4
In January 2019, the Toronto Region Board of Trade floated a trial balloon in an attempt to defuse the controversy that had swirled for a year and a half around Sidewalk Labs’ closely watched plans for Quayside, a twelve-acre brownfield site on the city’s waterfront.
Google/Alphabet’s smart city plan turned on fitting out the entire project, which would include numerous new buildings and public spaces, with thousands of sensors designed to gather massive quantities of data from a heavily wired neighbourhood. That information would then be sliced and diced in all sorts of ways, from specific energy and infrastructure operations to more open-ended applications, such as the management of public spaces.
The company had also promised that independent firms, including start-ups, could have open access to the raw data and use it to manage services and develop apps that could eventually be scaled up and deployed in other cities. Sidewalk Labs called this approach to digital city-building its ‘platform’ strategy – a business model not unlike Apple’s app store, and more than a little open-ended.
Activists and pundits had attacked Sidewalk’s proposal, zeroing in on a few fundamental questions: Who would own all that data? How was it to be used? Could Sidewalk’s sensors somehow identify individuals and target them for ads … or worse? Who would profit? And, finally, were Canada’s privacy laws adequate for regulating the collection of all this urban information?
During a period when the manipulation or outright misuse of individual data by tech giants like Google and Facebook had provoked a ‘tech-lash,’ it seemed clear that Sidewalk’s pitch would live or die based on how the company’s planners addressed these core issues.
The Board of Trade’s solution, dubbed BiblioTech, seemed beguilingly elegant and politically benign: entrust all that data to the Toronto Public Library, a well-loved local institution that happens to specialize in managing information. This ‘data hub,’ according to the board’s recommendations, would be overseen by the Information and Privacy Commissioner of Ontario, with the library in charge of developing policies for data collection and use.
Similar but subtly different ‘data governance’ proposals had also surfaced – among them a pitch from provincially owned tech incubator MARS for a ‘civic data trust,’ defined as ‘a trust that is established to manage the digital layer of a smart city.’ According to MARS, this new trust’s assets ‘may include the physical infrastructure (sensors and data warehouses), code base (database, standards, processing structures, and interface) and data that make up the digital layer. The civic digital trust may also manage financial assets to ensure the sustainable operation of the trust.’
With Sidewalk’s plans coming under intense scrutiny from critics who were skeptical about the company’s data strategy and its ulterior motives, it was not surprising that the company’s local supporters were talking about libraries and trusts.
Data is both the opportunity and the flashpoint in most conversations about smart city technology. Smart city hardware and software effectively soak up all sorts of data and transform it into intelligence that can, in theory, improve urban infrastructure, create new services, or add efficiencies to existing ones. Indeed, the promise of smart cities involves capturing very large tranches of so-called fast data and then applying sophisticated analytics to detect patterns or generate predictions. These findings can be used to make cities more livable or sustainable. That, in any event, is the vision.
As the global tech sector well knows, data has enormous monetary value, especially in large batches – the new oil, as the cliché goes. As importantly, data is the not-so-natural resource that is fuelling the development of lucrative artificial intelligence–based technologies. These run the gamut from voice-recognition algorithms and online language translation services to much more ambitious systems that can generate fine-grain recommendations about optimizing urban transportation networks or deploying police officers so they’re working in the areas most likely to experience crime. The development of algorithms that feed off urban data figured prominently in Sidewalk Labs’ plans. ‘The algorithm is where the value is,’ observes Natasha Tusikov, an assistant professor at York University who studies smart cities and data governance.
According to Kurtis McBride, CEO of Miovision, a Waterloo smart traffic signal firm, Canadian policy makers have yet to wake up to an economic reality that global tech giants like Google understand. By the 2030s, he predicts, most urban infrastructure will be fitted out with technology that generates data with significant commercial value. Either private firms will own and profit from it, or the value in those pools of urban data can be used to advance the public good, he says. ‘You have a decision about what kind of future you want.’ Government officials, McBride continues, ‘aren’t thinking about this.’
What further complicates the discourse about data is that the term itself is not only exceptionally broad – akin to talking about ‘mammals’ or ‘transportation’ – but also somewhat nebulous. Data covers everything from databases of building inspection records, recreation program registrations, and census track statistics to signals generated continuously by traffic monitors, smart phones, and GPS devices.
