By Veena Narashiman ’2020 Originally a portmanteau of the words hack and marathon, a hackathon typically occurs over a day or two, bringing together computer programmers and others to solve a puzzle or invent a creative solution. During these 24- to 48-hour periods, participants are encouraged to form groups and collaborate, completing the hackathon with a rough prototype or ideas that can be presented to judges for prize money. Over the years, the adrenaline rush that often drives these competitions have created some famous “hacks”: the messaging app GroupMe and the Facebook “Like” button were both conceived during hackathons. Although hackathons may feel new, they are nearing their twentieth anniversary. The concept was born in June 1999, when UC Berkeley alumni John Gage challenged attendees of a Sun Microsystems event to write a multi-user Internet program in Java for the Palm V. Almost two decades later, hackathons have been organized to advance all manner of technologies in practically every sector. And increasingly, hackathons have been launched to solve societal challenges, such as natural disaster preparedness and government transparency. But are “hackathons for good” really effective, given that rapid prototyping is rarely a fix for entrenched societal problems? For technologists like Luca Ibota, a former Apple employee who has been active in many hackathons for good, the most important aspect is “identifying the problem you have and the ideal outcome you want.” In fact, said Ibota, the key to a successful hackathon for good is “precisely defining a problem or challenge.” Still, some Cal students are skeptical about hackathons for good. The most cynical argue that incorporating buzz words such as “social impact” and “corporate social responsibility” at hackathons is a smart public relations move for technology companies looking to improve their public standing. Other Cal students insist that incorporating social good goals in hackathons is a testament to Silicon Valley’s aim to think more holistically and ethically about technology’s effects. For many, UC Berkeley hackathons that take on a social impact lens are seen as a reflecting a student culture that prioritizes hands-on learning and that seeks to solve grand challenges like climate change and food insecurity. For Swetha Prabhakaran, a UC Berkeley sophomore and computer science student, the 21st century requires companies, nonprofits, and individuals to make solutions to intractable problems a priority. Either way, the number of UC Berkeley hackathons focused on social impact is on the rise. Causes have ranged from building prosthetics for people with disabilities to developing apps that enable students to source fair trade goods. Student-run organizations have championed these hackathons as a way to ethically fill consumer gaps. In April 2018, a partnership between the Sutardja Dai Center for Entrepreneurship and the UC Regents Chancellor’s Association hosted Cal Innovates, a hackathon aimed to bridge the engineering and business disciplines. Prabhakaran, who organized the event, said attendance was high because engineering and business students have started to “soul search” for meaningful impact. “Berkeley students have a strong entrepreneurial spirit—you can see it everyday,” said Prabhakaran. “But conversations about using business to help others are happening on small scale. The hackathon is a way to help students do this on a bigger scale.” The Cal Innovates hackathon presented no strict problem to solve. It allowed participants—of which 40 to 50 percent were engineering majors and 30 to 40 percent were from the Haas School of Business—to build and plan deployment of prototypes. During the competition, participants listened to speakers or pursued their project with the help of guides. Professionals from SkyDeck, Cal’s startup accelerator, helped students with presentations and judged the finals, while employees of GoDaddy and students from BlockChain at Berkeley helped participants with technical aspects of their designs. Finalists included a project for civil engineers in developing countries to create sustainable bridges. Another example of a UC Berkeley hackathon geared toward positive social impact was EnableTech’s “Make-A-Thon.” Held in April 2018, it aimed to connect those building prosthetics with those who use them, per the club’s motto: “Build with them, not for them.” Spanning 48 hours, the event allowed participants, formed into groups of six people from different majors, to rank which of EnableTech’s active projects should be improved. Kyelo Torres, a rising senior in mechanical engineering and the event project leader, spoke of the hackathon’s purpose: “As students, we are taught only theories. The second we are asked to do something, we get lost. The idea of hackathons is to practice the real world aspect of things.” Amy Dinh, programs manager of the Jacobs Institute for Design Management, said she believes the reason for the uptick in hackathons with a social good emphasis is simple: “People seek a challenge, and there’s nothing more challenging than the wicked problems of the world.” Yet she dissuades students from expecting implementable solutions post-hackathon, highlighting instead the importance of the iterative process. “The point of a hackathon is to get creative juices to flow,” she said. “It’s not realistic for the ideas to be polished; rather the point is to kickstart a new team or

