 Hi, so I think we're going to get started today. First of all, welcome back and welcome to the new semester. And before I introduce our speaker today, I'd like to give you all kind of an overview of our lineup for the rest of the semester. OK, so we have many talks that address how data and technology mediate urban growth and planning, often reproducing and amplifying the physical and social realities and inequalities exist in the city. So next Tuesday, Jennifer Light will talk about her research on the gamification of city building and virtual worlds and augmented realities. The week after, on February 4th, Michael Batty will speak about his research on the digital representations of physical urban infrastructures and the increasing integration of the digital and the physical. More on digital mediation on March 3, Desiree Fields will talk about her research on digital real estate platforms and on March 31st, Danya Ribas Bell will talk about how to use machine learning and building footprint data to understand gradients of urban boundaries. On April 21st, Catherine DiGnazio and Lauren Klein will talk about their new book, Data Feminism, and how this notion can be better used to address questions of justice through data science and data visualization. OK, so it really wasn't planned this way, but we actually have three talks about the topic of the smart city with Benz today being the first one. Other ones include, on April 14th, Rachel Franklin will talk about algorithmic bias, surveillance, and socioeconomic inclusion aspects of data collection in the city. On April 28th, Xiaowei Wang will talk about how smart city technologies are permeating and transforming peripheral and rural spaces in China and the US. And on the subject of urban development on the periphery, Michael Waldrop will talk about the suburbs of Mexico City and London and the kinds of social priorities revealed through suburban development. That's on March 24th. And then going back to an urban context, on February 11th, Ingrid Gould-Ellen will talk about residential neighborhood stability in the context of gentrification, historic preservation, and rent control. On March 10th, Jacqueline Fong will talk about gentrification and how it influences residential mobility and displacement for disadvantaged residents in Philadelphia. And then finally, during career week in February, we have two panels on urban planning, past and future. On February 18th, there is a panel on the 1989 New York City Planning Commission, which broadened the commission's land use powers. And on February 25th, we'll have a panel of people in the urban tech startup world to discuss the role of private sector innovation in shaping cities. And if you didn't catch all of that, you can find all of these events available on GSAP's website. And with that, I want to welcome Ben Green, who's going to talk about his book, The Smart Enough City. Ben is a PhD candidate in applied math at Harvard, an affiliate at the Berkman Klein Center for Internet and Society at Harvard, and a research fellow at the AI Now Institute at NYU. He studies the social and policy impacts of data science with a focus on algorithmic fairness, municipal governments, and the criminal justice system. His book, The Smart Enough City, Putting Technology and Its Place to Reclaim Our Urban Future, was published in 2019 by MIT Press. So with that, let's welcome Ben Green. This is great. Thank you, Wenfei, for the introduction. And I'm really honored to be able to kick off what sounds like an incredible series this semester that touches on a lot of the similar themes that I explore in my book and hopefully I'll be able to make it back up for some of those later in the semester. So yeah, so I want to talk about my book that came out a couple months ago about the smart cities. And we can keep this informal. If any time you want to interrupt me or have questions, feel free to jump in. And we can talk about whatever is on your mind. But really, the effort of this book is to bring together a couple different areas that I've been working on over the last five or six years at this point. One is my academic work as a PhD student studying the implementation and the development and evaluation of algorithmic systems developed for city governments. I also spent a year as a data scientist in the city of Boston developing a variety of systems and efforts and policies. And I'll talk about some of that work. And analyzing all of that through the lens of scholarship around the social impacts of technology, particular through the lens of a field called Science Technology and Society, or STS, which looks at the ways in which society and technology interact and shape one another and what the social impacts of that are. So I'll roughly talk about three different ideas. First is the notion of technology as a solution to social problems, where I'll introduce the idea of smart cities. I'll talk about the limits of technology for social change, the dangers of viewing technology as a tool for solving social issues. And then I'll talk about how do we actually use technology well? How do we bring together a variety of engineering, social, legal, policy perspectives to understand both the risks and the opportunities presented by technology and actually navigate that sort of minefields responsibly and effectively? So we'll start with smart cities. And I like to start by just providing a definition. It's sort of this broad, loose term. But this is a definition that I like in the sense of giving a framework for what we typically mean when we talk about smart cities. By definition, smart cities are those that integrate information communications technology across three or more functional areas. More simply put, a smart city is one that combines traditional infrastructure, roads, buildings, and so on with technology to enrich the lives of its citizens. And so the smart city really mirrors this way in which the word smart is used across the board today. Of course, the most familiar would be your smartphone. But we have smart homes, smart toasters, smart toothbrushes, this idea of integrating digital technology, internet connectivity, data collection and analysis, artificial intelligence into traditionally analog objects. So what does this look like in practice? There can be a variety of different technologies from sensors on city street poles different parts of city infrastructure that collect data, understand the local conditions, self-driving cars, of course. Various use applications of our machine learning and artificial intelligence that try to take the various types of data that are collected by sensors and administrative functions of city governments to predict where things are likely to occur, to forecast events, to understand really the nature of what's going on within the city. And various technologies for civic engagement, connecting people to one another, providing information. And so the smart city is sort of a loose amalgam of all of these technologies. There's not a single sort of, if you have this technology or that technology, you have a smart city. But it's this rough sense of integrating technologies like these into urban life and urban governance, specifically, very much is a function of often city governments wielding this technology often in collaboration with private companies. And this is not just some sort of pipe dream of a couple of technologies. It really is a pretty strong consensus vision that's come out over the last five to 10 years around what is the future of city government look like. Everything from the federal government to major companies to governments around the world, chapters of mayors offices and local governments, and of course major news outlets talking about the smart city as what the future of cities will look like. Almost every city has some form, at least, and I should say that my talk today in the book really focuses in the context of the United States and North America, don't sort of at this point in time have too much work looking internationally, but I'm happy to talk in Q&A about some of the comparisons and other efforts going on around the world. But just about every city sort of has some sort of smart city effort branding themselves in that way. Kansas City calls itself the world's most connected smart city, San Diego says that it has the world's largest smart city platform. So there's lots of efforts to sort of brand themselves around the smart city bragging about just how smart cities really are in today. So we have these utopian visions of technology coming in to address all of these social problems. But there are a number of questions we need to actually ask of these systems. First, are the technologies actually capable of doing what they promise to do? And