According to scholar Rob Kitchin, author of Data Lives: How Data Are Made and Shape Our World (2021), smart city data is fundamentally different in character from older forms of static information that were commonly used for city-planning purposes. He noted in a 2016 article,
Such data [included] censuses, household, transport, environment and mapping surveys, and commissioned interviews and focus groups, complemented with various forms of public administration records. In general, this data is analysed at the aggregate level and provides snapshots of cities at particular moments. Increasingly, these datasets are being supplemented with new forms of urban big data.
Big data has fundamentally different properties from traditional datasets, being generated and processed in real-time, exhaustive in scope, and having a fine resolution. Rather than data being derived from a travel survey with a handful of city dwellers during a specific time period, transport big data consists of a continual survey of every traveller: for example, collecting all the tap-ins and tapouts of Oyster cards on the London Underground, or using automatic number plate recognition-enabled cameras to track all vehicles.
This transformation from slow and sampled data to fast and exhaustive data has been enabled by the roll-out of a raft of new networked, digital technologies embedded into the fabric of urban environments that underpin the drive to create so-called smart cities. (Kitchin 2016)
The idea, as Kitchin has explained, is that the city as a physical space is discernible on a continuous basis using various sensing and surveillance technologies, including devices – like smart phones – that leave a trail of data on someone’s movements and that can then be used to support urban systems. ‘The instrumented city,’ as he puts it, ‘offers the promise of an objectively measured, real time analysis of urban life and infrastructure.’
In some cities, raw, real-time information from big data streams – e.g., from air quality monitors, traffic sensors, wastewater flow rates – is gathered and marshalled, with the resulting ‘informatics’ pressed into service to describe quantitatively what’s going on in the city at any point in time.
Municipal officials use this kind of data, and the practice has raised questions about the technocratic nature of smart city information gathering. ‘Such instrumentation of the urban environment, however, is not by itself sufficient to have a meaningful impact on the quality, sustainability, and resilience of cities – or more broadly on urban policy and planning,’ Constantine Kontokosta, the NYU informatics scholar, observed in a 2016 paper. ‘Understanding the social, economic, and cultural dynamics of urban life requires both an appreciation of the social sciences and a substantive engagement with communities across diverse neighborhoods’ (67–84).
The installation of smart city sensors, in turn, can create information that didn’t exist in the (unmonitored) past. ‘The whole idea of the smart city is that every interface is a data collection space,’ says Anna Artyushina, a York University PhD candidate who specializes in data governance for smart cities, in an interview.
Case in point: sensors designed to detect if a parking spot is occupied or empty at any particular point in time. If there’s no monitoring device, the spot’s status – taken/vacant – is knowable only to someone who happens to be passing by. But what if there’s a connected digital system that registers the spot’s status in real time and makes this information available to transportation officials or anyone with an app? The resulting data could be used to alter parking rules – maybe the spot is always vacant and could be used for some other purpose? – or even generate revenue: after all, if you need to park, you may be willing to pay to find a location.
We already live in a world that’s programmed to track our movements, our consumer habits, our online behaviour, and our digital interactions, thanks to smart phones, apps, Google searches, social media platforms, and security devices in private spaces like malls and office buildings. Personal data is harvested, aggregated, analyzed, and then sold or shared, often without our knowledge or explicit consent.
Nor can data be considered neutral or apolitical. With the growing deployment of smart city technology, scholars and critics have raised important questions about the values or biases embedded in seemingly objective information. As Kitchin points out, ‘Data do not exist independently of the ideas, techniques, technologies, people and contexts’ that produce and use them in urban settings.
For those reasons, the management of personal and operational data gathered by smart city systems in public spaces (streets, parks, etc.) has become a hot-button topic, and rightly so.
Data and Privacy
Many critics of the Sidewalk plan for Toronto expressed grave concerns about privacy. Could sensors identify individuals who just happened to be on the street, or in a park, for example? There were also privacy questions about other types of systems, such as smart condo buildings that continuously collect energy consumption readings from individual apartments. Could that data be used to make inferences about the occupants’ habits? While Sidewalk initially retained former Ontario privacy commissioner Ann Cavoukian to evaluate its plans using a ‘privacy by design’ approach, she eventually resigned, citing concerns that Sidewalk would not live up to its pledges.