By Tamara Straus For the Pinoleville Pomo Nation of Ukiah, California, collaboration has not historically been a word used to describe interactions with white Americans. As late as 1950, native people were not permitted to walk on both sides of the street and signs in Ukiah’s storefront windows read, “No dogs or Indians allowed.” Angela James, vice chair of the Tribal Council for the Pinoleville Pomo Nation, remembered this history was she was approached by David Edmunds, her tribe’s environmental director, about a possible collaboration with UC Berkeley. The goal was to co-design a sustainable housing project for the low-income families of her 300-person nation. On the one hand, James, a mother of four, was eager to advance the education of young tribe members and teach them to live in two cultures. Yet her mind jangled with stories from her grandfather Smith Williams. He had told her about Ba-lay Ba-lin—the Bloody Run—an 1871 atrocity in which white settlers violently forced Native Americans off their land, turning the Eel River red with their blood. In 2007, when James first stepped onto the UC Berkeley campus to talk with Mechanical Engineering Professor Alice Agogino, now education director of the Blum Center and chair of the graduate group in Development Engineering, and graduate students Ryan Shelby and Yael Perez, she was aware that this was the place where Ishi, the so-called “last wild Indian,” became the research subject of anthropologist Alfred Kroeber. She also knew there was an ongoing dispute about the Hearst Museum’s return of 12,000 Native American remains to California tribes. James’ warm-up to the UC Berkeley engineers was slow. She recounted that because Shelby is African American and Perez is foreign (Israeli), she felt they might be worthy of her community’s trust. She also wanted to believe that “science can cross cultural barriers,” and she observed from Agogino’s classroom that engineering was no longer “just a field for white males.” At the same time, the interdisciplinary UC Berkeley group called CARES—Community Assessment of Renewable Energy and Sustainability—was seeking to tread new water in the field of development. “As we worked with the nation on the sustainable housing project, our understanding of development changed,” said Perez. “We realized technology, and technological ‘fixes,’ are not enough. We needed to start with what sustainability meant to the tribe. And they had a lot to say about sustainability, because of the way they view their connection to the Earth—resulting in unexpected design decisions around heating, water use, solar power, and the shape and functionality of their homes.” This insight about collaboration led the CARES group to a development methodology called “co-design.” The term, which builds on human-centered design, user-centered design, design thinking, and participatory design, goes further in empowering stakeholders in the decision making and design process to recognize that users (or locals or recipients of development assistance) are key participants in their own economic, environmental, and sociopolitical advancement, with significant contributions to offer. In essence, CARES, a forerunner of UC Berkeley’s development engineering graduate program, embraced co-design to address the disconnect between the creation of technological innovations by engineers and the needs, preferences, and cultural views of the people who will use them. Explains Agogino, “The 10-year collaboration with the Pinoleville Pomo Nation shows a number of things: It shows that development projects can and should be local, not just international. And it shows that development solutions can range from how we design to how we publish academic research. The journal articles that have come out of the PPN collaboration have notably been co-authored by PPN members.” Since 2008, CARES has collaborated with the Pinoleville Pomo Nation on engineering, architecture, and educational projects that have tested the boundaries of development. Co-designed results have included sustainable housing, renewable energy power systems, water restoration and management projects, and most recently science, technology, engineering, art, and math (STEAM) workshops for middle and high school students and a K-12 maker space. Like past collaborations, the recent educational one was the result of shared interests and available funding. Zhao Qui, project director of the Pomo Youth College and Career Success Project, explains that in October 2016 her organization received a Department of Education grant to fund new cultural and academic enrichment activities for native students; one area of concern was low access and achievement in math and science. At around the same time, the Blum Center received funding from the National Science Foundation to support development engineering students working