Smarter Cities: New Datasets Generated by City Governments and the Nerds Who Love Them

In case I win the Nobel Prize tomorrow, I wanted to make a couple of points because I could be real busy the next couple of weeks. In today's NY Times, Steve Lohr has a nice piece that will excite Supercrunchers. Cities are sitting on data goldmines. If we can overcome privacy concerns then academics and other nerds could do some really interesting research. The GIS mapping crew can map until the end of time but we could be more ambitious and make some progress on some causal questions.

Mayor Bloomberg, who knows something about the value of information, is leading the charge here. Will Mayors of smaller cities follow the leader? Suppose that the Mayor makes available the electricity bill of every commercial building in the city. Energy efficiency businesses could contact the buildings whose square foot consumption is high to see if these buildings are being "wasteful" or whether the activities taking place there (plasma TV?) just use a lot of power. The "green jobs" weatherizers should not be throwing darts at a map in choosing who to target. They should focus on high electricity consumers but do these consumers know that they are "high"? Equal easy access to information would make my answer "yes".

Lists of restaurants that have poisoned people recently will lead to improved sanitation and public health (see Phil Leslie's work). This is just the tip of the iceberg. Information is a public good and government does have a role here in providing it. I have been told that Milton Friedman opposed this role but I have never understood his view here.

Let the sunshine in and the nerds will run with it. Mayors will be accountable, the nerds will get tenure and the blogs will have a lot of funny things to talk about as the statisticians uncover some funny and policy relevant facts.

Now in terms of causal research designs, smart guys will have to solve initial conditions issues. Ideally treatments will be randomly assigned or we will need a model of the initial conditions to see how various treatments (such as introducing a Charter school) affect the spatial distribution of kid's test scores around the charter school and in nearby public schools.

But the dependent variable and easy access to it is a nice first step.