Boston Mayor Marty Walsh will be inaugurating the ultimate Big Data scorecard on city services, a single number reflecting how well city government is doing its job. The program, known as CityScore, was previewed in the New York Times. All of Boston’s services represented in a single number! Here’s my favorite quote from Mayor Walsh’s chief of staff Daniel Koh: “We need to figure out the best way to weight these things, and there will be some that will be somewhat political.”

Well sure it’s political; of course it’s political. But then numbers always are in politics.

That is one of the lessons of an invaluable book on public policy, Deborah Stone’s Policy Paradox: The Art of Political Decision Making. In politics numbers are never the endgame; they are a tool in political conflict. Numbers can present an argument for how well or poorly we’re doing, whether things are getting better or worse and for how long, signs of prosperity or of decay. Numbers give us a focal point to fight over, a device to assign our preferred causal interpretation, a way to push for our favored solution. Sometimes the same number is a sign of both progress and regress, depending on who is doing the interpreting. We have to decide what we are going to count, how to define it, and how to count it. Stone says that “Numbers, in fact, work exactly like metaphors. To categorize in counting or to analogize in metaphors is to select one feature of something, assert a likeness on the basis of that feature, and ignore all the other issues.”

Let’s measure our educational system. How is your child’s MCAS score? How is your school’s MCAS profile, improving or declining? Whoops, we are on to PARCC. Then again, the U.S. Secretary of Education said the other day that perhaps we really are testing too much. One complaint is that teachers “teach to the test” rather than inspiring children to love learning. Of course if you are going to measure teachers’ performance based on standardized tests, it is inevitable that they will teach to the test. Love of learning is wonderful, but Big Data does not measure love.

The Times reports that the commissioner of public works has detailed statistics on potholes. The city relies on citizens reporting potholes. Suppose one neighborhood supports many immigrant families (with limited access to technology) from countries where distrust of government runs high; it has lots of potholes but doesn’t report many of them. Another neighborhood, which is middle class and well-connected (politically and technologically), has multiple reports at the first sign of loose gravel. The multiple reports are good because then we fill one pothole and close 2-3 complaints. Our pothole metric is going great.

Or, another goal is fair representation on the city council. What is “fair”? The decennial redistricting is one of the most political decisions the city council makes. Usually there is litigation. Last redistricting the council passed a plan it deemed fair and Mayor Menino vetoed it. The council passed a different plan, the mayor signed and no lawsuit this time.

To return to Mr. Koh’s New York Times quote, he was referring to how to weight different services in the algorithm: 911 response time vs. tree trimming vs. schools (MCAS, PARCC?) vs. library books loaned vs. liquor licenses granted, vs. EMS service vs. animal control, and on and on and on. Just that exercise ought to impress us all with how much city government does. But how to weight each statistic into the final CityScore is an enormous task. And a political one.