on InFEWS (Innovations at the Nexus of Food, Energy, and Water Systems) for low-income communities facing extreme challenges. And the CARES team, energized by development engineering graduate students and Global Poverty & Practice (GPP) undergraduate students, was ready for a new collaboration. The task for the co-design was to integrate native activities and sensibilities into STEM education. Says George Moore, a UC Berkeley mechanical and development engineering graduate student, who taught at the summer workshop, “It became clear in conversations with the PPN that it was hard to get the native students to apply themselves in the STEM disciplines. Institutionally, it just wasn’t structured for them. But these students got really engaged at the workshops and are really good at math and science.” According to Moore—and other participating UC Berkeley students, including GPP students Dor Chavoinik, Grace Harrison, and Arielle Levin and Elena Duran, a PhD student in Graduate Group in Science and Mathematics Education—what works best is listening and not imposing views on what works in a STEM-based activities. Among the decisions were to teach engineering design through Pomo Pinoleville basket-making techniques and to engage students in 3D printing designs from local art and nature. Qui says the workshops and maker space are generating excitement among the students to get into STEM fields. “We have a college career counselor coming to the classroom,” she notes. “The students have a sense these are high paying jobs. Yet for native people, we’re not just looking at the pay. We’re looking at how we can use the STEM program to serve our own community around solar power, rain catchment, and other sustainable and environmental solutions.” For graduate student Pierce Gordon, the co-design approach is crucial for mechanical/development engineers like himself working in poor communities. Gordon says co-design is “de-colonizing,” as it simultaneously aware of the deep history of technological interventions and adamant that everyone be heard, understood, and acknowledged. Continues Gordon, “If we don’t do that, then we’re doing the very similar kind of harm that many people have done over the history of international development and interventionist work as a whole.” Gordon, who is finishing his PhD dissertation, which includes case studies of co-design efforts in the United States and Botswana, says the first priority of development engineering work is not to get research publications, funding, or material for teaching classes, but to benefit marginalized communities. “It is to figure out what the community wants, because this research moves at the speed of trust. Once you build up that trust, you have the opportunity to build up to collaborations and research outcomes and beneficial activities that you didn’t even know could have existed.” For Angela James, one very specific outcome of the 10-year collaboration is her children’s interest in STEM. “My daughter has been a participant in CARES since she was four. She’s 14 now. She’s very comfortable leaving Ukiah. She’s looking at a lot of different colleges. All her career interests are science-based. My son is right behind her and just the same.” James, who was on the UC Berkeley campus on July 13, 2018 with a group of Native American high school students, says the days of cultural and educational isolation can end for her tribe and others in California. “My goal has been to open the minds of our youth and introduce them to college and science, and teach them how to build positive working relationships with people outside their immediate circle,” says James. “It is important that the university has the right individuals involved in a collaboration—people who are willing to advocate for the human approach, get to know the individuals, and ask about background and culture. An important part of this collaboration has been that our voice is finally being heard.” Research about this collaboration was partially supported by the National Science Foundation’s Research Traineeship in Innovations at the Nexus in Food, Energy, and Water Systems (Award No.

By Veena Narashiman An injured soldier is rushed to a field hospital and is bleeding out. A surgeon needs to give the soldier meds to speed up her clotting. But too much or too little will kill her. The doctors rely on lab equipment to determine dosage; however, the large machine was never designed for field use. The surgeon is caught between an educated guess and blind dosing, putting the soldier’s life at risk. This was the story that Jeffrey Lu and Johnathon Li heard from a U.S. Air Force vascular trauma surgeon. The two old friends, UC Davis graduate students in biomedical engineering and animal science, realized they had possibly stumbled upon a market gap for a mobile blood clotting monitoring device. After conversations with UC Davis doctors, their hunch was confirmed. Not only did they learn that traumatic injuries which disrupt blood clotting are the second leading cause of preventable death in developed countries, they discovered that mobile blood clotting solution could had worldwide application—from