if they are capable of doing that, is that even a desirable future? Is the problem that they're trying to solve the right problem and is the solution that these technologies are providing the right solution? And I wanna talk, turn now to why the answer to those questions is no, why the smart city is really a vision full of false promises and hidden dangers and a particularly dangerous path to us to have as this imaginary of what we're trying to accomplish with the future of cities. So let's talk about what are the limits of technology as a lens through which to approach solving social problems? And sort of broad motif that I talk about throughout the book is this idea of tech goggles. The sort of vision that engineers in particular tend to bring to their work, the way that the world looks if you've been trained in computer science and other engineering fields. Both how do you understand problems? How do you understand solutions to those problems? And there are really two broad, what I call myths of what sort of how one perceives the world when looking through tech goggles. The first is this view of technology as the driver of social change, that it is technology that shapes the way the world evolves, that society evolves and that technology is both neutral and objective. That technology not just is the way that is the sort of driver of social change but that the social change it provides is somehow apolitical and socially optimal. That by bringing these tools of optimization to bear on social problems, we can address them in that manner. The former IBM president has this great description where he says if the leaders of smarter city systems do share an ideology, it is this. We believe in a smarter way to get things done. So the language and ideology here, there's no political ideology. There's just this idea that being smart is a thing that who could argue with that? Why wouldn't you wanna be smart? But of course in doing that we sort of lose sight of the contested nature of both the problems and the solutions and the sort of inherently contested nature of what a city is and what a city is striving to be. And so these notions of technology as a driver of optimal social change are particularly dangerous. And to see that, I like to turn to this example that is a simulation of what the city, a cityscape could look like in the future when we have full deployment of autonomous vehicles and it's a simulation of how you could have these vehicles zooming through city streets. You could have algorithmic systems that connect them all together. You get rid of traffic lights, you eliminate congestion. The researchers who put this together like to put this vision side by side with another vision of the status quo where you have a lot of cars waiting for each other and they turn red because everyone's angry and upset. And on first glance, this might look pretty incredible. No one really wants to sit in traffic but on closer inspection, some stuff might jump out at you. There's, it sort of hardly looks like a city at all. There are no people, there are no cyclists, there are no buses, no food trucks, no sort of any of the typical things you might imagine is not just being present in a city but actually essential to what that city is and what makes that city actually worth living in. All of those aspects of the world are not just ignored, they're sort of actively erased in order to create a simplified environment and abstraction of what the city looks like so that it can be optimized in such a sort of remarkable way. And what's particularly striking about this example is that it actually is supposed to represent a specific intersection in downtown Boston which looks something like this if you go there. Nothing like that, you spend a couple minutes at that intersection, you'll see tons of cyclists going through every time the lights turn green, there are pedestrians, it's along one of the major bus lines in the city of Boston and also along some of the major transit stops and transit lines. So suddenly we've turned, we are presented with this vision of what a smart city could look like, this technological solution to our social problem of congestion but when you actually go and study the context of that problem you see that there's much more going on. The problem is not simply one of congestion but that what is even the problem here in the first place? There are a great deal of overlapping intertwined people and needs and social challenges and social opportunities in this space that are sort of erased. And so there's this sort of key function of tech goggles that I think of as distortion where it takes a complex social reality such as any street intersection in a busy city and turns it into one where only the technological, certain technological aspects of that space are highlighted. In this case, what does the city look like through the lens of a computer where you might have sensors that are able to, and this I should say is not of that specific location but is a sort of graphic from a sidewalk labs which is a Google affiliated company that I'll talk about more later. This is like their vision, one vision of sort of the smart city that they're working towards. And I often have a slide especially for more urban planning audiences so I forgot that today where I look at, compare this vision of what does the city look like through a computer to some of the older images from the mid 20th century, sort of like what Le Corbusier's vision of urban planning is very much this vision of what does a city look like from an airplane and a city that was modeled after that vision and today we have what does a city look like through the lens of a computer, through the lens of the type of functionality and vision that a computer affords one to look at. So through this process of distortion, we move from understanding the complexity and the politics of life to these sort of abstract technical processes that seem very easy to optimize to provide incredible benefits. But of course, the problem is not necessarily the right problem and in doing this process of abstraction there and then optimization, there are a great deal of politics embedded in that process. There are winners and losers in terms of what gets deemed essential and needs to be kept within the abstraction and what gets deemed sort of inessential and can be thrown out. And we can see that in the traffic simulation where pedestrians are deemed inessential and in a future with those self-driving cars, it's hard to imagine that city being a place where one could actually have a thriving streetscape for pedestrians and cyclists and others. So to come back to sort of the intersection of cities and technology, I'll talk about three specific stories of things that I worked on in the city of Boston and how that embodies particular lessons we need to draw on the role of technology in improving urban welfare. So the first involves open data projects to take the data sets that the city of Boston has been collecting, various, you know, the different agencies that are collecting data, whether that's 311 information or property assessments or anything like that and bringing that out to the public to say, how can we, you know, have this data not just be internal, but open it up to anyone who wants to use it to enhance civic engagement and innovation and things like that. So we spent a lot of time not just in city hall but actually going out to public libraries, going out and talking to people to understand, you know, what sorts of issues did they have? What sorts of data did they want to see? Sort of specifically what data sets did they want and what would they want to do with that data? So we set up booths like this in the different branches of the Boston Public Libraries and we would stop people as they came in and out and ask them those questions about data sets. And what we found was that no one actually really wanted to talk to us. They would sort of politely wave us off or politely just say, I don't know. But it turned out, it was not that they were unfriendly, they weren't, you know, uninterested in talking to us but we weren't asking the right questions. We were asking them about data. When we asked them about their life in the city of Boston, how they liked the schools, what they were concerned about, how their kids liked the parks, all of the types of questions you might ask about someone's life in the city, we had these incredibly rich conversations, long discussions about the past, present, and future of the city of Boston. But it was never really functionally a conversation about data. I mean, we tried to bring it back to data as one resident put it, information is fine but I want a way to influence what's happening. We sort of came to see that the