In most big cities and especially in high-traffic core areas, public spaces have long been monitored by public and private CCTVS. In China, the government has installed widespread surveillance networks that extend from smart phones to the widespread use of facial recognition systems and, with the pandemic, location-based tracking apps. As the New York Times noted, ‘officials in some places are loading their apps with new features, hoping the software will live on as more than just an emergency measure.’
At the other end of the spectrum is the European Union’s General Data Protection Regulation (GDPR), which is considered to be the world’s ‘strictest’ privacy legislation, according to Anna Artyushina. California legislated a comparably strict consumer privacy framework that went into effect in early 2020, and the GDPR has influenced privacy reforms in other countries, including Canada. While the GDPR has broad applications in the private sphere (e.g., the law regulates the use of cookies and establishes the legal right for individuals to ‘be forgotten’), its core principles are also highly relevant for smart city applications.
Traditionally, privacy laws were constructed around the principle of consent, which means that entities collecting information on individuals had to obtain their agreement – a cumbersome process that involved the kinds of legal language found in the fine print of many forms of software.
Some of the most progressive versions of privacy law have rejected the consent approach and instead pursued a different way of framing legal privacy. According to an analysis by Artyushina published in 2020, the GDPR is centred on four pillars of data protection: purpose specification, data minimization, automated decisions, and special categories. As she writes,
The requirement of purpose specification states that personal data must be collected for a ‘specific, explicit, and legitimate’ purpose and cannot be further ‘processed’ in a way which is ‘incompatible’ with the original purpose. Data minimization means keeping data collection to the bare minimum required for data collectors’ operations. The notion of automated decisions grants European citizens the right to opt out of automated decision-making. The provision on special categories prohibits companies from gathering and processing data ‘revealing racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, and the processing of data concerning health or sex life.’ Additionally, the GDPR makes it a data subject’s right to transfer their information to another service provider or to require the data controller to delete certain information about them.
Outside the EU, some cities have adopted similar principles in their smart city strategies. The City of Boston, for example, says it ‘collect[s] as little data as possible to solve a particular problem’ and has solicited privacy advice from the American Civil Liberties Association.
Aspects of the GDPR can be found in Canada’s privacy legislation, including some recent amendments to national privacy laws (provincial governments also enact privacy laws that apply to municipalities). ‘However,’ Artyushina notes, ‘the Facebook/Cambridge Analytica scandal in 2018 demonstrated that the country’s privacy protection laws may be ill-equipped to deal with technology companies.’ York’s Natasha Tusikov adds that Sidewalk Labs’ plans for collecting what it called ‘urban data’ – i.e., all the various types of information gathered by sensors installed in public spaces and the buildings within the Quayside areas and then used to operate infrastructure or services – exposed a gaping hole in Canada’s privacy laws.
Data Bias
The best-known and best-publicized instances of data bias involve facial recognition apps. Because of the way the algorithms were developed, facial recognition systems were more likely to fail when asked to identify Black or Asian faces. The reason? The systems were ‘trained’ to make matches by using huge collections of images of faces that were predominantly white – an example of what’s known as ‘algorithmic bias.’
But data bias manifests itself in more subtle ways, too, and even within urban systems that aren’t normally considered to be political. For example, when cities across North America began setting up 311 call centres, they weren’t positioned as smart city systems. Rather, proponents saw 311 as a means of improving both citizen engagement and bureaucratic accountability. Over the years, 311 services have become increasingly tech-enabled, with social media accounts, apps, and the release of machine-readable complaint-tracking records through open data portals.
Municipalities now sit on huge troves of 311 call data – hundreds of thousands or even millions of requests per year – that can be mined and analyzed, and then used to inform municipal planning and budgeting. After all, a proliferation of calls about basement floods, missed garbage pickups, or dubious odours from a nearby factory can give officials important clues about what’s happening in a neighbourhood, as well as the performance of city departments. If scanned carefully for longer-term patterns, 311 calls may also offer predictions about future problems.
These call records certainly qualify as ‘big data.’ But the ways in which this information is or can be used also offers important lessons, both positive and negative, about applications for other large urban data sets that might be generated by smart city technologies, sensors, and other systems.
One obvious question: How do municipal agencies make decisions on how or when to respond to residents’ requests for service? Researchers who study 311 data have found that with many municipalities, such decisions tend to be made in a black box, with little transparency (e.g., first come first served, a triage system, etc.).