the frontlines of the war in Syria to rural areas in Sub-Saharan Africa. Lu and Li also learned that with current technologies, an injured soldier may not receive treatment for up to 24 hours; and in civilian hospitals, patient treatment can be delayed three hours. The technicians and surgeons they interviewed said they wanted a device they could use in the operating room, circumventing the time involved in sending samples to the lab. Surgeons especially complained that when they got back lab results, the information was often obsolete because the patient’s condition had changed from further bleeding out. “Current [blood-clot testing] devices are like using microwaves to cook,” said Lu. “It works if you don’t move it, and occasionally they come out great, but more often than not you’re just going to be disappointed.” Or as Dr. Joseph M. Galante, the trauma medical director at UC Davis Medical Center, put it: “Undiagnosed coagulopathy [bleeding disorder] in trauma patients is associated with greater transfusion requirements, longer intensive care unit and hospital stays, and greater incidence of multi-organ failure or death. Patients with uncorrected coagulopathy are eight times more likely to die within the first 24 hours following trauma.” Lu and Li spent part of their graduate school years working on the project they dubbed Innovis Medical. The partners began to understand their competition, their possible business model, and the people they needed to cultivate to make the best possible medical device. In October 2017, Lu and Li turned to the Blum Center’s Big Ideas student innovation contest, to further shape and fund their idea. Big Ideas is open to undergraduate and graduate students at all 10 UC campuses and had a contest category that fit their invention: Hardware for Good, made possible through the generous support of the Autodesk Foundation. The contest put them through a nine-month project incubation, mentoring, and application process that Lu and Li saw was crucial to their company’s development. “In entrepreneurship, you’re not selling your idea, you’re selling your network,” said Li. “Big Ideas participants enter Berkeley’s well-oiled machine, and their biggest advantage is their network.” By meeting new people and potential advisors, the Innovis Medical founders realized they needed to pivot their strategy. Lu and Li decided to prioritize the civilian market instead of battlefield situations and use more layman language to describe their product. At the 2018 Big Ideas Pitch Day before winning a first place prize, Lu went into the specificities of the device:  “Our solution is a portable medical device that uses a solid state sensor to track an electrical property of blood known as bioimpedance as it clots. Our device produces graphs and data similar to the current state of the art device, but without the bulky sensor mechanical components. The sensor itself is a disposable cartridge with no mess to clean up, no chemicals to work with. With a solution designed specifically to mobility, blood clot tests are no longer restricted to laboratories but can be used in a battlefield, operating room and even the comfort of your own home.” At the pitch, Lu further argued that tests during surgery, which take 30 minutes to receive back from the lab, could be performed in the operating room within four to 10 minutes. Post-surgery patients who previously needed to make a trip to the hospital every few weeks to have their blood thinner dosage checked, could run the tests themselves at home—much like diabetic patients who are able to track their own insulin levels.   In January 2018, Lu and Li joined UC Davis’ Inventopia, a makerspace for startups, where they were able to witness the production of their device’s sensor. The next month, they attended Meet the Experts Night at UC Berkeley, whey they were connected to Rhonda Shrader, director of the Berkeley-Haas Entrepreneurship Program, who referred them to contacts throughout the Bay Area. Innovis Medical estimates significant reach and cost savings. Its market could include an annual 672,000 U.S. military trauma cases, 15 million U.S. civilian cardiac surgeries, and 7 million disaster-related surgeries in the developing world. As for costs savings, Innovis estimates the 15 million annual civilian patients of cardiac surgery could save up to $7,000 per operation. In developing countries, Innovis believes its device could be crucial in setting where reliable energy and technicians may not be available. Lu and Li recently expanded their business plans through the UC Berkeley Innovation Corps course and were accepted to the national I-Corps incubator for scaling university-based innovations run by the National Science Foundation. The founders also have launched a collaboration with UC Davis Medical, where civilian and military surgeons are using the Innovis device to directly test human blood from cardiac patients alongside status quo devices. Lu said sensors are being deployed for clinical tests with the aim to iterate the device to address as wide a range of patients and blood types as possible. He and Li hope to get FDA approval by 2021.