data was not empowering on its own. There was this long tradition around stuff like open data that if you just released the information it could just immediately solve these problems. But it became clear and it's almost, you know, sort of an obvious insight but not one that's often brought sort of front and center in smart cities, which is that social problems are rarely technology problems. The types of issues that people were facing were not necessarily going to be solved by simply having a data set, let alone even providing some analysis based on that data set. And so we really had to move our analysis away from thinking about the data as the mechanism to thinking much more about relationships, much more about what role the data was playing in building community and in empowering different groups to actually advocate for the type of change that they wanted to see in their community. The second story involves an analysis that I did to try to understand how we could improve the city's investment in sidewalk repairs and sidewalk infrastructure. So we were looking at the data that typically comes, that data that comes in from 311 apps, these are apps where someone can, you know, open up their smartphone, take a picture of a pothole or something like that, notify the city of Boston that there is, you know, an issue on the street. And this is the way that the city had been identifying issues and managing its backlog of reports or managing its backlog of repairs over the last couple of years. And we wanted to look at does this actually provide us with a good assessment of where we should be going out and doing that work. So we compared on the left, what you see is our own internal assessment that some contractors had done of the sidewalk quality across the city. And on the right, we see where the requests for service are coming in. So in both cases, the red clusters are indicating aggregations of low sidewalks that are deemed to have a low quality or a need of repair. And it's pretty obvious that these are not the same map. And if you're familiar with the city of Boston, what you'll see is that on the left side, there's a great deal of sort of lower right cluster, you know, it's sort of evenly distributed across the city, whereas on the right, really all of the sidewalk repair requests are in downtown Boston, the sort of financial district back bay, the sort of wealthy business areas and where the more wealthy residents and wider residents of the city live. So clearly this data was very misleading. We had this idea that by having everyone submit these requests, we were getting this incredible data set on where we needed to repair sidewalks across the city. What it turned out was we were actually getting data on this much more sort of sociological phenomenon, which was where are there people who have a smartphone, who feel that the government will solve their problem that they have the right to notify the city when they have a problem like this and that it will be solved. And that's a very different story of where the data, of where sidewalks actually need to be repaired in sort of in reality. So we're relying on data that's being collected in cities, whether through administrative functions or apps or sensors. We need to understand that often the data that we think, the insight that we think we're gaining, the data might actually be telling us something very different than what we think it's telling us. And that's an important thing to understand about the limits of this data and the need to bring more than just a sort of narrow technological mindset to understand what other things might be embedded within that data. And finally, there's a project that I worked on with the city's emergency medical services. So this is the group that responds when you call 911 and they send an ambulance or an EMT to if you have a broken leg, heart attack, whatever it might be. And we started working with them because they were overburdened. There was a big rise, the city's population have been rising over the last decade and the staffing and resources of the department was not keeping up. So there was a huge real limitation in the amount of resources that they had and they wanted to figure out how they could best use their resources to respond efficiently to the variety of 911 calls that were coming into the city. So when I began working with them, the sort of initial instincts, of course, of a computer scientist coming into this problem would be to say, okay, well, this is just an engineering problem. You have a certain number of calls and we can statistically model those calls. You have a certain number of ambulances and we can just track where they're going to be and we can create a sort of matching function that can optimize where your ambulances should be stationed at all times and which ambulance you should send out at any given time based on our forecast of what's likely to occur over the next hour or two hours. But it became very clear that that type of solution would not have been a solution at all. When I spent time working with the department and actually studying their context, that sort of solution one would not have been feasible given their operations and it wasn't even actually the right problem at all. It turned out that the problem was not just that they were somehow misallocating their resources, but that there was a significant number of calls coming in that were really not about acute medical care but were about individuals who were suffering from homelessness and drug addiction and mental illness who, and the EMS department was increasingly the group that was called to deal with those incidents, to deal with someone who might be having a breakdown or might be passed out on the street. But the problem was that these people did not actually require, they didn't need to go to a hospital, what they needed was connection to social services and other forms of welfare and there was really this huge gap between the services that the city EMS department was providing, which is really an EMT and an ambulance and a trip to a hospital and the services that were needed. It was not a problem of inefficiency, it was a problem of a mismatch between the needs of the public and the resources that were being brought to address those needs. And so what we ended up doing was not simply just developing an algorithmic system to optimize the existing process but we actually developed a new team within EMS that was specialized in responding to specifically these types of incidents, especially trained in de-escalation for someone in a mental health crisis and in connecting individuals to the different shelters and other forms of social services that were available across the city of Boston. So, right, there's this sense that actually using technology well relies not just on having what we might think of as really sophisticated technology but actually pairing the insights from data and algorithms with various forms of non-technological policy innovation, understanding the variety of different ways that we can address these problems. So, thinking about these three stories, we can see three really significant pitfalls of the broad move towards smart cities. One is expecting technology to solve all of our problems, treating technology as this neutral force or a neutral tool for gaining insights and focusing on technology at the expense of other forms of innovation and change. So, and there are two sort of broad additional challenges that smart cities bring that I'll talk about now. One are the ways in which the deployment of this technology can really fundamentally reshape urban power and politics. People like to talk about the smart city revolutionizing urban life. And I would say that it will revolutionize urban life but not in the way that it's often held is doing not by creating these optimal utopian cities but by really transforming who is actually in control of urban life and who gets to make decisions and who has power and autonomy. All of the technology in smart cities really relies quite centrally on data collection. And so there are massive amounts of surveillance baked into smart cities, both from the government as well as from the private companies that are often the ones behind the sensors that you might see on different technologies. And of course, privatization. As more and more technology is integrated into city governments, that technology is almost always adopted or created and managed by a private company. Cities are creating some level of maybe data analysis and they have analytics engineers on their teams but they're not creating massive deployments of sensor networks. So increasingly technology companies are making a play for city space and city resources through these public