, there’s real people living imperfect and informal lives.

And the data they collect is, like in all statistics related fields, woefully biased. Whether the bias is driven by corporate profitability, middle management seeking promotions, or pure unintentional blindness, it is already impacting digital systems in cities across the globe. This expanded view of Smart Cities beyond the IoT quickstart “pressure and humidity sensor” fever dream more accurately captures how there can be many different ‘Smarts’ used in Smart City initiatives, and also welcomes the informality common in many cities across the globe.

What is driving the current focus of Smart Cities?

Behind the labyrinth of jargon of the field exists a slowly globulating, slowly actualizing set of focused use cases where investments are funneled and revenue is generated. However one can look even further upstream at the key technologies and innovations enabling modern Smart City initiatives. Ubiquitous Computing and the rise of cheap tiny computers, like the Raspberry Pi Zero, enables the tiny sensors that can be deployed in buildings at an extremely low cost. Fast development using high level programming languages like Python can be used by companies large and small for prototyping initial cloud-connected sensor designs. The explosion of AI / ML cloud computing services can extract novel insights from otherwise opaque datasets. And innovations in connectivity, through mesh networking, 5G, and high speed broadband, among others, enable the consolidation of previously isolated sensors and datasets.

The technologies are then deployed for business. Consider an elevator company. Its R&D department places sensors on all their new elevators, and after collecting data for months determine a way to predict through the data collected when a machine is about to break. A simple to understand and clear use scenario, IoT leaders coin this set of solutions Predictive Maintenance. This is the first key investment area.

Consider also the state department of transportation managing a freeway tunnel. The department installs a network of cameras and sensors along a tunnel, which sense if air particulates reach a high level, alluding to a traffic jam, and alert first responders if a crash is detected, or a cleanup crew if there’s debris in the roadway. These investments would improve commute times and in a small, impactful but nearly invisible way, help people’s lives. They are called Smart Infrastructure.

Finally companies have developed solutions using Computer Vision and high quality cameras to monitor people. A ‘Worker Productivity Platform’, where a company installs AI sensing cameras in a factory to detect when workers are taking breaks longer than they should be, or idling while on the job, can report inefficiencies to management and save the company money. While controversial and arguable unethical, cameras monitoring everyday life will continue growing in adoption for security and other purposes. Big Brother, some might say, and cry foul, and efforts have been made to legislate against this application of IoT in cities. But the pressure likely will not be enough to stop a large adoption of camera monitoring in Smart City spaces.

There are more examples, including the residential IoT solutions like Smart Home cameras or lightbulbs or switches, but I won’t get any further into these examples because their impact on the Smart Cities space is marginal. The other examples that I have described roughly capture the main uses of IoT solutions for Smart Cities to date.

The Problem with Today’s Vision for Smart Cities

As seen in Songdo and Rio De Jaineiro, the vast majority of current efforts to create Smart Cities from scratch or bootstrap existing cities into Smart Cities have failed. The few bright points of success for the industry paint a bleak picture of our ability to ‘Change the World’. Oftentimes it is the Big Brother application space that finds traction, as governments use CCTV and facial recognition to track individuals, citing their effect of encouraging good behavior. While cameras on police, buildings, and transit might theoretically create safer spaces, they agglomerate a huge amount of power within the hands of the organizations that control them, and can bolster the power of oppressive or manipulative regimes.

Ultimately as an industry, Smart Cities will continue to fail or empower the oppressor as long as it fails to confront it’s own blind spots. Principle among them is recognizing bias in data. Data processing tools quantize the world into digital metrics for modeling and decision making, which sounds good in principle. But when you quantize a signal, like with an Analog to Digital Converter, you lose data, and you transform the signal in sometimes entirely unknown ways. Smart City efforts have reduced complex problems to whatever data is available to process. And that sounds good to most of us. Data Driven, right? Yet rather than removing bias outright, we’ve simplify codified bias into our data collection, and lauded the subjective as objective. For cities, this is dangerous. The bias of large corporations and their municipal partners investing in Smart Cities risks entrenching and investing in ways that destroy the very fabric of cities, and exacerbate existing climate emergencies. Power brokers in the Smart Cities space, who oftentimes have no interest in acknowledging or disclosing the limits of their data driven processes, must be challenged.

Key Example: Electric Cars and Smart Roads

In the case of creating electric autonomous vehicles or improving traffic flow the signals analyzed, Well-to-Wheel C02 Emissions and Travel Time Reliability respectively, do not include data on Household Carbon Footprint or Pollution from Tyre Wear. The conclusions drawn from the first set of data, that we need to invest in electric vehicles and smart road-use technology, contradict the conclusions of second set of data, that rezoning and retrofitting urban areas for higher density should be prioritized above optimizing existing single occupancy vehicle infrastructure. This is not even considering data on the affect roadbuilding in America had on destroying Black communities, and removing independence for the very young and very old who cannot drive. The bias in data collection can very clearly skew the outcomes of decisionmaking, and IoT’s applications in Smart City design is no exception.

The Power Broker Bias

So I want to call this ability for powerful companies in the urban design space to bias existing data collection in harmful ways to society the Power-Broker-Bias (in part referencing Robert Moses, the NYC Power Broker who sought to destroy vast portions of the city in order to build more roads). When the Power-Broker-Bias is used to define the investments of Smart Cities, investments are made based on sound data analytics toward solutions that will ultimately create worse physical spaces. Microsoft has defined AI principles to seek to reduce the bias and harm caused by AI systems. We should as a community focused on Smart Spaces and Smart Cities define Smart City Principles. They could look something like this:

  • Balance — Smart City initiatives should balance Orchestration Empowerment and Implementation intelligence.
  • Inclusiveness— Data collection and decision-making should be shown to not statistically harm marginalized communities within cities.
  • Strength — Smart City initiatives should contribute toward creating Strong Cities by developing metrics and solutions that strengthen the community fabric of existing cities.

Whatever the principles end up looking like, the need for them is urgent. If we do not proactively guide the development of Smart City initiatives and IoT solutions for Smart Cities and Smart Spaces, there is a grave risk that new innovations and new efforts will entrench existing bias, exacerbate the climate emergency, destroy strong civic communities, and disproportionately benefit the powerful and rich. If that is what Smart Cities are for, that’s fine. We must be honest about it. But to say Smart Cities are about making the world a better place, about improving the lives of all people, when already the impacts are moving in the opposite direction, suggests it’s all smoke and mirrors.




Software Engineer passionate about the future of cities. Currently building libraries for Azure IoT.