Blum Center Education Director Alice Agogino has been named winner of the Presidential Award for Excellence in Science, Mathematics and Engineering Mentoring, the government’s highest honor for mentors who have worked to expand talent in science, technology, engineering and mathematics (STEM). The award was announced June 25 by the White House Office of Science and Technology Policy and the National Science Foundation. Agogino, the Roscoe and Elizabeth Hughes Professor of Mechanical Engineering at UC Berkeley, was one of 41 honorees to receive the award at a ceremony last week in Washington, D.C. Professor Agogino has had a long and illustrious history of mentoring university students and junior faculty as well as engaging with local schools, museums and organizations to engage K-12 students in STEM topics. To support engineering students at UC Berkeley, she created a tiered mentoring network, in which senior doctoral students advise masters and undergraduate students. Over the years, she has been in high demand as a mentor by those who want to use their STEM educations for positive social impact. She also has built a reputation for designing courses that attract a high percentage of women and under-represented minorities. At the Blum Center, Professor Agogino has been pivotal in creating the new field of Development Engineering, whose mission is to reframe development and the alleviation of poverty by educating engineering and social science students to create, test, apply and scale technologies for societal benefit. Development Engineering students, she has written, must learn “21st century skills”—interdisciplinary, team-based methods that are oriented to seeing problems from multiple viewpoints (quantitative, qualitative, ethnographic) and applying them through entrepreneurial pathways. Professor Agogino is not new to awards. She is a member of the National Academy of Engineering and is the previous recipient of an ASME Ruth and Joel Spira Outstanding Design Educator Award “for tireless efforts in furthering engineering design education.” At UC Berkeley, she has received Chancellor Awards for Public Service, a Chancellor’s Award for Advancing Institutional Excellence and a Faculty Award for Excellence in Graduate Student Mentoring. She was elected a Fellow of the American Society of Mechanical Engineers, has won many best paper awards and has been honored with a National Science Foundation Distinguished Teaching Award and a AAAS Lifetime Mentor Award, the latter for increasing the number of women and African- and Hispanic-American doctorates in mechanical engineering. Her work in decision-analytic approaches to engineering design led to a whole new field of research, and her research in mass customization became a patent-buster for licenses in database-driven Internet commerce. If that were not enough, Squishy Robotics, Inc., Professor Agogino’s startup company, recently was awarded a National Science Foundation Small Business Innovation Research grant to conduct research and development work on “Shape-Shifting Robots for Disaster Rescue, Monitoring and Education.” Professor Agogino has explained that she was inspired to become a mentor due to her own experience at the University of New Mexico, where she was the only female mechanical engineering undergraduate student, and at UC Berkeley, where she became the first woman to receive tenure in her field. Professor Agogino uses a mentoring approach that she calls “designing for diversity.” By emphasizing the social impact of solving research problems, this strategy helps students feel connected to their work and motivated to persist in

By Veena Narashiman and Blum Center News In 2013, the Blum Center and  the Center for Effective Global Action founded the Development Impact Lab, to launch the new field of Development Engineering and create a model for university-based poverty action labs. Since that time, the Development Impact Lab, with support from USAID, has tested over 135 innovations and engaged more than 500 students, 400 experts and 375 organizations, involving 16 universities in the United States, India and Uganda. On June 4, key representatives from the network met at the Blum Center to discuss their five years of findings and outcomes. Temina Madon, executive director of the Center for Effective Global Action, summarized Development Engineering as enabling doctoral students from multiple disciplines to research and test poverty solutions as part of their dissertations. A panel entitled “Institutionalizing the Field of Development Engineering” included UC Berkeley Mechanical Engineering Professor Alice Agogino and Economics Professor Paul Gertler, who attested to the long need for such a PhD minor. Rachel Dzombak and Sophi Martin, who earned their PhDs at Cal and are serving as the Blum Center’s doctoral fellow and innovation director, respectively, also offered anecdotes about rising interest in Development Engineering.  “Thanks to this new field, doctoral students considering international development now have a way to harness this aspect of their academic interests,” said Professor Agogino, who is chair of the Graduate Group in Development Engineering and Education Director of the Blum Center. Professor Gertler gave an overview of the Development Engineering journal, an open access,  interdisciplinary publication that applies engineering and economic research to the problems of poverty. The two-year old journal, he explained, is giving scholars academic credit for novel research that previously had not been widely acknowledged or disseminated. Katherine Dow, program manager of the Global Development Lab, said that the Development Impact Lab has helped USAID better understand emerging trends in science and technology for development and the role that university professors, researchers and students can play. She also said the UC Berkeley collaboration has helped bring quality data and data methodologies to government decision makers, allowing for a redefinition of problems. Data gathering around energy reliability in Ghana was a focus of one session. Hana Freymiller and Jeffrey Garnett from the U.S. Government’s Millennium Development Corporation, gave an overview of its $535 million, five-year project in Ghana, which aims to decrease the country’s energy outages by 20 percent by 2021.  