Urban data is used to trace issues even before they occur. Sounds comforting? That’s unless authorities are going to intervene on our predicted behavior.
There’s a long history of data being generated from citizens to understand city life, formulate policies, solve urban problems, guide operational governance, model possible futures, and tackle a diverse set of other issues. Consequently, the way cities are understood and managed has been data-informed for centuries. More recently however, there’s a change in the way cities are governed. As urban data creates new forms of data-driven steering, it helps to produce what’s been called smart cities. This is accompanied however by a threatening opportunity. With the potential to model possible futures, authorities will be able to judge and act on predicted behavior as well.
“If you want to keep a secret, you must also hide it from yourself” — George Orwell
Urban data constitutes a broad class of information that’s generated continuously and exhaustively from people and places. Such data is produced by networked sensors, transponders, cameras and other software-enabled devices around the city, the smart phones and apps that people use and interactions across networked systems. It’s a diverse range of public and private bodies producing that data, including utility companies, transport providers, environmental agencies, mobile phone operators, financial institutions, retail chains, security firms, and emergency services.
In almost all areas — health care, employment, public order, safety and public transport — city governance is increasingly underpinned by urban data. And the reasons are obvious. Through the smart use of data, cities can do their job more effective and less costly, with the ability to model possible futures and trace problems even before they occur. Besides, there’s a great amount of data readily available. Not only produced by public and private bodies, but also directly fed by social media — where almost every citizen exposes private information, like their interests on Facebook, work and education on LinkedIn, and their opinion on Twitter.
The flow of actionable data feeds into cities’ management systems and control rooms, set up to monitor and manage urban tasks and operational governance in real-time. This provides for imagining, simulating and predicting urban futures through big data analytics.
Urban Data Applications
Tangible examples are already out there. Public, real-time, urban data allows you to see what’s happening on the streets of Dublin, London, Rio de Janeiro and Amsterdam. While these dashboards seem to provide a simple, yet powerful, new data-driven way to know and govern cities, there are more ways in which urban data is put to use.
Utrecht is an example of a city that made great steps in data-driven management. Martin Jansen, their Data-Driven Management lead, says sufficient governmental support, a dedicated budget and pilots throughout various departments are key to its success. A few examples of the city’s new possibilities include:
- Predicting the maintenance of sewage pumping stations so maintenance can be performed promptly and proactively.
- Home burglary predictor.
- Applying 3D simulations to obtain better grip and insight into crowd management at large events.
Other Dutch cities are calculating people’s vulnerability to loneliness and depression, based on income records, unemployment rates, number of single-parent families, school drop-outs and buying power. Streets labeled with ‘increased risk’ may expect a counselor’s visit to discuss possibilities and provide help. The people visited completely unaware they were computer-picked; the counselor not knowing how the addresses made it to his list.
Ger Baron, responsible for digital technology experiments as the Chief Technology Officer of Amsterdam, emphasizes the possibilities with urban data. When real-time data is combined with historical data it gives the capability to predict what will happen, which in turn generates the opportunity to make choices: “What we can do right now is limited, but when we are able to predict accurately, we can also start to do interventions on the fronted.”
An exciting, yet thrilling thought, that needs to address at least a few matters before settling in.
What About Privacy?
Privacy is a sensitive political matter and the use of citizen’s data raises questions. In this context, privacy means: making sure individual data is used anonymously. As such, computer findings should only be applied on group level — highlighting streets or districts instead of individuals. But together with the increasing governmental authority to collect and use data, a new window of opportunities opens.
Baron doubts there’s any cause for concern. His city experimented with digital crowd control during the ‘Sail’ event. That means cameras visualizing flows of pedestrians, Wi-Fi sensors gauging passing mobile phones and social media teams measuring how visitors felt. At the same time, a campaign — large billboard throughout the city — promoted awareness, inviting people to reach out in case of questions or concerns. Baron: “We received no calls, zero.”
There’s no great deal of commotion or criticism on SyRI either. This computer model — short for System Risk Indication and not to be confused with Apple’s Siri — is used by the Ministry of Social Affairs in the Netherlands to identify potential tax fraud. Educational records, permits, welfare stats and health insurances are just a few examples of what SyRI is able to analyze for potential prosecution. But as the Ministry makes no mention of SyRI investigations because “it would undermine the methods and procedures of enforcement”, governmental concerns arise.
