Digital tracking systems – The smart city ‘is a promise to improve quality of life, even in Dutch cities. But recent studies show considerable pitfalls. Will an algorithm determine which area resident is a danger?
Hudson Yards is still the biggest construction site in Manhattan. The first skeletons of the skyscrapers look at first glance no different than any other in New York. But here no ordinary district is being built.
“We deal with sensors that measure air quality on every corner,” said Constantine Konto Kosta, one of the directors of the project. “We are also mapping temperature, light levels and constant noise.” He also wants to combine the data with data from municipal hotlines, social services, and possibly energy companies.
“We’re going to track messages on social media of residents so that we can measure their sentiment, and location data from mobile phone tracked via wifi networks so that we can see how they move through the neighborhood.”
Starting next year, Hudson Yards will be the first quantified community of the world: a neighbourhood in which everything and everyone at any time of the day is measured. “For the first time we can analyse a district in real time.” With the data, the municipality for example can provide rapid intervention where there is noise pollution.
In the long run, he also wants to collect information about exercise, health and weight of the neighbourhood residents, for example with data from wearables like wristbands. “Entirely voluntary of course.” And all data is stored anonymously. Hudson Yards is one of the most expensive areas of the city; the developers expect that there is a lot of demand for such a quantified life.
From our partners:
Around the world, citizens are digitally mapped, often under the name ‘ smart city ‘. Rotterdam is coming this fall with a ‘ smart city-approach ‘. Amsterdam has since last year a chief technology officer (CTO) – that does all kinds of experiments; like placing sensors for measurements in bicycle parking to the automated monitoring of crowds at events. Also, the municipality of Midden-Drenthe has its own smart city-app.
“The promises are great: the smart city can be very rewarding,” said Jorrit de Jong, academic director of the government innovation program at Harvard University. There is much more information available on which municipalities can coordinate their policies. “But there are also potential problems. That is not always well thought out. “
These are the four potential problems:
1. The smart city is hacked
Parts of smart cities are already hacked regularly: there are plenty of examples of smart street lights that was switched off remotely by unauthorised persons and traffic lights that were manipulated by hackers. Also, power systems and even lock gates are often broken into. The more sensors, infrastructure and operating systems have an Internet connection, the more vulnerable. “Not all small municipalities have the ability to control security as well,” said De Jong.
And besides the physical systems, the data needs to be protected. Getting data from smart energy meters can be used to deduce when someone is at home. Useful for burglars. Health data is a particularly sensitive subject, because they may be worth a lot for insurers. “The security and privacy protection of those data is absolutely essential,” says Konto Kosta. But he acknowledges that no system is 100 percent waterproof.
What’s important in addition to security: Who owns the data? It is technically complicated to separate data about different individuals. For now, the municipalities and the owners of the sensors possess the data.
2. The algorithms are uncontrollable
To find patterns in the mountain of data, algorithms are needed. Which search the data based on a fixed set of instructions. The instructions that have been programmed in such an algorithm, determine the outcome. But what if those instructions are ethically unacceptable? There are already many examples.
De Jong is not only researcher but also a director of the Harvard Project in three US cities. His team is using data to predict buildings where crimes may arise. How? By comparing data on those properties with buildings that have been derailed in the past, often to predict problems. “One of the predictive factors for example is that an inmate can not pay taxes.”
Other data that have predictive value include data about the noise level in the neighborhood, police data, and other personal information about residents.
The algorithm sends decisions that can have a major impact on the residents. De Jong: “It is very important that such an algorithm is valid.” Especially because potential factors such as ethnicity, age, or income level of the residents can play a role. “Then you come soon, for example, at ethnic profiling and other unconstitutional matters.”
Ger Baron, CTO of Amsterdam, recognises this. He develops a system that aid workers based on large amounts of data to help assess how safe certain neighbourhoods are. That is because the church firefighters and ambulance workers are sometimes attacked. The system uses color codes: If people live in a house or neighborhood that fit the pattern of people with violent behavior, for example, the code jumps on red and extra security personnel.
How do you determine the encryption algorithm that determines which remains neutral and not, for example, unjustifiably turns red at homes only because immigrants or foreigners live? “It could just be that the color code is often negative in neighborhoods with low incomes for example,” says Baron. It all depends as to what criteria the algorithm decided.
“Politicians and officials must find out more: where is that algorithm based on,” says De Jong. Politicians also read bills before they approve it; They must now increasingly understand the functional design of software applications and understand the algorithms. Drivers often do not have the technical knowledge to properly evaluate the technique, according to De Jong. “That need not be bad, but they need to check information thoroughly just as they do to bills.”
3. Companies have too much influence
Amsterdam and Eindhoven advertise both as urban living lab: an urban laboratory for new technologies. That sounds like the language of technology companies. And it is. In Eindhoven the French company Atos pays large-scale experiments. GER Baron of Amsterdam is, in his own words approached”daily” by companies who want to try out new technologies on locals. He rejects most.
Tech Companies like Samsung, Microsoft, IBM and Alphabet (formerly Google) have discovered the smart city and are closely involved in many projects, including in the Netherlands. In itself, it is not strange that they should join in, because those companies are also the innovators.
“But the interests of companies are not always the interests of citizens,” said Baron. They sometimes have different views on the ownership of the data, or the transparency of projects. For the local governments, openness and accountability is important, but companies sometimes want more secrecy to competitors.
The more smart city projects municipalities are going to do, the greater the risk that improper officials be swayed by technology companies. De Jong:
[infobox]“Partnering with businesses is important and useful for innovation. But the municipality should never lose control.”
[/infobox]
4. The government screwed up IT
Smart city projects in the core are just IT projects. IT and government are not always a fine combination; million-dollar projects often fail.
De Jong of Harvard conducted research on what determines whether a smart city project is a success or not. It shows that three factors are crucial. The first factor is the ability of municipalities to work together both internally and externally. This often proves difficult.
According to Konto Kosta, “Shared access to data can cause problems: for example, you rarely see that the agency for water talks well with the agency that collects energy data. And the combination of administrative bureaucracy with fast technology is not always good .”
The second factor that the study shows, is the ability of municipalities to set clear goals and to match them to measurable criteria.
Thirdly, a municipality should be able to analyze all the data. One solution is to hire more data scientists, which are now widely popular – even at companies that have deeper pockets than municipalities. “Municipalities will have to compete harder for those people,” says De Jong.
Telecom company KPN is launching this fall a special network in The Hague and Rotterdam that is intended to connect various sensors together.
Ger Baron also has a lot more smart city plans with Amsterdam in the short term, and Konto Kosta is not going to stop at Hudson Yards. He is with his team already working on other quantified community projects: in a poorer neighbourhood of Brooklyn and in the financial district of Manhattan.
This feature originally appeared in NRC.