The Millennium Development Corporation is collaborating with the Development Impact Lab and UC Berkeley’s Lab11 to better understand where and when outages are happening in Ghana—using a mobile app called Gridwatch. GridWatch enables utility customers to automatically report outages through sensor technology on their cell phones. The data collected is more accurate in many cases than what the utility company can gather, and can show a variety of possible solutions for energy investment, energy savings, economic development, and improved quality of life. “People are worried about the 1.1 billion people who don’t have access to electricity worldwide,” said UC Berkeley Business and Economics Professor Catherine Wolfram, a GridWatch project lead. “But what about those who don’t have reliable access? As urbanization trends move forward, it’s really important to understand reliability and measure how investment changes with reliability.” The daylong conference also featured presentations by faculty and current and former PhD students associated with the Development Impact Lab. UC Berkeley Bioengineering Professor Dan Fletcher summarized the progress of CellScope, an invention from his lab that adapts the camera of a mobile phone or tablet computer into a high-quality light microscope for disease detection in low-resource areas. Since 2008, Professor Fletcher has used CellScope to test more than 83,000 patients, analyzing the results in a dozen journal articles—and now is aiming to mass produce 10,000 CellScopes for use in rural areas in Africa. His insight was: “Don’t make technology for development as simple as possible; make it as automated as possible.” Erin Kelly, a PhD candidate in Agricultural and Resource Economies at UC Berkeley presented a mobile phone application called SmartMatatu, designed to prevent the high incidence of traffic accidents among privately owned minibuses in Nairobi, Kenya. The application uses GPS and other mobile technologies along with affordable, off-the-shelf car sensor devices to collect and send location, ignition, route, distances, speed, acceleration and deceleration information, to show Matatu drivers when, where and how they are driving unsafely. According to a six-month survey of SmartMatatu users, the app has helped drivers raised their profits and reduce repair costs–demonstrating that visibility of data can influence driver behavior and productivity. Danny Wilson, a Development Engineering alumnus, presented the outcome of his doctoral research: Geocene.com. With Development Impact Lab funding, Geocene has created hardware and apps for data logging and analysis. Wilson talked about findings from a survey of cookstoves users in Sudan in which people over-reported use of their indoor pollution-reducing stoves, making the survey findings unreliable. With Geocene, Wilson aims to create dashboards for monitoring cookstoves across many countries. The goal, he said, is to understand what are the best cookstoves for specific conditions, so people will actually use them. He noted that nearly two million deaths could be prevented annually by replacing cooking fires and inefficient, smoky stoves. Susanna Berkouwer, a PhD candidate at the UC Berkeley Department of Agricultural and Resource Economics and the Haas School of Business’s Energy Institute, reported on the latest findings from the Development Impact Lab’s Rural Electric Power Project, which is utilizing novel data collection and analysis tools. She explained that Kenya is working on a last-mile connectivity project to power all households by 2020, and is collaborating with her group to better understand how energy access does (or doesn’t) lead to higher consumption, income, health and education. The results, she said, could empower both citizens and governments. The other presentations included: Niall Keleher, a UC Berkeley School of Information PhD Candidate, presenting the results from Assistant Professor Joshua Blumenstock’s research applying machine learning to high-resolution satellite imagery to measure regional poverty in Africa; Chinmayee Subban presenting Berkeley Lab Water-Energy Resilience Research Institute’s progress on using charge-based salt water removal to clean brackish water; and Dana Hernandez, a Development Engineering student, summarizing UC Berkeley Civil and Environmental Engineering Professor Ashok Gadgil’s ongoing research on removing arsenic from groundwater. “The Development Impact Lab has had a tremendous five years,” said Heather Lofthouse, the Blum Center’s director of special projects. “By partnering with USAID and other governmental and nongovernmental organizations, we have been able not only to generate new ideas—we have been able to implement real-world

By Rachel Pizatella-Haswell, UC Berkeley Goldman School of Public Policy MPP ’18 Joshua Blumenstock is an Assistant Professor at the UC Berkeley School of Information, where he directs the Data-Intensive Development Lab, and a member of the Blum Center’s Development Engineering faculty. His research lies at the intersection of machine learning and development economics, and focuses on using novel data and methods to better understand the causes and consequences of global poverty. Blumenstock has a Ph.D. in Information Science and a M.A. in Economics from U.C. Berkeley, and Bachelor’s degrees in Computer Science and Physics from Wesleyan University. What can remote sensing and geographic information system data and cell phone data tell us about a person living in poverty? Blumenstock: We have partial answers to that question. The work that’s been done indicates we can estimate very basic things: population density, household average wealth, basic indices of relative socio-economic status. Of course, there are lots of different ways to measure poverty and inequality and welfare. People working in developing countries tend to like consumption because it seems to be most closely correlate to how someone is actually doing. There has been some work looking at