In June 2015, I published “Big Data In the Big Apple” a report on how London could learn from New York City’s Mayor’s Office of Data Analytics (MODA). MODA’s founding director, Mike Flowers (now based at Enigma Technologies), recently visited Policy Exchange to record an interview about his experiences of applying data analytics to urban environments. Below is an abridged transcript of that interview.
EC: Mike Flowers, welcome back to Policy Exchange. You’ve been here in London talking about the work that you did in New York City where you were the Founding Director of the Mayor’s Office of Data Analytics under Mayor Bloomberg. How did that role come about?
MF: Well, in 2009 the city’s population was continuing to grow. At the same time our budgetary resources were not increasing, so Mayor Bloomberg was basically pursuing any approach that would allow us to wring more efficiency and effectiveness from city service delivery mechanisms, and one of those assets that he believed was under-leveraged was our data.
EC: Were the approaches that he brought to cities very similar to what he’d been doing in the corporate world?
MF: Yes, absolutely. If you look at the Bloomberg Terminal, it’s simply a synthesis of different information streams for the purposes of making decisions in the financial markets. All we did was… copy that approach for government services.
EC: The model that you pioneered in New York was very much about using existing city data. Can you give an example of an area you applied that approach to?
MF: Certainly… what proved to be one of our most valuable sources of information as a predictor of where we should be allocating our resources was property tax records. Property tax records have been public for decades. They’ve only been really collected for the transactional purpose of recording that somebody actually paid off their tax obligations on a piece of real estate, but… that same information – whether you paid or not – proved quite valuable in terms of predicting where a fire might occur or other bad things that would require emergency response.
EC: What was the process of taking that data and using it to predict something useful?
MF: The actual process itself was… quite rudimentary and straightforward. It was simply a matter of putting it into a csv in Microsoft Excel as a data point among many data points that we looked at when cross-tabbed with a particular outcome we were trying to prevent or incentivise. So that’s a long-winded way of saying we just put it all in one place and took a look at its relevance.
EC: We often hear about ‘smart cities’ that are investing heavily in big, new technology to be part of a smart urban future; you’re talking about a relatively low-tech approach. Was that a deliberate choice?
MF: Leaving aside the term “smart cities” and what that even means… it was deliberate but it was a deliberate choice that was imposed on us. If I had gone to the Budget Office and asked them for a significant outlay of taxpayer money for a technology that we hadn’t even shown a need for yet, they would have laughed at me, and appropriately so. In fact I would have been happy, as a taxpayer, that they would have refused that kind of request.
So really we took a… “bootstrap method”… What that allowed us to do was a couple of things:
First it allowed us to show that there’s a lot of untapped value in existing [data] assets… Second it also allowed us to actually frame the questions we needed to ask of whatever technology we would ultimately buy. So instead of going in and relying on the vendors to tell us what we needed or didn’t need, we ourselves were able to shape and drive that conversation and make sure that we got the best value for our dollar.
EC: Cities round the world are struggling with establishing the business case for investing in smart initiatives. How did you approach deciding where there was going to be a return on investment?
MF: We took a proof of concept approach that would turn into a pilot, which was all a self-contained experiment… on whether or not it made sense to do things the way we were proposing to do them. And once that proved successful the next step… was to impose that across the enterprise for the entire agency or multiple agencies.
And at that point we had basically proved “Yes, this makes sense, this is something we can do, it’s not going to be disruptive to the organisation, and in fact [it] will provide real value.” And it was under those parameters that we would then go out and allocate funds to do that enterprise expansion.
And that’s when you’re talking about a technology spend. You don’t do a technology spend until you’re ready to really implement it in a way where it’s going to get used right away… It doesn’t mean it’s to the exclusion of technology. It’s just technology should not be driving the conversation.
EC: If it’s not about a big tech solution, what did you have in place in City Hall to make sense of that information?
MF: The most important thing I had in place was the Mayor. None of this ever could have happened without Bloomberg’s backing… Without mayoral support it’s just impossible. So that’s the first thing you need.
And once you have that, then you just have to go out and listen to the agencies on the ground and just ask them: “What are your problems? What are your challenges?” A lot of people think that if they implement key performance indicators as part of a performance management programme for their various cities… that’s the answer, and it’s really not. All those things do is create an external scrutiny on the agency for them to fix whatever problem is perceived to exist, without providing them any assistance.
Whereas the approach we took was going the next step, which is, “OK, we’ve identified this problem or this challenge… here’s how we’re going to go about fixing it. So I’m here to help you with an information product that will help you solve the problem that people are yelling at you to fix.”
EC: So spending time with front-line workers was vital?