private partnerships. One of the most significant is in Toronto where there's a major project where Sidewalk Labs is working to really develop and they call it develop from the internet up a city neighborhood in Toronto that's been sort of undeveloped for a couple of decades. I'm not sure how long. And there's a lot of pushback, not just about the idea that they're doing that essentially on Sidewalk Labs is essentially in the Google family. So a Google company coming in and developing the city neighborhood obviously raises huge surveillance and privacy concerns but also concerns about who actually gets to make a decision about urban life. What happens when an agency, normally you'd have a development board or the city government that's making decisions about zoning and planning and all of that. Suddenly you have Google coming in able to make those decisions in a very different way without any sort of enforcement for accountability or transparency. And perhaps in some ways more significantly are the ways in which these visions of technology as solving social problems can obstruct the types of more systemic reforms that are often needed to address these challenges because technology often can come in as optimizing these systems but it's really optimizing the status quo. It's optimizing the way that systems already work. And if we're not, and actually we can think back to the EMS example, if we had just come in and optimized the resources and the practices that the city was using, we would have really been on a, we would not have solved the problem because that would not have addressed the right problem. We had to actually change the broader model of what was going on and what practices we were actually using. And more importantly, recognizing that a lot of the problem that was leading the city into this space in the first place was the limited resources for EMS, right? The limited sort of the legacies and politics of austerity that had meant that urban city departments did not have the resources to keep up with demand. And so there's this pressing need to optimize them rather than actually provide them with more resources. And we saw, for example, with the simulation of self-driving cars how that can lead to a model of not continuing to move towards dense urban neighborhoods, not focusing on transit and buses and walkability, but emphasizing, well, we don't need to do any of that because self-driving cars are around the corner and they'll be this incredible thing so all of our congestion worries will go away. And there are actually a number of cities that have developed partnerships with Uber and Google and other companies and even started rather than investing in various forms of transit, just providing subsidies for people to take Uber trips or things like that. But it's a very different model of where does that lead us five, 10 years down the road, right? We're not actually addressing the deeper structural urban planning problems that are at the root of these issues of sprawl and greenhouse emissions and a lack of mobility. So, okay, so how can we actually do any of this well? This is all sort of dire and grim and are there actually things we can take away from all of this for how do we integrate technology into city governments, into urban life to address social problems? So the frame that I bring out in the book is what I call it smart enough cities. This idea of removing the notion of smart cities, which is very much, as I said, focused on technology, really positioning technology as the key thing, right? The way you get smarter is by having more technology. So by inserting this notion of smart enough cities, the goal that I have is not to throw away technology, not to say that all of this is terrible, and certainly not to say that the answer is to be smart or to be dumb, but to say that we need to really think about our technology from a much more pragmatist approach, right? Of what are we actually trying to solve? The technology is not an end in itself. The technology is a means towards achieving variety of social ends and is really only as valuable as to the extent that it's able to actually help us do that. So what are some principles that we can have to actually embody that? It sort of maybe sounds obvious, but how do we actually get to that point where we can approach these problems in that manner? The first is, and in many ways the most important to me is to approach this as addressing complex problems rather than trying to solve artificially simple ones. City life and urban processes are incredibly complex and to optimize systems in the way that we often want to with technology to solve them often requires creating artificial simplifications of those problems, right? We take a complex city street and we turn it into a space where all that there is are cars and then we can solve that problem. But these problems are never once to be solved. I like to use the language of addressing because I think once you start talking about solutions to complex problems, you end up simplifying the problem because a solution to the real problem is impossible. So in one city that has approached this in a very different way is Columbus, Ohio, which in 2015 or 2016 won a $40 million grant from the Department of Transportation to sort of create like a new smart transportation system. But rather than just focusing on ending traffic with automated vehicles, they really focused on the connection between social mobility and sort of physical transportation mobility and understanding the ways in which these problems are intertwined. They actually went out, went to different neighborhoods particularly low income neighborhoods where there are particularly acute public health challenges and tried to understand the variety of barriers to accessing mobility, accessing doctor's appointments and job interviews and all of that. And it was not simply just an issue of congestion but an issue of a lack of resources for childcare, a lack of information about what types of transit and transportation were actually available. Issues where certain types of transportation providers like Uber typically require a credit card and if you're unbanked, you are not actually able typically to get access to those. And so they, as they described after the fact, they said, we really needed to look at it from a more holistic viewpoint. As geeky technology people, we wouldn't have thought about these things had we not considered the whole picture. So they were able to come up with a broad array of solutions from improving access to internet, improving the accessibility of buses and the accessibility of different payment options and providing, making accessible even, paying for rides through Medicare and Medicaid often for people who are going to doctor's appointments, things like that. So really addressing the particular intricacies of why these issues of mobility were such a problem. And focusing on the real problems facing real people rather than trying to create these optimal solutions that sound really great on a press release. Second is implementing technology to address social needs and advanced policy. This sounds really so obvious to seem really trivial, but it's actually really the place where a lot of smart city projects go wrong where the allure of technology prompts cities to adopt the types of goals that align with technology because the technology seemed really incredible. Cities start thinking in terms of efficiency and optimization, which is what these technologies provide and what the technology vendors are selling them when those values may not actually be the things that they were striving for in the first place and certainly don't apply in all aspects of urban life. In many cases, inefficiency is quite valuable. So maybe I'll skip this and sort of to move on but we can talk about some of the civic engagement efforts underway if people are interested in the ways in which civic engagement in particular, understanding the value of inefficiency is incredibly important, right? Civic engagement, conversation is not meant to be a streamlined process but meant to be one of coming to terms with other people's perspectives and all of that. So places in city government and urban life where inefficiency is actually incredibly valuable and carving out space for that as a value to think not as something to be erased but something to be cherished. Number three and sort of along those lines is prioritizing innovative policy and program reforms above innovative technology. Again, thinking about the ways that a policy change is really the driver of any type of urban progress. It's not the technology on its own but the technology perhaps in conjunction with other aspects. So one of the major places of smart city development