whether you can estimate consumption and expenditures from remote data sources, and initial results are promising here too. Aside from measuring basic welfare, all sorts of work is being done to use these data to learn about migration, social network structure and the spread of disease, to give a few examples. What can data tell us about poverty indicators such as the incidence or depth of poverty? Blumenstock: What these models actually spit out are sub-regional estimates of welfare. We can define welfare however we want. In general, as long as you can measure it in the traditional way, you can use these non-traditional data and models to try to estimate it. However, depending on what you want to measure, and what data source you’re using – such as phone data or satellite data – your estimates may be more or less accurate. But once you have your estimate of the distribution of wealth, you can do all of the things you could do with traditional data. You can back-out the poverty incidence, Gini curves and other constructs you derive from the poverty distribution. In what ways are these data sources limited in their ability to measure welfare or other things? Blumenstock: I think we have to be careful because it is easy for people to get excited at the potential applications before really understanding the fundamental and practical limitations. Nonetheless, here are several margins where this could be a major improvement over the status quo. One is cost: it is a lot cheaper to collect cell phone data, for instance, than a nationally representative household survey, which costs tens of millions of dollars. Another is geographic resolution: budgets constrain the areas and number of people that you can survey, but satellites can quickly collect millions of images from a small region. Another is temporal resolution: again, because of the costs, you can only do a nationally representative survey every few years at best, but phone data gets updated every second and satellite data gets updated every day. So, if you can update your estimates of the things that you care about – the poverty incidence or the distribution of wealth – every day, that could be really useful. We can think about all of the applications: not just program targeting and impact evaluations but also program monitoring and disaster response. All of these things need up-to-date estimates of the distribution of welfare. Those are all of the reasons why I think people can be excited, including me, but we’re just at the very first baby steps of that long pipeline. There are basically two canonical papers out there: one that I worked on and one that a Stanford group worked on. Using the most up-to-date data, what we show is that cell phone meta data can be used to estimate relative wealth very accurately. The group at Stanford shows that daytime satellite imagery can be used to do the same thing. They also look at consumption, and find similar results. However, both of those studies include a very small number of countries at one point in time. We have no idea yet if those models that you calibrate at one point in time can generalize into future points in time. To do a lot of the tantalizing applications, like monitoring and impact evaluations, the first step, which has not been done yet, is to show that these estimates can reliably allow for inference over time. Poverty is heterogeneous across both space and time. Some of your work highlights the ability of machine learning to provide granular spatial assessments of poverty and assessments of poverty in real time. What are the implications of this for the delivery of poverty alleviation programs? Blumenstock: I don’t think this is going to solve any of these age-old problems relating to the shallowness of quantitative estimates. There’s always going to be trade-offs between quantitative and qualitative research. There’s the famous quote that says, “Not everything that counts can be counted, and not everything that can be counted counts.” Here, it’s no different. There is a latent characteristic of people, which is welfare or well-being, and that’s what we want to measure. However, you can’t measure it directly. You can either embed yourself in a community and really get a sense for it, or, if you want to measure welfare at scale, you have to rely on these instruments that make observations about one dimension of things that we think are correlated with this fundamentally unobservable state. And what we’re doing is even one-step removed from that. We have these instruments that are already imperfect measures of that underlying state, and we want to try to replicate that using data that we get for free at true scale. At best, what we’re doing is trying to replicate those already imperfect instruments, which adds another layer of imperfection. There certainly is nothing about what we’re doing that imposes homogeneity, though. We would never use a model that we fit on Rwanda and apply it to Kenya without knowing first how much we expect the model’s prediction to degrade by. Similarly, I would never expect a model that was fit in 2016 to be accurate in 2018. But that’s an empirical question, and one that we’re actively studying. Some of these are solvable problems, but I think there are these other unsolvable, philosophical questions like construct validity and if we’re measuring that thing that we care about. Those things are more fundamental. Household income changes overtime due to income shocks throughout the course of a year. How do you view this as a mechanism to correct problems in real time? Could this inform programs to better smooth consumption and target across time as opposed to just one moment? Blumenstock: I think there are some very compelling applications that look at inter-temporal consumption or changes in dynamics. For a lot of reasons, you might think that that’s a first order thing to target rather than a cross-sectional, stable measure of permanent income. In principle, this is exciting because with this line of research you can potentially get updated estimates at very high frequencies. Yet, that is a couple steps ahead of where we are now. In my mind, step one is seeing