These data sets also contain important patterns that could assist in making service delivery either more efficient or more equitable (which aren’t necessarily the same thing). The wrinkle is that cities need to understand the conditions that motivate residents to call 311. ‘We know that people don’t complain at the same rate,’ says Constantine Kontokosta. ‘[A]n individual accustomed to seeing rodents in their building may be less likely to complain than someone seeing a rodent in their apartment for the first time. In addition, individuals may have different levels of trust in government, differing expectations that the government will actually respond, and socio-cultural traits that make them more or less likely to report a problem.’
Another pattern, noted by a New York State Health Foundation/ Harvard research team in a 2020 study, found that spikes in calls about a particular problem may actually be orchestrated community campaigns meant to force municipal officials to address an issue. The study described the practice as a ‘misuse’ that could lead city officials to ‘erroneously’ conclude that an area was seeing some kind of decline.
A further evaluation, published by Kontokosta in 2017, looked at complaints by New Yorkers about hot water problems in their buildings. ‘We wanted to assess where bias was occurring, and to what extent,’ he says. Drawing on 311 data, inspection reports, census tract information, and other records, the study found that neighbourhoods with high rents, higher incomes, better educated residents, and larger non-Hispanic white populations ‘tend to over-report.’ ‘Based on these results, we find that socioeconomic status, householder characteristics and language proficiency have a non-trivial effect on the propensity to use 311 across the city.’
Still other analysts have mined 311 data sets to show how they correlate to broader trends, such as the spread of urban blight. Those patterns, according to a 2016 analysis by NYU and the Center for Urban Science and Progress, could theoretically be used to predict future real estate prices.
For planners, mining 311 call records holds out the potential to forecast service demand and also correct for biases that might lead city officials to be more attentive to complaints from more affluent or vocal communities. In 2017, a team of geographers and artificial intelligence scholars at the University of Illinois Urbana-Champaign used six years of Chicago 311 sanitation service requests (e.g., overflowing garbage cans) to develop what they said was the first algorithm capable of generating accurate predictions to help guide decisions about scheduling and routes.
Yet Kontokosta, in a 2018 study on biases in Kansas City’s 311 service entitled ‘Who Calls for Help?,’ offers up a caution, given findings that some individuals and communities – women, middle-income families, homeowners, parents whose kids attend private schools, households with internet access – were far more likely to call 311 than other demographic groups, as was the case in New York. ‘As such, training predictive city service delivery models on these data would lead to an inequitable distribution of service provision, leading to over-allocation of resources to households and neighborhoods that are more likely to report problems’ (Kontokosta & Hong 2018).
Open Data
Since the early 2010s, most city governments have taken to routinely releasing certain types of municipal information through open data portals – websites that allow users to download, for free, ‘machine readable’ databases (i.e., formatted so they can be queried with readily available software) that municipal officials have made public. The information ranges from registered pet names to air quality readings, overnight shelter usage, and the locations of urban objects, from signals and crosswalks to park benches. New York City’s huge open data portal even includes a Central Park squirrel census. These data sets are updated regularly. The contents are subject to privacy laws to ensure that no personally identifiable information is released.
Early on, the open data movement was regarded as a cause célèbre among digital open government advocates, who saw it as a way of unlocking public information tucked away in municipal servers and protected by bureaucracies. Cities hosted hack-a-thons for coders and app developers who would figure out how to use this treasure trove of data and create digital services for city dwellers. Some local governments made it a practice to aggressively release new data sets (New York City passed a law in 2012 mandating the disclosure of all municipal data from all departments by 2018) while others did it grudgingly or with little enthusiasm.
Some applications bobbed to the surface – e.g., a transit route app that maps the real-time movement of transit vehicles and can inform users when the next one will arrive. In other cases, the municipal data became the foundation of business ventures. Activists have marshalled large sets of data, often presented in visualizations, to advocate for non-commercial goals ranging from public space improvements to changes in police check practices.
Among the early users of open data portals, in fact, were municipal officials, who could finally gain access to operationally relevant but previously inaccessible information from other departments. For instance, databases of records of citizen complaints coming in to 311 call centres have been used to make service improvements, while data on taxi movements is used to assist in transportation planning.
In recent years, London, New York, and other cities have begun retooling their open data strategies to respond more quickly to requests and allow for the release of streams of live data.