is around policing and there's lots of efforts to create algorithms known as predictive policing to try to forecast where crime is going to occur and optimize where to send police to address those crimes and often in this way of responding to concerns about police discrimination, racial profiling, all of that but of course that type of system is not really addressing the underlying problem, it's simply addressing perhaps to some extent where police are going on a day to day basis but other cities are taking a very different approach so in Johnson County, Kansas for instance rather than thinking about it as how do we just optimize our existing policing system? They've spent actually the last 20 years really thinking about how do we change our policing strategy or really our public safety and social welfare strategy and understanding that the solution here is not simply to take someone who's in some form of distress and lock them up in jail but to actually provide them with different types of services that might prevent them from coming into contact with police in the first place. In many cities, especially local jails often are sort of a ton of people end up there because of unaddressed mental health and drug addiction crises and the police become the response to those problems but in places like Johnson County, Kansas they've really reshaped responding to those incidents with people who are trained in deescalation and bringing them to social services and then ultimately bringing together machine learning systems to really improve the functionality of that rather than simply waiting for a call to come in for someone who is maybe in crisis and have to be dealt with in some form and actually trying to understand how do we predict who's likely to end up absent to any sort of intervention who is likely to end up in jail for this type of mental health crisis or something like that. How can we proactively provide them with services, get them on a better track and prevent this from ever happening and by pairing both the policy change with the algorithm they've been able to really boost their ability to do this and their ability to address these problems before they even arise and this is a project that's been ongoing for several years and they're still working to implement this type of system but on one test they found that over 50% of the people that they identified as likely to end up in jail would have did in fact end up in jail and had they prevented all of those bookings they could have prevented 180 bookings into jail and over 18 years of total jail time so the gains here are pretty significant and in particular, the gains not just of a policy change but of a long-term planning vision, right? This is something that they had been working on for about 20 years in various ways so it's not just sort of, you know the way we talk about smart cities is often the technology come in you'll snap your fingers, the problem is solved it's much more that these problems take incredible amounts of time to get the right policies to build coalitions, all of that. So another aspect is ensuring that technologies design and implementation promote democratic values I described a little bit earlier how the way in which we conceive of these problems the way that we conceive of these solutions and think of what we're actually trying to solve and what needs to be kept in any sort of abstraction of urban life is really a source of incredible power to be able to say that these are the values that our systems are going to embody and these are the values that they're not going to embody and of course one of the really significant ones here is around privacy and all of the issues of surveillance I often show a map of New York City where many of you I'm sure have all seen the link NYC kiosks that are out and about around the city and those are also owned and operated essentially by Sidewalk Labs the Google company I described earlier and the way that that system operates is that the city does not pay for it the city has created a partnership with Sidewalk Labs Sidewalk Labs provides public internet to anyone in exchange for the ability to collect data and sell ads to people based on that data so what is in theory this great public service of free wifi something that is quite important is then sort of offered in exchange for the imposition of surveillance the imposition of data collection and this is something that really was never put up for any sort of public discussion another sort of very topical local thing is around algorithms many of you will likely maybe have heard of the New York City algorithmic decision-making task force and so that was an effort to do to turn the way so the New York City has a lot of different algorithmic systems from policing to schools to the fire department that manage various systems and again the question of who gets to develop these things does the public have any insight does the public get to audit these systems in what ways are these systems prioritizing certain groups or benefiting or harming different groups and how do we make sure that the public actually has insight and power over that so that was a process of about a year a little bit sort of dysfunctional and disappointing for those who are involved but that sort of wrapped up over the last couple months and there's a lot of ongoing debates around especially in New York City Council around how do we define the scope of what is an algorithmic system that needs to be regulated and how do we do that what sorts of city mechanisms what sorts of people need to be responsible for that there's an open call for a chief algorithms officer in New York City so this is a very sort of hot topic right now here in the city and something that there's a lot of opportunity to get involved in locally and then sort of finally is a piece on sort of thinking about it actually operationally what does it mean within city government to use data and technology effectively the visions for smart cities as I described are sort of this magic drop of a hat you'll have the technology your problems will be solved but typically it's not about having the best technology and really using this well as about implementation and integration of technology and policy and people there's a fun story and from Boston when they were had lots of Boston the city leadership would have lots of conversations with vendors and at one point a major tech company came in Giant Fortune 500 company comes in and gives them a great pitch and when the city leadership asks okay so we see all this technology what's it actually going to be used for the sales person just says that's the exciting part we give you the platform and the data and you get to figure out all the ways you can get value from it not exactly a great pitch for a system that would have costed millions of dollars up front without any clear process of how to use it with complete sort of lack of attention paid to those aspects so to sort of close out another local story actually I'll talk about sort of why you need to have attention to integrating all of this together based on something that happened in New York City in 2015 there was a massive outbreak of Legionnaires disease around August of 2015 which is essentially like a acute form of pneumonia so you don't want to get it and it was incubating in the cooling towers on top of large apartment buildings and places like that that sort of managed the air conditioning systems and so this happened there are a couple people I think had already died from it and there the city, the public health department really had to figure out how do we stop this from spreading how do we make sure that this doesn't become a really big issue and the starting point because it was incubating in those cooling towers the city had to actually go out and inspect and clean every cooling tower across the city to make sure that they were had it under control the problem was they didn't actually know the location and owner and identity of every cooling tower across the city that was not something they had ever collected they had various types of data about apartment buildings and owners split across a number of different departments and they could sort of infer which ones might have cooling towers but not others so there was first of all this incredible challenge of actually bringing together data sets that had never been brought together before information from the department of buildings the department of finance typically they have their data no one really thinks about it outside of the department so it was incredibly onerous to actually bring it all together but ultimately they managed to some extent they were able to build some machine learning algorithms based on that helping to