I suggest that anybody who is talking about ‘big data’ and ‘data driven policy’ or using grotesque statements like data is the new oil to revisit the foundations of scientific research and the embedded vision on data.
Without elementary insight in the way scientists arrive at their conclusions ‘data driven policy’ can have disastrous consequences. The city of Chattanooga has build a digital twin. That is a digital model that is connected to reality with the help of sensors. Such a dynamic model can be used for simulation purposes if the connections between the variables have been established. Here things can go wrong. In Chattanooga the model was used to simulate the impact of flexible lane assignment and traffic light phasing. It turned out that this could result in a 30% decrease of congestion.
Had this experiment been carried out in the real world, the result would probably have been disastrous. Traffic experts note time and again that every newly opened road gets satiated after a short time, while the traffic on other roads hardly decreases. In econometrics this phenomenon is called induced demand. In a study of urban traffic patterns between 1983 and 2003, economists Gilles Duranton and Matthew Turner found that car use increases proportionally with the growth of road capacity: Every road user reacts differently to the opening or closing of a road. Those reactions can be to move the ride to another time, to use a different road, to ride with someone else, to use public transport or to cancel the ride. To understand this pattern data must be collected from e sufficient large sample of road behavior of individual drivers.
What the computer scientist in Chattanooga did wrong is assuming that only the adding of a single lane and changing the intervals of the traffic lights would cause all drivers’ behavior change into the same direction, as if they were metal balls, reacting upon a change in the magnetic field. If the ICT-experts had collaborated with traffic experts, the digital twin might have been fed with an empirical justifiable model, that incorporates the assumption of induced demand.
In essence, data is useless without a theory, based on already established insights or views.
This post based on by the new e-book Better cities, the contribution of digital technology. Interested? Download the book here for free (90 pages)
Content:
Hardcore: Technology-centered approaches
1. Ten years of smart city technology marketing
2. Scare off the monster behind the curtain: Big Tech’s monopoly
Towards a humancentric approach
3. A smart city, this is how you do it
4. Digital social innovation: For the social good
Misunderstanding the use of data
5. Digital twins
6. Artificial intelligence
Embedding digitization in urban policy
7. The steps to urban governance
8. Guidelines for a responsible digitization policy
9. A closer look at the digitization agenda of Amsterdam
10. Forging beneficial cooperation with technology companies
Applications
11. Government: How digital tools help residents regaining power?
12. Mobility: Will MaaS reduce the use of cars?
13. Energy: Smart grids – where social and digital innovation meet
14. Healthcare: Opportunities and risks of digitization
Wrapping up: Better cities and technology
15. Two 100 city missions: India and Europe
Epilogue: Beyond the Smart City
Science for Society 