Some players, in turn, have begun to ponder the monetary value of all this publicly generated information and whether municipalities should be giving it away. Miovision’s Kurtis McBride says that companies like his use public data streams – e.g, traffic counts – and transform them into a profitable business model. ‘The more data I have, the more the data is worth,’ he said last year during a public consultation session entitled ‘Realizing the Value of Data.’ ‘The public sector needs to think about whether open data is a lost opportunity.’
Municipal Dashboards
Since they first began to appear in the late 2000s, urban ‘dashboards’ have become not just de rigueur for cities aiming to position themselves as smart and evidence-driven but also, in the words of one analyst, ‘an object of desire’ for municipal politicians. On their face, city dashboards serve up a buffet of indicators about all aspects of urban life, from traffic and jobs to investment, housing, safety, and so on. The data on dashboards appears to be constantly updated, and drawn from sources like development applications, economic indicators, crime stats, and even through water bills or transit fares.9
While dashboards are publicly accessible and thus available to anyone, the users tend to be civil servants, local politicians, media, businesses looking to invest in a city, and other specialized audiences. Produced typically by outside consultants working with teams of municipal officials, dashboards can be seen to contribute to a city’s public image in the way that international city rankings do – a quantifiable and apparently objective snapshot of a particular city at a moment in time, the perfect tool in a world where the phrase ‘evidence-based decision-making’ has become an indispensable disclaimer for a wide range of political choices.
Yet in a 2021 essay in New Media & Society, Jathan Sadowski, a Melbourne-based social scientist who studies smart urbanism at Monash University, exposed the backstory of one Australian city dashboard and found it to contain a lot more political messaging and image-making than met the eye.
Sadowski had spent two years working on a revised dashboard for Parramatta, a booming suburb of Sydney that wanted to promote itself as a smart city. The old dashboard, he explained, had ‘been left to die.’
Its primary purpose, Sadowski soon discovered, was to provide the city’s chief executive officer with ammunition for holding lower-level managers to account. ‘People resented being browbeaten with KPIS,’ he noted. ‘Thus, by extension, their ire was directed at the corporate dashboard; it became a despised tool of control.’ A new CEO eventually took over and didn’t care about the dashboard’s KPIS, which meant the lower-level managers in charge of providing the data simply stopped updating it.
Parramatta officials decided to start from scratch and assembled a working committee of local politicians, municipal managers, and data visualization consultants. A mock-up version with fake data was eventually presented to the committee, whose members, Sadowski notes, became preoccupied with the placeholder data – evidence, he observes, of the power of these visualizations. ‘The dashboards are meant to represent reality, but they can influence perception so much that they bend reality,’ he explained.
As Sadowski concluded: ‘These are not the stories of technological development that we usually hear. Stories of failure and frustration, of delays and dead ends, are extraordinarily typical features of the work done in government and technology. But these stories run counter to the narratives of innovation meant to sell smartness’ (Sadowski 2021).
Data Governance
In a light-filled event space in Toronto’s historic Corktown neighbourhood, dozens of reporters and Sidewalk Labs officials gathered one sunny morning in June 2019 for the long-awaited release of the company’s so-called Master Innovation and Development Plan (MIDP). Monitors were situated around the edges of the room, and Sidewalk’s brash CEO, Dan Doctoroff, was holding forth at the front.
Members of the media had been given hard copies of the MIDP – a four-volume, richly illustrated box set of documents weighing in at about ten kilos, with ‘Toronto Tomorrow’ emblazoned in block letters on the side. Those volumes – also available online – seemed to cover everything one could possibly want to know about Sidewalk’s plans, including copious details about architecture, environmental features, and technology, as well as appendices laying out the fine print: how this place would be managed, who would be in charge, what legal exemptions were required, and so on.
By the time the reporters gathered to watch the presentation, many had already been contacted by the media relations officials with Waterfront Toronto, which had brought Sidewalk to town two years earlier, to give them a heads-up that the agency’s chair would be rebuffing key elements of the plan.
The most theoretical and legally opaque details involved the governance of this new community – Sidewalk had made it clear that it wouldn’t be the City of Toronto – and the handling of all the data that would be collected within its 12 acres. According to the MIDP, that information would become the responsibility of something it called an ‘Urban Data Trust.’ The trust, the company stated, would be tasked with storing and protecting the data collected within Quayside, ensuring privacy and monetizing the IP created for this futuristic neighbourhood. The trust would also devise data-sharing standards that would be applicable for other smart cities – a hint that information generated in Toronto might not stay in Toronto. But in the spirit of establishing a mutually beneficial partnership, Sidewalk said the city would get a cut of the action.