direct the various city agencies to do the inspections and the cleaning and stop the crisis before it got out of hand but what became clear from this process was that there was a huge limitation in the city's ability to actually take advantage of the various resources that it had they didn't know what data they had they didn't know people in one department wouldn't know what information the other departments had often didn't know how to use it all of these issues and so they developed two different frameworks for dealing with that one is what they describe as data drills which you can think of as a normal fire drill you would pretend what happens in a fire they would have data drills where they would pretend what happens in a crisis and have 24 hours where they would get the heads of a bunch of different departments to actually come in and say okay there's a blackout in Brooklyn we need to bring together all of our data that we have to understand where we need to send help and who's going to be most vulnerable based on the information they have so really practicing how to work with different types of data that's spread across different agencies building systems to actually create some of that infrastructural backbone so that information about the different buildings and parcels across the city could actually be integrated across different agencies in a sustainable way and ultimately they sort of have this framework of data management leading to data support leading to the analytics and everyone wants to talk just about the sort of top part of this, the analytics but it's really the management and support that makes all of that possible so hopefully I've sort of convinced you of the many dangers of smart cities the value of shifting to an approach of smart enough cities really emphasizing the need to integrate technology into holistic visions of democratic and egalitarian urban futures thank you I'm happy to take questions, comments any thoughts? I'll also note that the book is also available, open access online so if you don't want to actually buy it you can just read it all online for free and access it that way so yeah any questions? Yeah I really appreciate it I have a question about like you're in the 5th version and then you're in the tech office at the start yeah and then thinking about what are your thoughts on times when taking off the tech goggles for example the discussion around an official recognition ban which I'm talking about for a reason that seems to me like specifically being off the tech goggles not replacing it with a different solution necessarily but I'm curious about other thoughts on approaches that aren't just like here about decisions but also specifically have to be changing the tech much better yeah yeah absolutely I mean so so the the subtitle I almost went with was taking off our tech goggles to so I mean and I would say that the yeah the sort of five sort of design principles I describe are less about you know better tech and are all specifically about how do we approach these problems without tech goggles right what does it look like to not to take them off to not have that approach and to not think of that that we need to have better tech but how do we actually move the conversation away from the sort of lens of technology where we're you know able to debate okay well this technology is imperfect but then how do we just create a better version of that technology to a conversation of it's not clear the technology let's let's go back to square one and say what is the problem that we're trying to even solve what are the ways of solving that problem and then you know maybe technology has some role there but really expanding the frame of analysis which is exactly that point about about taking them off so yeah yes there seems to be an underlying assumption or there is a threshold like the city has to have certain kind of resources and kind of capacity but actually and so in your case studies some of these are really places whether you think technology actually is worth the money I mean you said we're not ditching so I'm just curious in a cost-benefit way is smart enough to have the technology for choice whatever the reason is yeah I mean there are absolutely significant differences in the resources that you know a small city has versus a place like New York City but I think in many ways you know the point one of the main points that I try to show is that it's actually not about having the sort of fanciest most sophisticated technological sort of rollouts and often it's relatively simple things like just pulling your data together and having some you know data scientists build a system on top of that that is actually often most effective and so I find that I mean so you know looking at the example for example of like Johnson County, Kansas that's not a massive city with huge amounts of resources or Columbus, Ohio I mean that's a pretty big city but not you know the system is really the approach is really less about you know it's exactly saying we don't need to have link NYC, we don't need to have the massive infrastructure investments in many cases that are millions and millions of dollars those are often sort of a red herring and a wrong approach and it's sometimes the smaller things that are more effective and sort of building practices from the ground up in cities there are definitely challenges that you know the smaller cities face just in terms of getting resources and actually having even one or two people who can focus on that work but you know I think that the the goal at least is for the idea of smart enough cities to be sort of flexible enough to think about less it's it's less about you know it's moving away from that idea of having the most technology to thinking about what type of technology is actually useful and actually sort of tailored to our particular needs yeah what does enough means and who gets to define what is enough is it like you want to define it so quickly you have to find on the speed level are there like best practices approach so what is enough I changed effort but I was like enough is a big yeah yeah yeah I mean it's it's huge and I wouldn't you know I think locally real I mean to me that the entire question is right who gets to decide and the you know I see it less is about you know I sort of have the the media maybe like mezzanine level principles for how to approach these things but certainly I wouldn't want to come in and say you know this is the type of technological system or pairing that is always good because it has to be very contextualized I think and I am particularly really focused in a lot of my work on making sure that more and more people are able to have those be part of those conversations so a lot of the efforts that really have even come about and really gained momentum in the last couple years since I wrote the book and it's come out have been you know stuff like the New York City Council effort on algorithms various surveillance oversight ordinances facial recognition bans all efforts that are actually bringing more people in providing more transparency about these systems and allowing more of the public to have a voice and saying you know what that's we don't want facial recognition to be used by the police there was just and I think maybe last month there was a hearing for something called the Post Act in New York where a bunch of advocates came and talked about why we needed more information about various algorithms and technologies being used by the police department so to me the question of who gets to decide is so central and I'm more interested in sort of creating the mechanisms for more and more people to be part of that conversation than sort of specifying exactly where the threshold might be of you know enough versus not enough or something like that yeah the emerging trend of the community technology within the field of smart cities where the technologies instead of the technology they actually work with the social actors like the existing social infrastructure like the nonprofit the churches and libraries like who has been working inside the neighborhood like that is that understand the social and structural divide I guess my question is like especially from your case studies that you mentioned Ohio and Kansas you seem to be advocating for building kind of a new layer of social infrastructure same thing for example like almost like from like a so I guess like the question is like are you suggesting to say creating new services new support and management to specifically work with knowledge or what do you think of like the current approach of just like working with the existing social organizations in that way you know they wouldn't be like standing in the way of the structural reform