What About Governance?
From a political and governmental perspective, urban data is often critiqued for enacting and promoting a form of technocratic governance and expressing a particular set of normative values.
Traditionally, authorities respond to citizens’ actions — if you do something, they react — either with a grant, a penalty or something else. However, with the goal of analyzing urban data to predict behavior, authorities are able to intervene before citizens act. And that’s where the shoe starts to pinch. It’s no longer what a citizen does that matters, but it’s about predicting what he intends to do.
That way urban issues are treated as technical problems that can be solved through technical approaches. Such approaches are based on the premise that cities are largely mechanical, rational, linear and hierarchical systems and can be disassembled into neatly defined problems that can be fixed or optimized through computation. It promotes a technocratic, top-down form of governance in which city managers steer and control the city through a set of data levers. It ignores the fact that data driven steering is full of normative assumptions and effects — this includes notions about what should be measured and for what reasons, with the information produced being used to influence decision-making and modify institutional behavior.
What About Transparency?
Data is often processed and analysed by black-box algorithms, lacking a desired level of transparency.
Computers generally measure correlation — the occurrence of two or more situations at the same time — not causality. If that happens enough times, the computer assumes there’s a connection. Why there’s a connection, and how it’s established, is of no interest to the computer. Humans on the other hand need those explanations to understand what’s going on. In other words, if someone asks ‘why do you think my daughter is going to be a school drop-out?’, it’s hard to give a different answer than ‘because the computer said so’.
There’s an important organizational issue in the way of citizens being able to see which data is collected and what for. It’s simply hard for cities to share an overview of all their data. Most of the times a central register is not available, and different departments don’t even know what others are doing. Critics note that even if urban data is accessible to the public, it’s not neutral, reductive, and puts complex relationships into simple measures. Visualizations are repeatedly open to ecological fallacies that significantly shape interpretation — the same data displayed at different scales can show considerably different patterns.
What About Errors?
Computer based behavioral predictions are often considered untouchable through their appearance of impartiality. Nonetheless, from a technical perspective there are numerous concerns with the data itself and the analytics used. Much data is plagued with issues of accuracy and fidelity, and is often published with little or no metadata that enables analysts to judge data quality. Besides, people’s preferences do not always match their real life decisions. I mean, is your Facebook profile a real-time reflection of your life? Your past (data) doesn’t guarantee your future actions (predictions).
There are examples of citizens being falsely reported to the Immigration Office, as an algorithm identified them as potential illegal immigrants. What the algorithm didn’t take into account was the freedom of travel for EU citizens — meaning citizens of every EU member state are free to travel to all other member states. A computer code checks one law, but fails to include the legal framework concerned.
Rectifying errors in algorithms is proven to be hard. The only way citizens are able to reverse decisions, is to step up to the Court of Justice. Problem is, the algorithms behind these decision are hard to read, and therefore it’s difficult to judge their legitimacy. Only a handful of people know how to operate algorithms and understand its visualizations. The more sets of data a computer model uses, the harder it is to understand its calculations. That being said, a model is performing better with more variables — often dozens, sometimes even hundreds — and a human is then soon lost.
Urban data helps understand city life, formulate policies, solve urban problems, guide operational governance, model possible futures, and tackle a diverse set of other issues. It undoubtedly provides a wealth of timely, longitudinal and citywide information, but it’s also partial, framed, normative, and enact a particular form of technocratic governance. There are substantial transparency- and technical shortcomings that should be recognized, advocating data is used contextually and in conjunction with other forms of urban knowing and governance. Authorities intervening on predicted behavior is a dangerous game to play.
It’s important to remember that the data generated does not exist solely of the ideas, instruments, practices, contexts, knowledge and systems used to generate, process, analyse and interpret them. It’s not objective and neutral, but shaped by a wider social perspective and invested with the values of those that create and analyze them. Focusing narrowly on framed factual information de-contextualize a city from its longer historical context, politics, governance and culture.
Cities are fluid, open, complex and multi-level systems, full of competing interests and wicked problems that are not well captured in dashboards and are often best tackled through political and social solutions, policy interventions, and citizen-centered deliberative democracy. The authorities should try and find a balance between utility and limitations.