if data from a single point in time is accurate. We can sort of do that now. Step two is seeing if the estimates can remain accurate across different points in time. We haven’t quite gotten there, but we’re working on it. Step three is what you’re talking about: real-time estimates. I think those need to be done in that order. Unless we know the right way to generate dynamic estimates and, then, know the right way to layer those onto a real time streaming data set, real time estimates will be wrong. That won’t be for another few years. You mentioned that you wouldn’t fit a model from one country to another country. What are the ways forward to potentially be able to generalize from one country to another using this data? Blumenstock: Generalizability can mean a lot of different things. One is generalizing over space, like from one country to another or even within one country from one region to another. Another is generalizing over time. A third is generalizing from a population that you observe to a population that you don’t, even if it is in the same space at the same time. Of the three, I think that generalizing from one country to another is the easiest. That is an empirical question. We can collect data from two countries, or 10 countries (I’m working on a project where we’re collecting data from 50 countries), and we can just see if we train the model in one country and apply it to another. Then, we need to determine if the degradation of the model’s estimates depend on things that we observe: whether the countries are on the same continent or how far apart they are; whether the ethnic composition of the two countries is similar; or if the distribution of wealth is similar or not. These are things that we can measure. So, we don’t want to just apply the estimates from one country to another, but rather, we want to have a sense for how to correct for translation errors, or at least know when and where such errors are likely to exist. We’ll hopefully have answers to some of these questions in the next few months. We’re also actively working on this project to have a sense for the ability for phone- and satellite-based estimates to generalize over time.  But realistically, it will be a while until we have a conclusive answer. The hardest one is generalizing from an observed to an unobserved population. That said, there are a lot of techniques from traditional econometrics that you might apply to this problem. If you know the process that governs whether someone is observable or not, then you can “reverse engineer” a statistical correction. For instance, if you only observe people from population A, you can only reliably estimate the distribution of wealth in population A. But, say you want to be able to estimate the wealth of population B, but they’re not visible in your data (because they don’t have phones, for instance). In this scenario, if you know something about how the distribution of wealth of population A relates to that of population B, you apply a transformation to the estimates of population A to get estimates of population B. These are not big data things or new data things. These are old problems with sample selection and construction, to which we have partial solutions. What are some of the ethical concerns with using machine learning to track poverty that you’re confronting and how do you confront these issues, especially in consideration of the inherent vulnerability of those who your work is meant to serve? Blumenstock: The thorniest ethical questions for me are more philosophical ethical questions: things like the legibility of populations and the possibility of misappropriation. Is it a good thing to make people easier to measure? What if an authoritarian regime takes the papers that I’m writing and uses them to weed out political dissenters? For these sort of questions, I try to look at the scale of negative use cases and positive use cases, and focus on problems where the positives outweigh the negatives. Another important issue is privacy, in the sense that we often deal with data that people generate without a full understanding of how it can be used to draw inferences about them. Here again there is an ethical concern and a practical one. The practical one is easier to address: at least as far as our research is concerned, we do our work in a controlled research environment. We put in place rigid data protection procedures, like removing personally identifying information prior to conducting analysis, to ensure that, to the extent that we can, we are safeguarding the privacy of the people we study. But we can’t control the privacy practices of others, like industry or government. So, the ethical issues are bigger, and boil down to a similar calculus as I mentioned earlier – do we think the benefits outweigh the risks? There’s an assertion that there’s tension between technocratic solutions and human-rights based solutions. Beyond the implications for poverty alleviation, in what ways does or could machine learning contribute to good governance or protection of human rights? Blumenstock:  It could contribute in a lot of ways — both through the applications that have been developed in the last few years, as well as those that are on the horizon. I can see why these “big data measurement” methods may seem more natural a fit for technocratic approaches. But bottom-up governance structures need data too. You need to know who your constituents are. A lot of the things that we’re measuring now are largely motivated by more of the holistic, softer things that have been off limits to technocratic fixes. For instance, we’re working on a project now that tries to quantify the extent to which violence disrupts the social fabric of places like Afghanistan — using phone data to observe the social fabric in ways that wouldn’t be possible with other methods. Technocrats are limited by what they can observe, and one of the things they have a hard time observing is community cohesion and fragmentation. If we can provide ways for people to more directly observe that then it can create a bridge between what technocrats are equipped to do and the more bottom up approaches to good