is that they actually work with the community organizations although the cavity might be like those social organizations they might not be tech savvy and they might not really have the background enough for that like I guess what's your what's your stand on that yeah yeah that's that's a super important topic and I very much view it as the existing organizations both within cities within the community community groups I mean that's really the place to start very much in opposition to the way that it's often you know the private companies Google Cisco etc that are coming in and sort of parachuting in with these various technologies so then the question becomes you know how do we shift to that model how do we actually enable technology to be to the extent it should be sort of integrated into those into those institutions one model that I think has been pretty successful is when city governments actually bring in data science teams internally that's sort of the group the type of group that I was in in the city of Boston where we had a team of data scientists and others who were working sort of but working in the city department but then spread across the city agencies in terms of what types of projects we worked on so I describe you know working with EMS working with public works the fire department all of that and really to this point of integration the point is to have the technological expertise and knowledge and capacity on the inside of these organizations rather than on the outside either as a completely independent layer or completely residing with the companies that are coming in and then there's been a real shift over the last you know maybe five to ten years where you know in 2012 a you know big company would come in they would pitch something to a city and no one in the city really knew enough about technology to actually ask the right questions to actually sort of like call them out on false or misleading claims things like that or even to know what they wanted and now over the last couple years it's completely reversed where companies are coming in with you know sort of these not that impressive solutions or approaches and now there are enough experience with technology and enough people who have who actually are trained specifically in building technology to know you know this isn't actually an effective system your claims about the performance actually don't hold up and that's very much in the internal side you know obviously a lot of this and a lot of I think the most exciting work is happening outside of city governments you know groups like the ACLU have been super important and other groups like that really bringing in more and more technologists and I in Massachusetts when I was there worked a lot with the mass ACLU and they do great work pushing you know bringing different community organizations who care about racial justice about policing reform to the table to push for legislative change push for democratic controls on this type of technology so you know those sorts of efforts are really key and you know the challenge then is continuing to just bring more technological really just fluency into community groups and places like that less because they necessarily like need to have more technology but because of nothing else they need to understand the ways in which technology is intersecting with them right so whether you know if you're I mean so even just to give one example over the last year there was a major effort in response to there was an apartment building where the the landlord wanted to put facial recognition systems in the building this is like the Atlantic Towers if you want to look up more about it but the residents you know wouldn't you know push back on this they worked with my group at NYU and now and some others and was able to you know push against both this system and actually now lead to various types of legislation saying that facial recognition cannot be used in public housing things like that so again this is an issue you know if you're if you care about housing justice and then you might not think you need to care about technology but the technology is starting to come to you if you care about policing and racial justice and economic justice and any of these issues the technology is certainly coming to you so in some ways it's sort of shifted to being an offensive quote unquote where it's unique to be building the technology although there's value in that to at least being able to understand the technology to push back on the ways in which technology is imposing on you yeah data technology like a epistemology question I'm down but maybe but you kind of started your talk in the beginning and like kind of the data bias kind of the presentation algorithm bias of data and technology and I wonder you know and like kind of how that kind of inherent like inherent issues with data and technology can lead to kind of certain unenvelopment outcomes and thus the reason kind of why you need to work with and you know government and office etc but I wonder if you know it can challenge that notion of it and ask like what if we you know were in a scenario where we had that call or you know like constantly consequences like quantify human self or I forget what it's called where you have a sensor on kind of everything on building on people on you know streets etc so what if you were able to kind of like collect all the data right wouldn't that kind of lead to more accurate less bias kind of algorithms etc great question okay so I'll leave aside any of the like privacy surveillance aspects of that let's just think about from yeah so so there yeah I mean so the main way I would talk about think about that is that there's an a there's an assumption often when we talk about the limits and failures and harms of technology that the issue is a technical problem right that the issue is oh we just don't have enough data the data is actually biased if only we had more representative data and more of that representative data we could solve these problems but the problem is that the so much of the game so to speak is an actually understanding what the problem is in the first place being able to have these broader discussions and so in many cases it's you know so in my area of thinking about criminal justice reform and algorithms and there's lots of conversations about why these algorithms are biased for example but the real problem is not that the algorithm is biased the real problem is that the algorithm is actually not an effective tool for achieving the types of criminal justice reform that are advocated for that it's even supposed to be creating similarly you know of facial recognition the problem is not that the underlying datasets are biased and so that it's less accurate for one group than another the problem is that it's creating this massive surveillance network and the more accurate it is the more legitimate it becomes as a tool to deploy you know across our public spaces so a lot of my actually more recent work is diving into exactly these questions of why do we have to move you know I sort of talk about it in terms of epistemic reform right move from not just thinking about reforming the technology but reforming the lens with which we're approaching the role of technology in creating social change and moving from a model where we see everything as we're sort of tweaks within the within the bounds of what it means to do engineering to actually opening up to other modes of analysis and understanding that the best possible engineering in the world is not going to solve these problems because it's not the right approach and yeah I actually particularly like the model of imagining okay we can critique the technical aspects of this system but let's imagine that the system is perfect and let's talk about why it's still bad and that being I think a really important sort of mode of analysis because if we just talked about the technical issues we invite the like well here's a slightly better technical solution and what we need is to change the approach to even understanding what the problem is in the first place is that yeah just a quick question you talked a lot about democratic values and about how we also have to involve the community to avoid the pitfalls of like tech goggles and everything could you maybe go through like an example or two of what like good community involvement looks like or like some cases you've come across because I think with the New York City Task Force report one criticism was that community voices they weren't sufficiently heard I think the Toronto example community voices were kind of bopped it by like city and sidewalk laps to Chris so what would like a good model of community involvement of those projects? Yeah yeah I mean those