Big Data and the City
One of the central themes in our first set of discussions was the role advanced technologies and big data would play in accelerating the mutation and evolution of the city.
The idea is simple enough. Using high-quality data to measure the positive or negative effects of changes in design or regulation should ideally allow us to learn what works, and to act accordingly. Given enough experiments on the target population, planners should be able to build a more inclusive future based on the needs and revealed preferences of those that will be affected. The success of this trial and error approach relies on, and indeed assumes, that the objective function that we are optimizing for, remains constant. Only then is it possible to compare different outcomes and assess if one meets the defined objective better than the other.
But as we have seen, these objectives, as well-intentioned as they might be, are not infallible. Even if intended goals are achieved, a number of unintended negative consequences can spring up unexpectedly and existing biases can be perpetuated. A good example of this was Amazon’s recent attempt to roll out single day delivery in Boston. In order to remain cost effective, they intentionally targeted only the areas where they expected to make a profit given the volume and frequency of existing orders. But the move ended up excluding a number of already socio-economically marginalized residential areas. The goal of providing one-day delivery to the maximum number of customers in the city was met, but the further exacerbation of unequal access to private and public resources was an accidental corollary.
To that end, the key elements that I believe are missing from the present discussion on the place of data in defining the future of our cities are focusing on asking the right questions and a comprehensive definition of data transparency.
Given the potential impact of policies in this space, and the difficulty in knowing a priori what consequences there might be, the focus of our trial and error should not be on what approach generates the best results for a given question, but on the arriving at the right social objective. The question we pose of the data we have should itself be as mutable as the experiments we intend on running.
In addition, the role of transparency in ensuring that outcomes are inclusive cannot be understated. Although a lot of data used for public policy is openly available, the methods and algorithms used to build models and make decisions are often not. This line is further blurred by corporations with privately collected data, such as Uber and Facebook, able to run “experiments” at scale that have significant impacts on the lives of citizens. We need to extend our definition of transparency to apply not only to source data but also what models are being used in the public domain. This will require reversing a number of embedded legal and philosophical conceptions on what the ownership of data means, and how much of it can justifiably be withheld in the name of protecting competitive advantage.
As with all industries, the increasing role of data and technology in urban policy and design is inevitable. But for there to be sustainable positive outcomes we will need all involved stakeholders to be responsible stewards of our collective future.