MF: That was the primary source of work for my office. The only times we ever did work that was top-down actually ended up being pretty useless, in the sense that we would come up with a solution and then go to the agencies without really understanding that problem… We’d get there and they’d go: “No, that’s not our problem. This is our problem.” So we would toss whatever solution we’d come up with and come up with a solution that actually worked for them.
EC: How did you get over the technical challenge of bringing together different data sets which may not be easy to compare?
MF: We took it on a matter-by-matter basis, rather than trying to come up with a universal ontology [i.e. open standards for data]… More important to me was simply extracting the information that was locked in the silos, getting it into a singularly accessible place, accessible by all, and then, on a project-by-project basis, we would synthesise these ontologies. Most of the time for New York City that just meant synthesising locations, like getting them all on a map in the same place.
EC: I understand there are several different ways of recording location in New York, is that right?
MF: Quite a few! Latitude and longitude, post code, building number, housing number, utility account number. There’s dozens of ways in New York City… of saying “Where am I?”. But the answer to that question invariably involves [asking] “Why do you want to know?” …That was why I felt it didn’t make any sense to spend a lot of iron on hammering out a universal identifier first. …I think rather a problem-driven approach served us quite well.
That said, there is a technology piece to that. [But] you don’t really know what you need from them [the technology providers] until you go in and start answering these questions.
EC: So, not trying to implement open standards for data across the piece, but focusing on specific outcomes that matter?
EC: And MODA had the expertise to do that data crunching in order to compare different datasets?
MF: That was our primary expertise… To be fair, I personally didn’t have that skill set and still don’t, but what I was looking for were data scientists… If you’re any good as a data scientist you have the ability to pull that information together and join it however it needs to be joined for the task in hand. The agencies themselves had the subject matter expertise so they would tell us specifically what they wanted out of it and they would tell us the anthropology behind the dataset so we would do the right interpretation.
EC: I know you’re a fan of the London; you’ve mentioned its similarities to New York. If you were standing in front of the Mayor of London and trying to give some advice on where to start, what would you say?
MF: What I would say, I guess, would be… find a problem that’s really upsetting from a resource allocation standpoint or an efficiency standpoint, to both you as a government actor as well as the citizenry that you are obligated to serve, and fix it.
Whatever that problem is, figure out a way to leverage your data assets to optimise the return on GLA resources and then that will drive an iterative discovery process that will tell you how you want to craft your office. I think waiting to get the data in the right place and in the right way is just an eternal wait. You’re never really going to get fully there, and why would you even want to? So the most important thing… is just to get going, pick a problem; solve it.
EC: And the experience from New York suggests that that doesn’t have to be expensive, doesn’t require big tech solutions, doesn’t necessarily even require a big team. Is that right?
MF: Yes, you’re absolutely correct. No, it did not cost a ton of money. It was the cost of my salary and the salary of a couple of analysts, and we leveraged existing data sources…. Our big data tool – at least at the very beginning – was Microsoft Excel. It’s pretty much the same business tools that are available on pretty much everybody’s desk at this point.
EC: In New York you now have Mayor Bill de Blasio, who has a very different set of priorities to his predecessor. Will MODA endure? How do you see the model evolving?
MF: There was a successor appointed to my position, my old office still exists, the data infrastructure and the analytics infrastructure that we put in place still exists…
The agencies themselves are leveraging the foundation that we laid in the Bloomberg administration… The Parks Department just announced their first Analytics Director as an example. The Department of Health continues to be very robust in terms of data. So it’s scaling and it’s… taking on a life of its own within the agencies, which is exactly what I wanted.
…One of Mayor de Blasio’s first initiatives was the expansion of Universal Pre-Kindergarten, certainly a laudable goal… In order to get there, they had to leverage the information infrastructure that we had pulled together in the Bloomberg administration to roll it out in a way that was successful. So proving my point that it [the MODA model] is policy agnostic, but also proving that once you light the fuse it will take on its own life within the civil service if you execute it correctly.
EC: You’re now based at Enigma Technologies, which provides analytics infrastructure and advice to a whole variety of organisations including cities. What advice would you give UK cities on where they should begin with harnessing their data?
MF: I think they should look at what they’ve already got before they start picking up the phone and calling vendors.
Every locality generates data. It’s impossible at this point not to be generating data because you probably have a utility system that generates water, because you’ve got to bill them somehow.
You have an ambulance service. You have a fire service. You have a police service. You have businesses that are licensed in your town. You’ve got roads that people travel. etc. etc. etc.
You actually have in those a tremendous untapped resource that you can look to to solve the problems that you’re facing in terms of meeting your obligations to deliver those services.
So I’d tell cities the first step is to look inward. The answers really are already there. They just need to decide to take that first step.