are those are great examples of what it doesn't look like or really what a superficial version of it looks like I think it's very easy to do to do it in a way that sounds like you're being inclusive and having these conversations but you're not actually you know I I think it's really hard there's not like a great singular model of what it looks like and I would say there's sort of there's two different ways of thinking about it probably more than two but one would be right the model of sort of proactive civic engagement things like that the other is more of a thinking about it in terms of like I don't know like a legalistic model where it's like okay something like the the city council legislations in a variety of cities that say you know before you want to use the surveillance technology you have to have a you have to have a report you have to have a city council hearing you need city council approval that is one sort of mechanism with teeth for bringing more people into the conversation and I think that's an incredibly powerful one as we've seen sort of and many of you probably have noticed much more deeply than I do around histories of urban planning and urban civic engagement which just sort of bringing people into the table is not always enough and often can be sort of you know avail for explicitly ignoring that type of input but I do think one story I actually skipped over was in Chicago they were developing a set of sensors known as the array of things sort of a sensor network to collect a bunch of data and you know raises all of the alarm bells about data collection and surveillance but they actually did a pretty I would say really probably the most robust job I've seen of any city to have conversations about privacy and what people wanted these city what these technologies to look like actually releasing draft versions of policy proposals and privacy policies and actually allowing the public to comment on those adjusting their processes based on that holding sort of educational info sessions to teach people about this technology so but you know a couple aspects that are worth noting there one and a really important thing is right the structure of this system so that system was not owned by company it was actually through academic institutes it was funded by the National Science Foundation so you know lacking the profit motive to collect a lot of data it's much easier to be you know sort of magnanimous in having these conversations and taking care of that input because there were research agenda involved but not sort of the fundamental bottom line in a from a profit perspective that you know makes one far less interested in hearing about ways to collect less data and you know another aspect being that that was not necessarily a set of civic engagement around you know at the first part of the process saying you know to what extent do we want this technology at all how do we want you know if we were to have a bunch of scientists and engineers and millions of dollars in funding to do something what would we want to do you know there's still a limit of how much they're brought into sort of the the broader conception of the project but I think that that one and that's I read about that in the book is a good example of sort of like towards the best case example of doing something like that but again it relies very much on the good faith actors and the lack of you know sort of the institutional structure of what's involved there and I think there's a lot of need for things that are also more sort of with teeth and enforcement to actually say you need to have these people at the table you need to have these conversations before actually embarking on this sort of technological usage yeah one of the things about that I see of particularly all of the things that we remember is what is completely new yeah the new thing you figure out what to do with it but on the other hand technology in many forms is completely new but on the other hand technology in many forms is completely new kind of the city since the end of the system so I wonder how you think about those kinds of historical precedents in the way in which you would look like yeah yeah I think there there are actually a couple of different sort of histories that I draw out in the book that I think are really instructive one I sort of briefly alluded to sort of the histories of you know high modern urban planning like babuzia and those and how we've long had this vision of cities of you know we can create optimal cities based on the technology of today whether you know a hundred years ago that technology may have been very different than what it is today then you had last century the idea of you know cars come in and we have this whole pitch around the motor age and structurally very similar right that we have companies with a huge desire to sell and create this vision to create a market a huge market for themselves so that everyone went by cars and they created these incredible utopian visions of you know the futuristic cities looking forward to the 60s or 70s or 80s without traffic and you'd have these beautiful freeways and cars going through and there's like great videos online of that so totally different technology but very much the same story and I think that those are really powerful and I think we this is not the first time we've thought that there's a utopian future made possible by technology and all of those have turned out pretty badly and we're like acting this you know case of cars we're like really actively trying to undo all of those mistakes so let's like try to see those parallels and think about how we can avoid in you know 21-20 being like wow we really messed up with all of those smart cities and let's try to deconstruct all of those now if we even make it that far but you know I think that those are really they're definitely there and they're quite powerful I don't always know how convincing they are different mode of argumentation for different people but I find them compelling yeah yeah I guess you talked a little bit about the emotional or institutional reasons like for example I imagine that a group of people who read your memory you say you can still be doing this you know because there's a lot of incentives they're great for metrics or you can do that whether that's in here or not you know I'm sure there are ways to look into contact with a very high degree you know you know I'm sure I imagine there are a lot of institutional and personal reasons why even as we're and really what you say you know a lot of compelling reasons but hopefully we just have some contact with that you know you know because and do you have any thoughts on that and also is it the role of those five of those strategies you're talking about to adapt to the language like that's for actually you know like specific results or to kind of change you know all together ensure like someone agreeing with what you say and the reason there are still reasons why you know you're feeling what it means to that that this is about performance just about it being you know that they're yeah yeah so I'll I'll quickly answer that I know it's two thirty so I'll give a quick answer we can talk more after I think I shrug it with that all the time where I sort of increasingly hear city officials talk in the language that embodies things that I talk about but then in practice they don't really do anything that different there's definitely political incentives I think one reason to go sort of more towards a public audience engagement sort of a a book like this is to try to shift some of the dynamics where for a city government it's not no longer the incentive where you can like do this technology whether or not it works you're going to get some great press and all of that that even now there is much more of a model where like the press will be like well wait there are privacy concerns but wait look at this other city and there's enough fluency now to know and distrust in big tech companies so that some of those incentives are shifting and then you know I think to some extent I try to make things legible within the language that sort of city operate managers are often working in but you know a lot of this ultimately is actually about breaking out of that mold and realize in the limitations of like a really narrow performance management type framework in city government and that's a large task that goes beyond just smart cities but I think that that sort of mode of thinking is absolutely sort of at play here especially in as a barrier from a mode of thinking that sounds good to actually doing something different about it and a lot of what I do is try to think about how to bridge that divide so thanks all right that's 230 I'll let you all go thank you