Sidewalk Labs pledged that this trust would respect Canadian privacy laws, but critics had a laundry list of questions about how the trust would work, its legal obligations, and its ability to meaningfully guarantee privacy while ensuring that people living, working, and visiting Quayside would consent to sharing their information.
According to a critique by Anna Artyushin, ‘the term urban data was assigned a new meaning: information “gathered in the city’s physical environment, including the public realm, publicly accessible spaces, and even some private buildings.” The document categorized urban data into four categories: personal information, de-identified data, aggregate data, and non-personal data, with each and every one to be managed by the trust.’ She also debunked Sidewalk’s claims about the legality of this scheme: ‘Of the four types, only personal information is subject to Canada’s legislation.’
Other critics questioned the idea that a private entity, like a trust, should be put in charge of information gathered in the public realm. ‘Promises to self-regulate must be viewed with skepticism especially because of the way technology companies have expanded their data collection and use practices,’ wrote Rutgers University law professor Ellen Goodman in a 2019 review of Sidewalk’s plan commissioned by the Canadian Civil Liberties Association as part a lawsuit seeking to block approval.
Goodman co-founded the Rutgers Institute for Information Policy and Law and studies issues such as the ethics of artificial intelligence in smart cities. As she states in her brief, Sidewalk Lab’s digital governance plans contained promises about privacy and responsible use. But, she argued, the gaps in Sidewalk Labs’ proposal, coupled with Alphabet/Google’s history of misusing personal information, raised warning flags.
She cited some technologies envisioned for Quayside – apartment energy schedulers, a system for tracking each apartment’s waste to generate ‘pay-as-you-throw’ bills, and self-driving parcel delivery robots operating in tunnels beneath the buildings. All would gather personal data (i.e., name, address) that could theoretically be used to figure out when, for example, someone was home, what might be in their waste stream, and insights about their consumption habits. The systems, embedded in the design of the Quayside buildings, relied on algorithms fed by data generated by residents themselves. ‘These particular algorithmic regulatory systems are likely to be at the core of the Quayside infrastructure, influencing how the built environment is arranged and functions,’ Goodman cautioned. ‘Once they are in place, it may be difficult to unwind the data flows.’
Sidewalk’s proposed digital governance leaned heavily on anonymizing personal data before it is shared. But Goodman noted the growing research that shows how privacy violations – ‘re-identification’ – can occur despite such steps. Transportation information – tap-on/tap-off transit cards or trip data from ride hailing services – can be cross-referenced with other public sources of information to generate inferences based on a user’s behaviour (e.g., regular trips to a health clinic). Citing privacy violations by Apple, Amazon, and YouTube (Google), Goodman concluded: ‘There are too many examples of technology companies promising to anonymize personal information, but then compromising that anonymity, to rely on assurances of de-identification.’
Smart city technologies are complex, but technical difficulty isn’t necessarily an impediment. Natasha Tusikov cites earlier examples of public governance of highly technical systems, such as the Canadian Radio-Television and Telecommunications Commission or the Atomic Energy Control Board. ‘It seems that after all these scandals with tech companies, people have reached a point where there’s a role for government,’ she says.
Sidewalk’s decision to pull the plug on Quayside shortly after the World Health Organization declared an international state of emergency in the early months of the COVID-19 pandemic rendered moot the abstruse debate about the urban data trust. Still, says Anna Artyushin, there are lessons to be learned. ‘While Sidewalk Labs’ plans for data collection were very much in line with the privatization of urban governance and the normalization of ubiquitous surveillance that takes place in smart cities around the world, the idea to put data in a public trust was rather novel. Using trusts to give the public a share in profits and an oversight role in the governance of personal data in smart cities may seem like a viable solution, but there are some significant limitations to this approach,’ she notes. The story, she concludes, ‘is a cautionary tale.’
9. Toronto’s Dashboard, which lives on the City of Toronto website, displays data on sixty-five key performance indicators in six categories: community vulnerability, crime, development and construction, economy, revenue, and services. Each KPI is shown in its own colour-coded box – red, yellow, and green – with the latest year-to-date information, an arrow indicating the trend line, and the percentage change from the previous year. Users can also click on individual boxes to get even more granular information, such as historic data, charts, and explanatory notes.